Color and shape nonadditively enhance leaf masquerade in the wild

To determine how leaf masquerade improves individual fitness under ecologically relevant conditions, we exposed wild avian predators to artificial wing targets varying in color to mimic naturally occurring green leaves, the brown bark substrate upon which they were pinned, or an unnatural blue control (Fig A in S1 Text; see Materials and methods). Doing so meant we could confirm that any survival benefits incurred by these stimuli were a result of misclassification of the targets and not due to either enhanced crypsis of the targets against their background or to neophobic responses. Models of each color also varied in shape to mimic an oval leaf, an elongate leaf, or a rectangular control (Fig 1B), and were designed based on morphological measurements taken from the tegmina of an existing leaf-masquerading katydid species (see Materials and methods). Survival of each individual target at sampling was assessed based on the presence or absence of an edible mealworm bait, which served as the “body” of the target. To minimize potential effects of predator learning through search image formation—a phenomenon known to improve foraging efficiency in birds [17–20] —we deliberately performed the experiment in a locality where avian predators are naïve to leaf-masquerading insects. This approach, which mimics previous studies testing the adaptive value of protective coloration in tropical species using naïve predators in UK field environments [21,12], recreates the ecological conditions under which leaf masquerade first conferred a selective advantage, allowing us to isolate its functional properties, independently of learned recognition. Since masquerade relies on the implicit assumption that prey are misclassified as their models, performing this experiment in a community of “inexperienced” predators that do not routinely encounter leaf-masquerading prey enabled us to better ensure a leaf-like prey object would be classified as unprofitable by an insectivorous bird. Overall, 39.6% of targets (513) were attacked by birds (51.9% showed signs of nonavian predation, 1.9% were lost, 6.3% survived until the end of the trial). We found that the specific combination of green color and oval shape increased fitness, rather than each trait improving survival in isolation (Fig 1C and Fig B in S1 Text).

The interaction between color and shape predicted survival rate (mixed-model Cox regression: χ²₄ = 9.882, N = 1,296, p = 0.043), but neither factor was significant as a main effect (mixed-model Cox regression: color, χ²₂ = 2.147, N = 1,296, p = 0.342; shape, χ²₂ = 5.878, N = 1,296, p = 0.053). Furthermore, color affected survival rate only for oval-shaped targets (mixed-model Cox regression: oval, χ²₂ = 7.143, N = 432, p = 0.028; elongated oval, χ²₂ = 2.058, N = 432, p = 0.357; rectangle, χ²₂ = 2.814, N = 432, p = 0.245; Fig 1C and Table A in S1 Text for post-hoc tests). The “green oval” targets did not match the background color, and their shape were specifically designed to closely resemble that of an existing, but not local, leaf-masquerading katydid species, and likely other locally abundant leaves at the study site. Therefore, their increased survival compared to targets with brown, background-matching coloration and those with atypical leaf shapes strongly suggests that their survival advantage resulted from predators misclassifying them. This field experiment thus confirms an assumption central to the definition of masquerade that distinguishes it from other forms of camouflage [9,10,22].

Composite trait evolution in Neotropical leaf-masquerading katydids

Our field predation experiment suggests that leaf masquerade is a composite trait reliant on the synergistic effects of color and shape for its adaptive function of causing predator misclassification. To assess whether color and shape evolved synergistically in real leaf-masquerading species, we studied the tegmina of 288 wild-caught individuals of 58 sympatric katydid species (196 male, 92 female; see Table B in S1 Text) from a diverse Panamanian rainforest community (Fig 2A). The coloration of each species was classified (and independently verified; see Materials and methods) as being green or brown in color and pigmented or not (Fig 2B; katydid tegmina can be translucent or opaquely pigmented irrespective of their general color, with the former being the most likely ancestral state for Tettigoniidae). Our particular interest was in tegmina that were green and pigmented, since green targets provided better masquerade in the avian predation experiment above. Spectrophotometry from a random subset of tegmina confirmed minimal ultraviolet (UV) reflectance, indicating that the appearance of the artificial targets used in the avian predation experiment are comparable to real katydid tegmina, in that both reflect minimal UV (Fig C in S1 Text; see Materials and methods). All studied katydid species were cryptic in coloration and available data suggests they occupy similar rainforest microhabitats [24]. Therefore, evolutionary shifts in color and pigmentation do not appear to be driven by habitat-specific background matching.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Synergistic evolution explains the origins of leaf masquerade in katydids.

