Heliconius Butterflies Research Articles

Heliconius butterflies use wide-field landscape features, but not individual local landmarks, during spatial learning.

Spatial learning is vital in foraging ecology. Many hymenopteran insects are adept spatial foragers that rely on visual cues contained within broader wide-field scenes for central place foraging from a central nest. By contrast, for butterflies, which lack central nest sites, visual cue use during spatial foraging is less understood. Heliconius butterflies, however, exhibit stable nocturnal roosts, strong site fidelity and a sophisticated capacity for spatial navigation. This study furthers our understanding of Heliconius spatial learning by testing whether H. erato can associate a spatially informative visual cue with artificial feeders. We explored the relative importance of a visual local landmark compared with broader, wide-field visual cues, through experiments with (i) a fixed rewarded feeder with a local landmark; (ii) a mobile rewarded feeder with the landmark as the sole reliable cue; (iii) the same setup while blocking visual access to external landscape features. Our data suggest that Heliconius butterflies learn static feeder locations without relying on a local individual landmark. Instead, we suggest they integrate broader landscape and celestial cues. This suggests that Heliconius butterflies and central place foraging hymenopterans likely share similar visual navigation strategies, using wide-field, low-resolution views rather than focusing on specific individual landmarks.

Inference of Phylogenetic Networks from Sequence Data using Composite Likelihood.

While phylogenies have been essential in understanding how species evolve, they do not adequately describe some evolutionary processes. For instance, hybridization, a common phenomenon where interbreeding between two species leads to formation of a new species, must be depicted by a phylogenetic network, a structure that modifies a phylogenetic tree by allowing two branches to merge into one, resulting in reticulation. However, existing methods for estimating networks become computationally expensive as the dataset size and/or topological complexity increase. The lack of methods for scalable inference hampers phylogenetic networks from being widely used in practice, despite accumulating evidence that hybridization occurs frequently in nature. Here, we propose a novel method, PhyNEST (Phylogenetic Network Estimation using SiTe patterns), that estimates binary, level-1 phylogenetic networks with a fixed, user-specified number of reticulations directly from sequence data. By using the composite likelihood as the basis for inference, PhyNEST is able to use the full genomic data in a computationally tractable manner, eliminating the need to summarize the data as a set of gene trees prior to network estimation. To search network space, PhyNEST implements both hill climbing and simulated annealing algorithms. PhyNEST assumes that the data are composed of coalescent independent sites that evolve according to the Jukes-Cantor substitution model and that the network has a constant effective population size. Simulation studies demonstrate that PhyNEST is often more accurate than two existing composite likelihood summary methods (SNaQ and PhyloNet) and that it is robust to at least one form of model misspecification (assuming a less complex nucleotide substitution model than the true generating model). We applied PhyNEST to reconstruct the evolutionary relationships among Heliconius butterflies and Papionini primates, characterized by hybrid speciation and widespread introgression, respectively. PhyNEST is implemented in an open-source Julia package and is publicly available at https://github.com/sungsik-kong/PhyNEST.jl.

Weighting of sensory cues reflect changing patterns of visual investment during ecological divergence in Heliconius butterflies.

Integrating information across sensory modalities enables animals to orchestrate a wide range of complex behaviours. The relative importance placed on one sensory modality over another reflects the reliability of cues in a particular environment and corresponding differences in neural investment. As populations diverge across environmental gradients, the reliability of sensory cues may shift, favouring divergence in neural investment and the weight given to different sensory modalities. During their divergence across closed-forest and forest-edge habitats, closely related butterflies Heliconius cydno and Heliconius melpomene evolved distinct brain morphologies, with the former investing more in vision. Quantitative genetic analyses suggest that selection drove these changes, but their behavioural consequences remain uncertain. We hypothesized that divergent neural investment may alter sensory weighting. We trained individuals in an associative learning experiment using multimodal colour and odour cues. When positively rewarded stimuli were presented in conflict, i.e. pairing positively trained colour with negatively trained odour and vice versa, H. cydno favoured visual cues more strongly than H. melpomene. Hence, differences in sensory weighting may evolve early during divergence and are predicted by patterns of neural investment. These findings, alongside other examples, imply that differences in sensory weighting stem from divergent investment as adaptations to local sensory environments.

Genealogical asymmetry under the IM model and a two-taxon test for gene flow.

