henhaobio
Sign in

Philosophical Transactions of the Royal Society B Biological SciencesAnimal Behavior & CognitionPeer-reviewed journal article

Modality specific memory enhancement in Heliconius butterflies

Elizabeth A. HodgeiD, Amaia Alcalde AntoniD, Louise BesteaiD, Greta HernándeziD, Jane Margereth AguilariD, Max S. FarnworthiD, Denise Dalbosco Dell’AglioiD, W. Owen McMillaniD, Stephen H. MontgomeryiD

PDFDOIPreprint
In-depth full-text digest available · Jump to digest ↓

Abstract

How animals perceive, process and respond to environmental cues is tightly tuned to the species-specific demands, and reflected by the structure of neural systems. In the Neotropical butterflies, Heliconius, the mushroom bodies, insect learning and memory centres, are significantly expanded compared to their closest relatives. This expansion coincided with the evolution of a novel diet, pollen feeding, and a spatial foraging behaviour consistent with trap-lining. Previous research has shown that Heliconius have more accurate visual long-term memory than other Heliconiini. Here, we tested whether this enhanced memory stability is specific to visual cues by conducting a long-term olfactory memory assay in two Heliconius species and two outgroup species. We found no differences in the long-term olfactory memory between Heliconius species and the outgroup species. Combining data from olfactory and visual memory trials confirms a significant shift in performance among sensory modalities between Heliconius and outgroup genera. In contrast, tests of how Heliconiini prioritise olfactory and visual cues when in presented in conflict also show no consistent shift in attentiveness to sensory cues between species. Our data provide a rare case where memory performance has been compared across species and sensory modalities, to identify evidence of a modality specific shift.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (15 Sept 2024) · CC BY 4.0

Heliconius butterflies retain odour associations over eight days no better than outgroup Heliconiini, contrasting with their superior visual long-term memory, consistent with a modality-specific memory enhancement.

Heliconius mushroom bodies are roughly four times the size of those in related Heliconiini, and earlier assays found more durable visual memory in Heliconius, though a general memory gain across senses had not been ruled out. Repeating that visual protocol with odours in the same species gives a rare comparison of memory across both species and sensory modalities.

Key results

  1. ShownAll four species learned odour-reward associations and stayed above chance after eight days without odour exposure. Long-term olfactory memory did not differ among species, and Heliconius did no better than the outgroups.

    Naive-to-recall preference shift: χ² = 400.817, d.f. = 1, p < 0.001. Species did not differ in learning (χ² = 3.024, d.f. = 3, p = 0.388) or in long-term memory (LTM) after eight days (χ² = 2.445, d.f. = 3, p = 0.485). Being Heliconius gave no LTM advantage (χ² = 0.280, d.f. = 1, p = 0.597). Accuracy fell from recall to LTM (χ² = 57.398, d.f. = 1, p < 0.001) but stayed above chance. · i. Proficient long-term memory of olfactory…

  2. SuggestedPooling these data with the earlier visual experiment shows a modality shift. Heliconius long-term visual memory was better than both their own olfactory memory and the outgroups' visual memory, whereas the outgroups retained odours better than colours.

    Pooled binomial generalised linear mixed model (GLMM) with group or species, trial and experiment as fixed effects. Figure 1 values come from a model with a significant group × trial × experiment interaction. Visual learning matched or exceeded olfactory learning across species (χ² = 28.331, d.f. = 1, p < 0.001). Groups did not differ in olfactory memory (χ² = 2.579, d.f. = 1, p = 0.612). · ii, Long-term memory recall varies across…; iv. Statistical analysis; Figure legends

  3. SuggestedWithin species, both Heliconius species were more accurate with colour than with odour at recall and in the long-term test. D. iulia learned both cues equally and scored higher on odour in the long-term test. A. vanillae learned colour better but retained both cues equally.

    The species-level contrasts from the pooled GLMMs are reported in Tables S6, S7 and Figure S5. The main text gives no test statistics for them. · ii, Long-term memory recall varies across…

  4. ShownWhen colour and odour cues conflicted, H. erato chose the trained colour above chance, H. melpomene leaned the same way without reaching significance, and D. iulia showed no preference. Species did not differ significantly, so cue prioritisation did not shift consistently.

