Introduction to the Heliconiini system used in this study.

A: The Heliconiini feature a previously reported expansion of the mushroom bodies in Heliconius sp. concomitant with a dietary innovation of pollen feeding. Shown are 3D segmentations of the OL, AL, MB, CX, AOTU and POTU in the brain of four exemplary species. B: Examining the CX of these four species already indicates that CX volume seems conserved, relative to variation in the MB. Shown are separate 3D segmentations of the MB and AOTU as well as the CX and POTU, with the CX and POTU consistently enlarged (i.e. not to scale to the MB models). C: To assess divergence in volumetric investment in these structures we used a dataset of 307 individuals of 41 species of Heliconiini. Depicted is the phylogeny 109 with species names, appearance and number of individuals per species. Species shown in A and B are indicated here by a circle at the end of each edge. Scale bars are always 250 μm. Abbreviations: OL optic lobe, AL antennal lobe, MB mushroom bodies, CX central complex, AOTU anterior optic tubercle, POTU posterior optic tubercle, PB protocerebral bridge, FB fan-shaped body, EB ellipsoid body, NO noduli.

The central complex, the AOTU and POTU do not show pollen-feeding linked patterns of volumetric expansion or increased evolutionary rates.

A/B: Assessments of the effects of mushroom body size interacting with pollen feeding and with previously identified expansion events in the sum of all CX neuropils (total CX), AOTU and POTU, and in separate CX neuropils (see Figure S2). Shown are each neuropil against a measure of whole brain size, rCBR (A), as well as against MB size (B). Results are indicated in each panel; first, the effects of each interaction by assessment of DIC differences, and then the significance of the relationship between the MB and each neuropil. Colour coding was done according to clade differences in mushroom body size identified previously 29. C/D: Analyses of evolutionary rates of MB size relative to rCBR (C), and CX size relative to rCBR (D), which reveals two very distinct evolutionary patterns. In the MB (C) particularly high rates of evolution have previously been identified on the branch leading to Heliconius 29, coincident with the innovation of pollen feeding. At the same branching point we see relatively low rates of evolution in CX size (D). Both phylogenies are shown at the same scale (see Figure S3 for a CX tree with absolute scale of evolutionary rates). The pollen feeding photograph is kindly provided by Sebastian Mena. CX central complex, AOTU anterior optic tubercle, POTU posterior optic tubercle, MB mushroom bodies, rCBR rest of the central brain. PF pollen feeding.

Central complex neuropils and associated areas show different patterns of scaling with each other.

A: Tests of significant scaling relationships between each substructure on the left, as dependent variable, and all others as independent variables. Each cell indicates the test statistic (posterior mean) and the P value in parentheses, and if significant is shown in bold. As values for NO were only available for parts of the total dataset, noduli were not included as independent variables. B: Prominent cell types that interconnect central complex neuropils and the AOTU and POTU, which may potentially explain patterns of neuropil scaling, if positive scaling relationships indicate co-evolution among functionally interdependent structures. Cell type depictions are examples with localisation inside each neuropil being purely visual. Namings first refer to the connected neuropils, then in brackets to established nomenclature, based on current literature 21,45,46,51–55. C: A depiction of where our data showing significant scaling relationships (A) matches (check mark), or mismatches (X), expectations that are based on prominent (mostly columnar) neuron classes and connections (B). Where expectations are unclear, we have annotated the comparison with (?). Specifically, we would normally not expect positive scaling between AOTU and FB, but this may be explained by findings in Figure 7 and an increased population of ER neurons projecting to the FB. Generally, this diverse pattern indicates that more variation occurs throughout the system than can be captured in this volumes-based analysis. AOTU anterior optic tubercle, POTU posterior optic tubercle, PB protocerebral bridge, FB fan-shaped body, EB ellipsoid body, NO noduli, GA gall.

Tract and synaptic labelling reveal a conserved central complex architecture in Heliconiini butterflies.

A/B: Double labelling of tubulin (green) and HRP (magenta) was used to reveal the synapse-dense areas of the central complex as well as its underlying tract systems. C/D: Noduli architecture reveals, tentatively, a conserved pattern of four main domains, but with very pronounced asymmetry between hemispheres. E/F: Tract bundles close to the PB associate with DM1-4/WXYZ tracts (arrows). Tangential neuron bundles project across hemispheres through the PB (arrowheads). HRP alone (grey) allows a rough approximation of vertical columns in the PB (a-h). G/H & K/L: Labelling reveals two distinguishable layers in the fan-shaped body while additional staining elsewhere reveals additional detail (arrows). Thicker tract conflations indicate the columnar architecture determined through the four columnar neuron bundles (arrowheads). I/J & K/L Labelling reveals two pronounced layers in the EB (arrows), while obvious columns could not be indicated. PB protocerebral bridge, FB fan-shaped body, EB ellipsoid body. A anterior, P posterior. Scale bars are 50 μm.

Mass injections in Dryas iulia reveal columnar architecture and input into the fan-shaped body.