Methods for detecting gene flow between populations often rely on asymmetry in the average length of particular genealogical branches, with the ABBA-BABA test being a well known example. Currently, asymmetry-based methods cannot be applied to a pair of populations and such analyses are instead performed using model-based methods. Here we investigate genealogical asymmetry under a two-population Isolation with Migration model. We focus on genealogies where the first coalescence event is between lineages sampled from different populations, as the external branches of these genealogies have equal expected length as long as there is no post-divergence gene flow. We show that unidirectional gene flow breaks this symmetry and results in the recipient population having longer external branches. We derive expectations for the probability of this genealogical asymmetry and propose a simple statistic (Am) to detect it from genome sequence data. Am provides a two-taxon test for gene flow that only requires a single unphased diploid genome from each population, with no outgroup information. We use analytic expectations and simulations to explore how recombination, unequal effective population sizes, bidirectional gene flow and background selection influence Am and find that the statistic provides unambiguous evidence for gene flow under a continent-island history. We estimate Am for genome sequence data from Heliconius butterflies and Odocoileus deer, generating results consistent with previous model-based analyses. Our work highlights a signal of gene flow overlooked to date and provides a method that complements existing approaches for investigating the demographic history of recently diverged populations.

Mating preferences act independently on different elements of visual signals in Heliconius butterflies

Abstract Mating cues are often comprised of several elements, which can act independently, or in concert to attract a suitable partner. Individual elements may also function in other contexts, such as anti-predator defense or camouflage. In Heliconius butterflies, wing patterns comprise several individual color pattern elements, which advertise the butterflies’ toxicity to predators. These wing patterns are also mating cues and males predominantly court females that possess the same wing pattern as their own. However, it is not known whether male preference is based on the full wing pattern or only individual pattern elements. We compared preferences of male H. erato lativitta between female models with the full wing pattern and those with some pattern elements removed. We found no differences in preference between the full wing pattern model and a model with pattern elements removed, indicating that the complete composition of all elements is not essential to the mating signal. Wing pattern preferences also contribute to pre-mating isolation between two other Heliconius taxa, H. erato cyrbia, and H. himera; therefore, we next compared preferences for the same models in these species. H. erato cyrbia and H. himera strongly differed in preferences for the models, potentially providing a mechanism for how pre-mating isolation acts between these species. These findings suggest that contrasting levels of selective constraint act on elements across the wing pattern.

Genomic evidence reveals three W-autosome fusions in Heliconius butterflies.

Sex chromosomes are evolutionarily labile in many animals and sometimes fuse with autosomes, creating so-called neo-sex chromosomes. Fusions between sex chromosomes and autosomes have been proposed to reduce sexual conflict and to promote adaptation and reproductive isolation among species. Recently, advances in genomics have fuelled the discovery of such fusions across the tree of life. Here, we discovered multiple fusions leading to neo-sex chromosomes in the sapho subclade of the classical adaptive radiation of Heliconius butterflies. Heliconius butterflies generally have 21 chromosomes with very high synteny. However, the five Heliconius species in the sapho subclade show large variation in chromosome number ranging from 21 to 60. We find that the W chromosome is fused with chromosome 4 in all of them. Two sister species pairs show subsequent fusions between the W and chromosomes 9 or 14, respectively. These fusions between autosomes and sex chromosomes make Heliconius butterflies an ideal system for studying the role of neo-sex chromosomes in adaptive radiations and the degeneration of sex chromosomes over time. Our findings emphasize the capability of short-read resequencing to detect genomic signatures of fusion events between sex chromosomes and autosomes even when sex chromosomes are not explicitly assembled.

Behavioural changes in aposematic Heliconius melpomene butterflies in response to their predatory bird calls

Prey-predator interactions have resulted in the evolution of many anti-predatory traits. One of them is the ability for prey to listen to predators and avoid them. Although prey anti-predatory behavioural responses to predator auditory cues are well described in a wide range of taxa, studies on whether butterflies change their behaviours in response to their predatory calls are lacking. Heliconius butterflies are unpalatable and form Müllerian mimicry rings as morphological defence strategies against their avian predators. Like many other butterflies in the Nymphalidae family, some Heliconius butterflies possess auditory organs, which are hypothesized to assist with predator detection. Here we test whether Heliconius melpomene change their behaviour in response to their predatory bird calls by observing the behaviour of male and female H. m. plessini exposed to calls of Heliconius avian predators: rufous-tailed jacamar, migratory Eastern kingbird, and resident tropical kingbird. We also exposed them to the calls of the toco toucan, a frugivorous bird as a control bird call, and an amplified greenhouse background noise as a noise control. We found that individuals changed their behaviour in response to jacamar calls only. Males increased their walking and fluttering behaviour, while females did not change their behaviour during the playback of the jacamar call. Intersexual behaviours like courtship, copulation, and abdomen lifting did not change in response to bird calls. Our findings suggest that despite having primary predatory defences like toxicity and being in a mimicry ring, H. m. plessini butterflies changed their behaviour in response to predator calls. Furthermore, this response was predator specific, as H. m. plesseni did not respond to either the Eastern kingbird or the tropical kingbird calls. This suggests that Heliconius butterflies may be able to differentiate predatory calls, and potentially the birds associated with those calls.