(A) A continuous trait plot reconstructing patterns of human perceived “leafiness” (“mean leafiness score”) per species across a pruned molecular phylogeny of 58 katydid species [23]. The presence (dark green) and absence (gray) of green pigmentation is shown at the tips, and the mean tegmen aspect ratio (au) per species is indicated by black horizontal bars. Exemplars from each subfamily (highlighted blocks) are shown on the right. Photos courtesy of Ciara Kernan, Dr Laurel Symes and Dr Hannah ter Hofstede. (B) Transition rates between coloration states (above: green pigmentation, below: pigmentation). Arrow thickness is scaled by the transition rate value which is also given beside each arrow. (C) Nonadditive effects of tegmen pigmentation (presence: dark gray, absence: light gray) and mean tegmen aspect ratio on the mean leafiness score per species (N = 58). The inset shows mounted tegmina of the species with the lowest (Caulopsis micropora) and highest (Aegimia maculofolia) mean aspect ratio, with the measured minor (W) and major (L) axis labeled for each. Scale bars = 5 mm. (D) Evolutionary association between tegmen aspect ratio (N = 288) and the presence and absence of green tegmen pigmentation across the 58 katydid species. Thick bars indicate medians, boxes indicate interquartile ranges, and whiskers show values within 1.5 interquartile ranges. (E) Phylogenetic path analysis illustrating the directional and simultaneous evolutionary acquisition of pigmentation, green coloration, and reduced aspect ratio during the evolution of leaf masquerade. Human perceived “leafiness score” (out of 10) per species is the response variable. Arrows indicate the direction of the interaction, and values adjacent to these arrows are the respective path coefficients indicating the strength of the estimated relationship. Positive coefficients (blue arrows) indicate a transition from absence to presence in the parent trait (tegmen pigmentation and green coloration) whereas negative coefficients (red arrows) indicate a decrease in the parent trait (mean tegmen aspect ratio). All data underlying the graphs shown in this figure can be found in https://doi.org/10.5281/zenodo.14585161.

https://doi.org/10.1371/journal.pbio.3003468.g002

Shape was quantified by measuring the tegmen aspect ratio (maximum length/ maximum width; see Materials and methods). Thus, species with wider, more rounded, “leaf-like” tegmina had lower aspect ratios [25] (Fig 2C). The ecological relevance of our shape treatments in the avian predation experiment was supported by the observation that the species with the lowest aspect ratio (Aegimia maculofolia, aspect ratio = 2.177) closely matched that of the ‘oval’ treatment in our field experiment, while the species with the highest aspect ratio (Caulopsis micropora, aspect ratio = 8.841) closely matched that of the ‘elongated oval’ treatment. Expanding upon previous categorisations of Neotropical katydids [15], two “human leafiness metrics” were obtained by asking 64 naïve human participants to both classify (presence versus absence) and score (out of 10) leafiness of the 58 species’ tegmina via an online survey (see Materials and methods). Unlike previous categorisations of leaf masquerade that relied on specific morphological features [4,14], scoring “leafiness” in this way provides a holistic estimate of how leaf masquerade is perceived, without relying on any one component trait. Using a recent molecular phylogeny [23], we then reconstructed the evolution of leafiness and its shape and color component traits.

We found that the evolution of leaf masquerade is under strong phylogenetic constraint. Both green pigmentation and shape show a strong phylogenetic signal (green pigmentation, χ²₁ = 0.000, λ = 1.000, p = 1.000; mean aspect ratio, χ²₁ = 46.132, λ = 1.042, p < 0.001) and transition rates between presence and absence of green pigmentation are equivalent (Fig 2B; Fig D in S1 Text and Tables C and D in S1 Text). The most likely pattern of aspect ratio evolution followed a Brownian motion (BM) model (BM versus Ornstein–Uhlenbeck [OU]:ΔAkaike’s information criterion (ΔAIC) = −2.226, BM versus early burst ΔAIC = −1.786; Fig D in S1 Text). Modeling the human leafiness metrics in the same way showed a similarly high phylogenetic signal, but this time, an OU model was best fitting (Table D in S1 Text). These results indicate a strong degree of evolutionary inertia, potentially limiting the axes of adaptive morphological evolution among leaf-masquerading species [26]. Despite these constraints, the best-fit OU model for overall leafiness suggests that this trait has evolved under directional selection across the phylogeny. Consistent with previous work, ancestral state reconstructions support multiple independent elaborations and reductions of ‘leafiness’ and both its component traits across the 58 studied species [14] (Fig 2A).

Coloration and shape have nonadditive effects on human-perceived “leafiness”

In the face of this phylogenetic constraint, we sought to test whether the functional synergy of coloration and shape, indicated by our predation experiment, is concordant with human perceptions of leafiness. While humans are not natural predators of Neotropical katydids, many insectivorous New World primates are [27–29], and behavioral experiments have suggested conserved mechanisms underlying object recognition across birds and primates [30–32]. Because tegmina were presented to human participants against a white matte background (see Materials and methods), there is no biological reason why green tegmina were more likely to be classified as leaf-like over brown tegmina. Therefore, accounting for phylogenetic effects, we tested whether human perceptions of leaf masquerade depended on the interaction between shape and the presence or absence of notable tegmen pigmentation, independent of color.