A: A single injection into the superior median and lateral protocerebrum (SMP/SLP) of Dryas iulia has revealed prominent labelling of the fan-shaped body, particularly layers I, II (dorsal projection) and IV (ventral projection). B: Input pathways into these layers from a dorsal set of projections through four input tracts into the fan-shaped body (arrowheads), originating from the SMP/SLP region (C, arrow), whereas A’ and B’ are close-ups of A and B, highlighting the unique columnar architecture of the fan-shaped body revealed through this labelling. D: Ventral projections also seem to originate from the SMP/SLP region but are hard to trace. a-c depict close-ups of an interesting input pathway from γ lobe and lobelet into the fan-shaped body through the ventral projection. E: Schematic of labelling in generic central complex. Abbreviations: FB fan-shaped body, lob lobelet of the mushroom body. Scale bars equal 100 μm.

Comparative anatomical analysis of the Heliconiini central complex examining dopaminergic and serotonergic neurons and Fasciclin-II expression supports a largely conserved neurotransmitter expression and anatomy.

Shown are original data and a schematic of the central complex summarising expression patterns inside the synaptic region of the central complex. A-C: Patterns of Tyrosine Hydroxylase (TH) labelling of dopaminergic neurons depict weak labelling of fan-shaped body layers I and II, spot-like layering of ellipsoid body layers I-III, as well as projection pathways from the superior medial protocerebrum into the fan-shaped body (arrows), ventral projections into the ellipsoid body (arrows) and projections from the bulb into the ellipsoid body (arrows). D-F: Patterns of 5-hydroxytryptamine (5-HT) labelling of serotonergic neurons depict prominent labelling into layers II and III of the fan-shaped body, from projections in the superior medial protocerebrum (arrowheads), as well as showing spot-like labelling of the ellipsoid body layers I and II from ventral projections as well as dorsal projections (arrowheads). G-I: Fasciclin-II labelling of a subpopulation of columnar neurons across all lineages DM1-4 emerging into prominent labelling of layers II-IV of the fan-shaped body and all layers in the ellipsoid body. In addition, projections into the gall from the ellipsoid body are shown (arrowheads). Abbreviations: PB protocerebral bridge, FB fan-shaped body, EB ellipsoid body, NO noduli, MB-LO mushroom body lobes, BU bulb, DM1-4 dorso-medial lineage 1-4, GA gall. Scale bars equal 100 μm.

Anatomical and statistical analysis of GABA-ergic DALv2 lineage-derived ER neurons showing higher innervation of the FB in pollen feeding Heliconius melpomene versus Dryas iulia.

A-C: Patterns of GAD (glutamate decarboxylase) labelling of GABA-ergic neurons depicting spot-like labelling in the fan-shaped body layers II and III, complete labelling of the ellipsoid body and spot-like labelling of nodulus layers I and II. D/E: Location of the GABA-ergic cell group and its projections in focus relative to the brain. F/G: Position of cell bodies, nestled between the mushroom body lobes and antennal lobe at the anterior surface of the brain, along four different positions of the anterior-posterior axis (“1-4”). H: Quantification of cell body numbers in both species revealed a significant increase in Heliconius. I/J: ER neurite close to the cell bodies at two different positions (“1-2”). K/L: Projection of ER neurites into the central body, with a focus on the projections into the FB. “I/II” indicate two tracts that project ventrally and form a fibre that sits ventrally to the EB (“III”) M/N: A perpendicular view onto the projections labelled with “III” show the relative position of the beginning synaptic EB neuropil as well as the ER projections and ramifications leading into the FB. O/P: Focus on the ramifications of the ER-FB neurons inside the FB at two different A-P axis points (“1-2”), as well as a lateral view (“3”). Q/R: 3D segmentations of the ER neurites with focus on the FB contributing portion, with three different viewpoints (“1-3”). Light blue depicts the massive projections into the light-grey labelled EB, with green showing the projections and ramifications into the FB. All immunostaining depictions are based on an anti-GAD labelling. Abbreviations: BU bulb, LEa lateral ellipsoid body tract, ER ellipsoid body ring neurons, MB mushroom bodies, AL antennal lobe, LX lateral complex, FB fan-shaped body, EB ellipsoid body. A anterior, P posterior. Scale bar equals 50 μm, except for the central body panels in A/B where it is 100 μm.

Anatomical analysis of Allatostatin A expression across five representative species of Heliconiini reveals variability consistent with behavioural innovations in Heliconius.

A: Summary of the dataset used to assess Allatostatin A variability, with the phylogeny of sampled species and indication of when pollen feeding evolved. B: General expression pattern of Allatostatin A that is present throughout all species. C: Schematic representation of the interspecific variation present in Allatostatin A expression in fan-shaped body, ellipsoid body and noduli, with major differences indicated in orange lettering. If consistent with a shift in Heliconius, this was indicated by the pollen feeding photograph. Lower-case letters correspond to figure panels in D-F, where this variability of expression is depicted. Note that Dione juno varies in its noduli expression even between individuals, hence the double-headed arrow. D-F: Variation in Allatostatin A expression (green or grey) across the five Heliconiini species. Synapsin was used as co-stain, depicted in magenta. Lower-case letters correspond to the schematic in c, with schematic depictions of the central complex neuropil in question indicated where consistent differences between species was identified. Arrowheads specifically point to difficult to identify expressions of Allatostatin A in the ellipsoid body of some species. Arrows point to differences described in the schematic and in text. Abbreviations: PB protocerebral bridge, FB fan-shaped body, EB ellipsoid body, NO noduli. Scale bars in D and E equals 100 μm and in F 25 μm.