Selection drives divergence of eye morphology in sympatric Heliconius butterflies.

When populations experience different sensory conditions, natural selection may favor sensory system divergence, affecting peripheral structures and/or downstream neural pathways. We characterized the outer eye morphology of sympatric Heliconius butterflies from different forest types and their first-generation reciprocal hybrids to test for adaptive visual system divergence and hybrid disruption. In Panama, Heliconius cydno occurs in closed forests, whereas Heliconius melpomene resides at the forest edge. Among wild individuals, H. cydno has larger eyes than H. melpomene, and there are heritable, habitat-associated differences in the visual brain structures that exceed neutral divergence expectations. Notably, hybrids have intermediate neural phenotypes, suggesting disruption. To test for similar effects in the visual periphery, we reared both species and their hybrids in common garden conditions. We confirm that H. cydno has larger eyes and provide new evidence that this is driven by selection. Hybrid eye morphology is more H. melpomene-like despite body size being intermediate, contrasting with neural trait intermediacy. Overall, our results suggest that eye morphology differences between H. cydno and H. melpomene are adaptive and that hybrids may suffer fitness costs due to a mismatch between the peripheral visual structures and previously described neural traits that could affect visual performance.

Hybrid speciation driven by multilocus introgression of ecological traits

Hybridization allows adaptations to be shared among lineages and may trigger the evolution of new species1,2. However, convincing examples of homoploid hybrid speciation remain rare because it is challenging to demonstrate that hybridization was crucial in generating reproductive isolation3. Here we combine population genomic analysis with quantitative trait locus mapping of species-specific traits to examine a case of hybrid speciation in Heliconius butterflies. We show that Heliconius elevatus is a hybrid species that is sympatric with both parents and has persisted as an independently evolving lineage for at least 180,000 years. This is despite pervasive and ongoing gene flow with one parent, Heliconius pardalinus, which homogenizes 99% of their genomes. The remaining 1% introgressed from the other parent, Heliconius melpomene, and is scattered widely across the H. elevatus genome in islands of divergence from H. pardalinus. These islands contain multiple traits that are under disruptive selection, including colour pattern, wing shape, host plant preference, sex pheromones and mate choice. Collectively, these traits place H. elevatus on its own adaptive peak and permit coexistence with both parents. Our results show that speciation was driven by introgression of ecological traits, and that speciation with gene flow is possible with a multilocus genetic architecture.

Adaptive introgression of a visual preference gene.

Visual preferences are important drivers of mate choice and sexual selection, but little is known of how they evolve at the genetic level. In this study, we took advantage of the diversity of bright warning patterns displayed by Heliconius butterflies, which are also used during mate choice. Combining behavioral, population genomic, and expression analyses, we show that two Heliconius species have evolved the same preferences for red patterns by exchanging genetic material through hybridization. Neural expression of regucalcin1 correlates with visual preference across populations, and disruption of regucalcin1 with CRISPR-Cas9 impairs courtship toward conspecific females, providing a direct link between gene and behavior. Our results support a role for hybridization during behavioral evolution and show how visually guided behaviors contributing to adaptation and speciation are encoded within the genome.