When individual participant data were analyzed using phylogenetic generalized linear mixed models, the interaction between mean aspect ratio and pigmentation predicted the human leafiness score, echoing results from the avian predation experiment (MCMCglmm: P-mean = −3.840, 95% CI = −7.522 to −0.136, P_MCMC= 0.045; Fig 2A and 2C and Table E in S1 Text). When species were subset by pigmentation status, participants consistently awarded high leafiness scores to low aspect ratio tegmina, but the strength of this association was stronger for pigmented species (MCMCglmm: pigmented, P-mean = −8.930, 95% CI = −10.495 to −7.234, P_MCMC< 0.001; not pigmented, P-mean = −6.144, 95% CI = −11.705 – −9.959, P_MCMC= 0.025). These results were confirmed by equivalent models constructed using phylogenetic generalized least-squares (PGLS; see Table F in S1 Text). No interaction between pigmentation and shape was found when human-perceived leaf masquerade presence/absence was used as the dependent variable rather than “leafiness” score, possibly because the former metric was too coarse to capture continuous variation in leafiness (MCMCglmm: P-mean = −4.772, 95% CI = −10.805 to 1.531, P_MCMC= 0.120; Tables E and F in S1 Text). While many New World primate predators of katydids are also trichromats, thus enhancing the ecological relevance of our human-based assessments of leafiness, these metrics might represent a conservative estimate of how color and shape interact in the eyes of dichromatic species, whose visual systems are thought to improve the ability to find camouflaged prey [29,33,34]. Overall, in concordance with the conclusions derived from our artificial prey experiment, perception of leaf masquerade appears reliant on the co-occurrence of appropriate coloration, shape, and possibly other independent components, for exploiting both avian and mammalian recognition systems.

Coloration and shape co-evolve synergistically to produce spectacular leaf masquerade

Given that the interaction between coloration and shape enhances the efficacy of leaf masquerade, we next tested whether nonadaptive stepwise or simultaneous acquisition of these trait components explain its evolution. In other words, were coloration and shape selected independently at different timepoints, one maintained by selection for functions unrelated to leaf masquerade and later co-opted to form a composite masquerade trait with the other? Or did they both evolve simultaneously and result in this form of camouflage? Nonadaptive stepwise evolution would imply no evolutionary correlation between coloration and shape across the phylogeny—that is, these traits would not tend to arise at the same time during the evolutionary history of these species because neither would enhance masquerade in isolation. The opposite is predicted if these traits evolved simultaneously—that is, they should tend to arise concurrently during the species’ evolutionary history. After controlling for phylogeny, MCMCglmm analysis on individual data and a phylogenetic ANOVA on species means confirmed that the evolution of green pigmentation was significantly associated with reduced tegmen aspect ratios (MCMCglmm: P-mean = 0.062, 95% CI = 0.004 to 0.117, P_MCMC= 0.035; phylogenetic ANOVA: F₁ = 16.127, p = 0.025; Fig 2D). These results are consistent with an evolutionary scenario whereby coloration and shape simultaneously evolved under directional selection. This model also revealed that female tegmina exhibit significantly lower aspect ratios compared to males (P-mean = −0.014, 95% CI = −0.023 to −0.005, P_MCMC= 0.002), suggesting that males and females likely experience different selection pressures, but the interaction between green pigmentation and sex was not significant (P-mean = −0.002, 95% CI = −0.016 to 0.012, P_MCMC= 0.847).

Given the evolutionary correlation between trait components, we next assessed the order of pigmentation, green coloration, and aspect ratio acquisition across the phylogeny using phylogenetic path analysis (PPA). This method involved constructing structural equation models (SEMs) to represent alternative evolutionary sequences by which these traits were acquired stepwise across the phylogeny. We also modeled a scenario in which all three traits were acquired concurrently, as well as hybrid models that combined elements of simultaneous and stepwise acquisition of the three traits. When both leaf masquerade probability and mean leafiness score were used as the dependent variable, the path analysis yielded strong support for models of simultaneous acquisition and loss of all three traits, supporting their coordinated, synergistic evolution during the origin of leaf masquerade (Fig 2E; Fig E in S1 Text; Table G in S1 Text). Taken together, the evolution of leaf masquerade in our studied species is best explained by the simultaneous, synergistic acquisition and loss of trait components, under directional selection.