    All three species learned the colour-odour pairing (Tables S8, S9), and choices in the conflict test differed from recall (Table S8). Conflict vs colour-only test: H. erato estimate = -1.334, z ratio = -2.435, p = 0.040; H. melpomene estimate = -2.157, z ratio = -2.440, p = 0.039; D. iulia estimate = -0.766, z ratio = -0.922, p = 0.626. No species effect in the conflict test (Table S9). · iii. No consistent shifts in prioritisation of…; iii. Conflict assay

  5. ShownIn later trials, and most clearly in the conflict test, D. iulia made fewer feeding attempts per individual than either Heliconius species. The authors read this as a possible sign of greater uncertainty when deciding.

    A Poisson GLMM of attempts per individual gave D. iulia an estimate of -1.28 relative to H. erato (z ratio -5.627, p < 0.001) and -1.211 relative to H. melpomene (z ratio -4.332, p < 0.001); see Table S11. · iii. No consistent shifts in prioritisation of…; iv. Statistical analysis

Methods, data and reproducibility

Butterflies were trained for four days to link lemongrass or citrus odour with a reward versus quinine, then tested at recall and again after eight days without the odours, mirroring an earlier visual memory protocol. A second assay trained colour-odour pairs and then set the learned cues in conflict. Feeding attempts were scored from video in BORIS and analysed with binomial generalised linear mixed models (GLMMs).

  • Olfactory LTM assay butterflies, Smithsonian Tropical Research Institute insectaries, Panama (2023); 37 non-learners removed before analysis (10, 13, 8 and 6, in the species order of n) · Heliconius erato, Heliconius melpomene, Agraulis vanillae, Dryas iulia · n = Tested: 65 A. vanillae, 50 D. iulia, 45 H. erato, 51 H. melpomene
  • Colour-odour conflict assay butterflies raised the same way (2024) · Heliconius erato, Heliconius melpomene, Dryas iulia
  • Visual LTM data from Young et al. (2024): the olfactory assay followed that protocol as closely as possible, with four of the same populations/species in another year; pooled here (Table S5) · Heliconius erato, Heliconius melpomene, Agraulis vanillae, Dryas iulia
Methods & tools
BORIS, GoPro Hero 5, R v4.3.0, lme4 v1.1-33, DHARMa v0.4.6, emmeans v1.8.6
Reagents & environment
R v4.3.0 with lme4 v1.1-33 (glmer), DHARMa v0.4.6 and emmeans v1.8.6
Reproducibility gaps
The text has no statement on code or data availability and no links to either. Many test statistics are only in supplementary tables (S1 to S13), which are not in the text. The conflict-assay sample sizes are not stated.
Limitations (5)
  • Testing several free-flying species is labour-intensive, which limited the stimuli to two ecologically neutral odours and two equally non-preferred colours. · stated by the authors
  • The olfactory and visual memory experiments were run in different years. Pooling them relies on the same cage design, environmental conditions and wild-derived stock origin. · stated by the authors
  • It is still unknown whether better visual long-term memory in Heliconius comes from more Kenyon cells or from changed plasticity. The share of Kenyon cells receiving visual versus olfactory input has not been delineated. · stated by the authors
  • Only two Heliconius and two outgroup species were tested, and the conflict assay had a single outgroup (D. iulia) because A. vanillae stocks were unavailable. The genus-level contrasts therefore rest on few lineages. · noted in this summary
  • The main text does not give sample sizes for the conflict assay, so readers cannot judge the statistical power behind its finding of no species difference. · noted in this summary

Read this if you study the evolution of insect cognition or mushroom body specialisation and want a cross-species test of whether a memory gain is confined to one sense.

Citations
12
FWCI
3.78
Field percentile
top 6%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISHodge '26Young '23Young '23Toure '20Borrero '24Foley '26
83% semantic matchEvolutionary Biology
Elizabeth A Hodge, Denise Dalbosco Dell’Aglio, Antoine Couto+3 more·Evolution·Jan 17, 2026· 2 citations