Complexity of Chemical Emissions Increases Concurrently with Sexual Maturity in Heliconius Butterflies

Pheromone communication is widespread among animals. Since it is often involved in mate choice, pheromone production is often tightly controlled. Although male sex pheromones (MSPs) and anti-aphrodisiacs have been studied in some Heliconius butterfly species, little is known about the factors affecting their production and release in these long-lived butterflies. Here, we investigate the effect of post-eclosion age on chemical blends from pheromone-emitting tissues in Heliconius atthis and Heliconius charithonia, exhibiting respectively free-mating and pupal-mating strategies that are hypothesised to differently affect the timing of their pheromone emissions. We focus on two different tissues: the wing androconia, responsible for MSPs used in courtship, and the genital tip, the production site for anti-aphrodisiac pheromones that affect post-mating behaviour. Gas chromatography-mass spectrometric analysis of tissue extracts from virgin males and females of both species from day 0 to 8 post-eclosion demonstrates the following. Some ubiquitous fatty acid precursors are already detectable at day 0. The complexity of the chemical blends increases with age regardless of tissue or sex. No obvious difference in the time course of blend production was evident between the two species, but female tissues in H. charithonia were more affected by age than in H. atthis. We suggest that compounds unique to male androconia and genitals and whose amount increases with age are potential candidates for future investigation into their roles as pheromones. While this analysis revealed some of the complexity in Heliconius chemical ecology, the effects of other factors, such as the time of day, remain unknown.

Cis-regulatory modes of Ultrabithorax inactivation in butterfly forewings.

Hox gene clusters encode transcription factors that drive regional specialization during animal development: for example the Hox factor Ubx is expressed in the insect metathoracic (T3) wing appendages and differentiates them from T2 mesothoracic identities. Hox transcriptional regulation requires silencing activities that prevent spurious activation and regulatory crosstalks in the wrong tissues, but this has seldom been studied in insects other than Drosophila, which shows a derived Hox dislocation into two genomic clusters that disjoined Antennapedia (Antp) and Ultrabithorax (Ubx). Here, we investigated how Ubx is restricted to the hindwing in butterflies, amidst a contiguous Hox cluster. By analysing Hi-C and ATAC-seq data in the butterfly Junonia coenia, we show that a Topologically Associated Domain (TAD) maintains a hindwing-enriched profile of chromatin opening around Ubx. This TAD is bordered by a Boundary Element (BE) that separates it from a region of joined wing activity around the Antp locus. CRISPR mutational perturbation of this BE releases ectopic Ubx expression in forewings, inducing homeotic clones with hindwing identities. Further mutational interrogation of two non-coding RNA encoding regions and one putative cis-regulatory module within the Ubx TAD cause rare homeotic transformations in both directions, indicating the presence of both activating and repressing chromatin features. We also describe a series of spontaneous forewing homeotic phenotypes obtained in Heliconius butterflies, and discuss their possible mutational basis. By leveraging the extensive wing specialization found in butterflies, our initial exploration of Ubx regulation demonstrates the existence of silencing and insulating sequences that prevent its spurious expression in forewings.

Cis-regulatory modes of Ultrabithorax inactivation in butterfly forewings

Hox gene clusters encode transcription factors that drive regional specialization during animal development: for example the Hox factor Ubx is expressed in the insect metathoracic (T3) wing appendages and differentiates them from T2 mesothoracic identities. Hox transcriptional regulation requires silencing activities that prevent spurious activation and regulatory crosstalks in the wrong tissues, but this has seldom been studied in insects other than Drosophila, which shows a derived Hox dislocation into two genomic clusters that disjoined Antennapedia (Antp) and Ultrabithorax (Ubx). Here, we investigated how Ubx is restricted to the hindwing in butterflies, amidst a contiguous Hox cluster. By analysing Hi-C and ATAC-seq data in the butterfly Junonia coenia, we show that a Topologically Associated Domain (TAD) maintains a hindwing-enriched profile of chromatin opening around Ubx. This TAD is bordered by a Boundary Element (BE) that separates it from a region of joined wing activity around the Antp locus. CRISPR mutational perturbation of this BE releases ectopic Ubx expression in forewings, inducing homeotic clones with hindwing identities. Further mutational interrogation of two non-coding RNA encoding regions and one putative cis-regulatory module within the Ubx TAD cause rare homeotic transformations in both directions, indicating the presence of both activating and repressing chromatin features. We also describe a series of spontaneous forewing homeotic phenotypes obtained in Heliconius butterflies, and discuss their possible mutational basis. By leveraging the extensive wing specialization found in butterflies, our initial exploration of Ubx regulation demonstrates the existence of silencing and insulating sequences that prevent its spurious expression in forewings.

Divergent warning patterns influence male and female mating behaviours in a tropical butterfly.