Conclusions

Functional synergy between color and shape of leaf-like prey causes misclassification by naïve predators, and phylogenetic reconstruction of these component traits supports their simultaneous, synergistic evolution in Neotropical katydid tegmina. Our field experiment demonstrates that resemblance in both color and shape is necessary for this type of masquerade to be functional [35]. While we isolate and demonstrate the functional benefits of masquerade in a natural context and illustrate the synergistic evolution of its component traits, other sources of selection are likely to act upon the tegmina of Neotropical katydids, including sexual selection, as male tegmina are used to produce sound, and other forms of ecological selection. The latter might explain the sex differences in tegmen shape that we observed. In nature, masquerade likely also co-evolves with behavioral adaptations such as resting orientation and microhabitat selection, as has been observed in other masquerading organisms [8,36–38]. Although selection for background matching may have contributed towards the evolution and maintenance of leaf masquerade, disentangling relative strengths of these effects, particularly with little ecological data, is challenging. It is irrelevant to our arguments how much leaf masqueraders benefit from crypsis through background matching; their often-uncanny specific resemblance to leaves suggests an additional benefit through being misidentified.

By integrating experimental evidence with comparative analyses of extant species, we demonstrate how composite adaptations can originate through the simultaneous recruitment of individual trait components, rather than by stepwise evolution. Our findings validate and expand upon patterns of contingent wing pattern evolution observed in leaf-masquerading butterflies [39,40], providing a plausible evolutionary mechanism for explaining the origin of integrated phenotypes, leaving its genetic basis an alluring topic of future study. In many species, leaf masquerade appears to have been further optimized through the evolution of elaborate venation patterning, false necrotic spots, and false holes [15,21,41,42]. The sequence and pattern by which these more sophisticated leaf-specific traits are acquired across species, and their relative fitness advantages, warrants further investigation. The adaptive significance and evolutionary mechanisms underlying within-species polymorphisms remain an important avenue for further targeted study [15]. Color polymorphisms in particular are common in many of our sampled species when comparing populations across their geographic range, but here we used species for which intraspecific polymorphism was absent at our sampling locality, enabling species-level phylogenetic comparative analyses. Although not part of our analysis, genera within the subfamily Pterochrozinae such as Mimetica and Typophyllum are exceptional examples of leaf masquerade, and these do exhibit striking within-species differences in both coloration and shape, even within the same population [15,42]. It is not known whether this variation is heritable or the result of environmentally induced plasticity, However, such extreme phenotypic lability might hint that color and shape have become evolutionarily decoupled in this lineage. Testing this will require large within-species sample sizes. It would also be interesting to test if predator learning shapes the relative fitness of these morphs among communities of experienced predators.

To conclude, the appearance of exquisitely sophisticated ‘design’ in the natural world is an enduring issue in evolutionary biology that propels both historical and contemporary debates [4,5,43]. While the dominant, gradualist paradigm envisions sequential optimization of adaptive traits that eventually build complex functions, we suggest that alternative scenarios warrant more serious investigation; that is, those involving the co-occurrence of traits with synergistic effects, which may be rarer, but subject to stronger selection.

Wild katydid sampling

Adult katydids were sampled at night from lights around research station buildings on Barro Colorado Island (BCI), Panama (9.1647° N, 79.8367° W) in March–April 2024 and January–May 2025 between 04:00 and 06:00 and 20:00 and 01:00. Individuals were identified to species level using public resources [44,45] and custom ID keys provided by Dr Laurel Symes and Dr Hannah ter Hofstede. A total of 288 individuals were sampled across 58 species (51 species were sampled in 2024, with a further 7 species sampled in 2025; Fig B in S1 Text), with representatives from subfamilies Conocephalinae, Phaneropterinae and Pseudophyllinae (Figs 1A and 2A). We collected more than three individuals of 31 species. Studying a single diverse community of katydids with seemingly generalist habitat preferences [24] allowed for robust comparative analyses by eliminating variation in background vegetation composition, which might influence the evolution of different leaf masquerading strategies.

Captured individuals were placed in a freezer at −20°C for 10 min, after which morphological data were collected (body length, femur length/width, thorax height). Whole tegmina were removed with microscissors. The external surface of the tegmina was photographed using a Nikon D3300 camera (Nikon Corporation, Tokyo, Japan) with a Sigma 17–50 mm f/2.8 lens (Sigma Corporation, Kanagawa, Japan) in a photography light box (DUCLUS, model DU5032U, Guangdong, China) under LED illumination. Images were taken against a matte white background, at a constant distance (~25 cm) and magnification, and contained a ColorChecker passport (X-Rite Grand Rapids, MI, USA) for future coloration analysis. Tegmina were preserved as dry specimens, and the remaining body tissue was either preserved in 95% ethanol or contributed to the Museo de Invertebrados Fairchild de la Universidad de Panamá (MIUP). Samples were exported to the United Kingdom under export permit collection nos. PA-01-ARG-040-2024 and PA-01-ARB-039-2025, obtained from the Ministerio de Ambiente, Panama.

Field experiment

Comparative morphology