Traits under divergent ecological selection that also function during mating can be important in maintaining species boundaries. Few studies have considered mutual mate choice, where both males and females base mating decisions on the same trait. Wing colouration in Heliconius butterflies evolved as a warning signal but also functions as a mating cue. We investigated the contribution of visual preference to assortative mating in an aposematic butterfly Heliconius cydno in the context of reproductive isolation with its sympatric, visually distinct relative Heliconius melpomene. Heliconius cydno have conspicuous white bands on their forewings, whereas those of H. melpomene are red in colour. We predicted that both sexes of H. cydno contributed to assortative mating by exhibiting visual preference towards conspecific wing colouration. We analysed published and new data from preference experiments, in which males were presented with conspecific and H. melpomene females. We also recorded female responses and mating outcomes in choice experiments, involving conspecific males with either the original white or artificially painted red forewing bands. Both sexes of H. cydno responded more positively towards the conspecific colouration, and males strongly preferred females of its own colours. In contrast, male colouration did not predict mating outcomes in female choice experiments. As courtships are initiated by males in butterflies, our findings suggest that female visual preference might be of secondary importance in H. cydno. Our data also suggest that the contribution of visual preference to reproductive isolation might be unequal between H. cydno and its sympatric relative H. melpomene.

Metazoa-level USCOs as markers in species delimitation and classification.

Metazoa-level universal single-copy orthologs (mzl-USCOs) are universally applicable markers for DNA taxonomy in animals that can replace or supplement single-gene barcodes. Previously, mzl-USCOs from target enrichment data were shown to reliably distinguish species. Here, we tested whether USCOs are an evenly distributed, representative sample of a given metazoan genome and therefore able to cope with past hybridization events and incomplete lineage sorting. This is relevant for coalescent-based species delimitation approaches, which critically depend on the assumption that the investigated loci do not exhibit autocorrelation due to physical linkage. Based on 239 chromosome-level assembled genomes, we confirmed that mzl-USCOs are genetically unlinked for practical purposes and a representative sample of a genome in terms of reciprocal distances between USCOs on a chromosome and of distribution across chromosomes. We tested the suitability of mzl-USCOs extracted from genomes for species delimitation and phylogeny in four case studies: Anopheles mosquitos, Drosophila fruit flies, Heliconius butterflies and Darwin's finches. In almost all instances, USCOs allowed delineating species and yielded phylogenies that corresponded to those generated from whole genome data. Our phylogenetic analyses demonstrate that USCOs may complement single-gene DNA barcodes and provide more accurate taxonomic inferences. Combining USCOs from sources that used different versions of ortholog reference libraries to infer marker orthology may be challenging and, at times, impact taxonomic conclusions. However, we expect this problem to become less severe as the rapidly growing number of reference genomes provides a better representation of the number and diversity of organismal lineages.

Geographically variable mate preferences shed light on the processes maintaining inversion polymorphism

Abstract An important question in evolution is to understand the mechanisms that maintain phenotypic diversity, despite selection that should drive homogeneity. For example, selection by predators may promote the convergence of colour patterns among defended prey, resulting in the formation of mimetic communities. However, certain aposematic species, such as the Neotropical butterfly Heliconius numata, are polymorphic. In H. numata, wing pattern polymorphism, which is associated with chromosomal inversions, may be maintained via disassortative mating preferences, thought to favour the co-occurrence of individuals with different mimicry phenotypes in Peru. To test whether environmental variation due to geography influences mate choice, here we investigate the occurrence of disassortative mating among the two coexisting forms of H. numata in French Guiana, and its potential role in the maintenance of this polymorphism. Our experimental approach demonstrates that the two forms display weak and slightly asymmetrical disassortative mate preferences. Modelling and simulations suggest that this pattern of mate preference alone is not sufficient to maintain polymorphism, and predict the loss of the choosiest form, unless this form enjoys a survival advantage. In this aposematic species, such an advantage could arise from mimicry, but further studies into the benefits of mimicry and predator generalization are needed to test this hypothesis. More importantly, our results suggest that the balance between selective forces influencing polymorphism may vary across geographical and ecological contexts, and this warrants further study.

A Butterfly-Inspired Multisensory Neuromorphic Platform for Integration of Visual and Chemical Cues.

Unisensory cues are often insufficient for animals to effectively engage in foraging, mating, and predatory activities. In contrast, integration of cues collected from multiple sensory organs enhances the overall perceptual experience and thereby facilitates better decision-making. Despite the importance of multisensory integration in animals, the field of artificial intelligence (AI) and neuromorphic computing has primarily focused on processing unisensory information. This lack of emphasis on multisensory integration can be attributed to the absence of a miniaturized hardware platform capable of co-locating multiple sensing modalities and enabling in-sensor and near-sensor processing. In this study, this limitation is addressed by utilizing the chemo-sensing properties of graphene and the photo-sensing capability of monolayer molybdenum disulfide (MoS2) to create a multisensory platform for visuochemical integration. Additionally, the in-memory-compute capability of MoS2 memtransistors is leveraged to develop neural circuits that facilitate multisensory decision-making. The visuochemical integration platform is inspired by intricate courtship of Heliconius butterflies, where female species rely on the integration of visual cues (such as wing color) and chemical cues (such as pheromones) generated by the male butterflies for mate selection. The butterfly-inspired visuochemical integration platform has significant implications in both robotics and the advancement of neuromorphic computing, going beyond unisensory intelligence and information processing.

Major patterns in the introgression history of Heliconius butterflies

Gene flow between species, although usually deleterious, is an important evolutionary process that can facilitate adaptation and lead to species diversification. It also makes estimation of species relationships difficult. Here, we use the full-likelihood multispecies coalescent (MSC) approach to estimate species phylogeny and major introgression events in Heliconius butterflies from whole-genome sequence data. We obtain a robust estimate of species branching order among major clades in the genus, including the ‘melpomene-silvaniform’ group, which shows extensive historical and ongoing gene flow. We obtain chromosome-level estimates of key parameters in the species phylogeny, including species divergence times, present-day and ancestral population sizes, as well as the direction, timing, and intensity of gene flow. Our analysis leads to a phylogeny with introgression events that differ from those obtained in previous studies. We find that Heliconius aoede most likely represents the earliest-branching lineage of the genus and that ‘silvaniform’ species are paraphyletic within the melpomene-silvaniform group. Our phylogeny provides new, parsimonious histories for the origins of key traits in Heliconius, including pollen feeding and an inversion involved in wing pattern mimicry. Our results demonstrate the power and feasibility of the full-likelihood MSC approach for estimating species phylogeny and key population parameters despite extensive gene flow. The methods used here should be useful for analysis of other difficult species groups with high rates of introgression.

Major patterns in the introgression history of Heliconius butterflies.

Gene flow between species, although usually deleterious, is an important evolutionary process that can facilitate adaptation and lead to species diversification. It also makes estimation of species relationships difficult. Here, we use the full-likelihood multispecies coalescent (MSC) approach to estimate species phylogeny and major introgression events in Heliconius butterflies from whole-genome sequence data. We obtain a robust estimate of species branching order among major clades in the genus, including the 'melpomene-silvaniform' group, which shows extensive historical and ongoing gene flow. We obtain chromosome-level estimates of key parameters in the species phylogeny, including species divergence times, present-day and ancestral population sizes, as well as the direction, timing, and intensity of gene flow. Our analysis leads to a phylogeny with introgression events that differ from those obtained in previous studies. We find that Heliconius aoede most likely represents the earliest-branching lineage of the genus and that 'silvaniform' species are paraphyletic within the melpomene-silvaniform group. Our phylogeny provides new, parsimonious histories for the origins of key traits in Heliconius, including pollen feeding and an inversion involved in wing pattern mimicry. Our results demonstrate the power and feasibility of the full-likelihood MSC approach for estimating species phylogeny and key population parameters despite extensive gene flow. The methods used here should be useful for analysis of other difficult species groups with high rates of introgression.

Parallel shifts in flight-height associated with altitude across incipient Heliconius species.

Vertical gradients in microclimate, resource availability, and interspecific interactions are thought to underly stratification patterns in tropical insect communities. However, only a few studies have explored the adaptive significance of vertical space use during the early stages of reproductive isolation. We analysed flight-height variation across speciation events in Heliconius butterflies, representing parallel colonizations of high-altitude forest. We measured flight-height in wild H. erato venus and H. chestertonii, parapatric lowland and mountain specialists, respectively, and found that H. chestertonii consistently flies at a lower height. By comparing our data to previously published results for the ecologically equivalent H. e. cyrbia (lowland) and H. himera (high altitude), we found that the species flying closest to the ground are those that recently colonized high-altitude forests. We show that these repeated trends largely result from shared patterns of ecological selection producing parallel trait-shifts in H. himera and H. chestertonii. Although our results imply a signature of local adaptation, we did not find an association between resource distribution and flight-height in H. e. venus and H. chestertonii. We discuss how this pattern may be explained by variations in forest structure and microclimate. Overall, our findings underscore the importance of behavioural adjustments during early divergence mediated by altitude-shifts.