Association of flower color with pollen reward may explain increased bumblebee visitation to the Scotch broom yellow morph

Given that pollinators usually visit flowers for hidden rewards, they need to rely on floral traits that indicate reward status (“honest signals”). However, the relationship between pollination, honest signals, and floral rewards is little documented in natural conditions. The Scotch broom (Cytisus scoparius) is an invasive shrub with polymorphism in the color of its flowers that can be yellow, orange, or red. In three areas dominated by the Scotch broom, we described the abundance of the floral morphs and estimated bumblebee (Bombus terrestris) visitation rate. We examined whether bumblebee visitation to the floral morphs was related to pollen reward. We collected flowers and classified their stamens according to their function: reward or pollen export. Then, we measured anther size and estimated pollen quantity. The yellow morph was more abundant and more visited by bumblebees than the orange and red morphs. The yellow flowers did indeed offer more pollen than the other morphs and this occurred only for rewarding anthers, suggesting that bumblebees could use yellow color as an honest signal to visit the most rewarding flowers. We discuss whether innate and/or learned preferences of bumblebees can explain why the yellow morph is more visited, pollinated, and abundant, while the other morphs are maintained at a lower frequency. This is one of the few field works that shows that variation in intra-specific floral traits is associated with variation in floral reward and pollinator visitation rate, helping to understand the foraging preferences of pollinators and the coexistence of floral morphs in nature.

Clinical trials registration: Not applicable.

     

Fruit colour polymorphism in Acacia ligulata: seed and seedling performance, clinal patterns, and chemical variation

Fruit colour polymorphisms are widespread in nature, but their ecological and evolutionary dynamics remain poorly understood. Here we examine Acacia ligulata, a shrub of the Australian arid zone which exhibits a red/orange/yellow aril colour polymorphism. We asked whether the polymorphism had a genetic basis; whether selection acted differentially on morphs during the seed and seedling stages; whether geographic variation in morph frequencies was correlated with environmental factors; and whether morphs differed in physical or chemical characteristics that might influence selection on them. When grown to maturity in a common greenhouse environment, maternal families of seeds showed phenotypic patterns consistent with biparental genetic control of the polymorphism. In contrast to other fruit-colour polymorphic species, progeny of A. ligulata morphs did not vary in rates of seedling emergence or survival in a common garden. Sampling along a 580 km transect revealed clinal variation in morph frequencies. Frequencies of the yellow morph decreased, and frequencies of the red morph increased, across a gradient of decreasing temperature and increasing rainfall. Morphs did not differ in seed mass, aril mass, or in profiles of fatty acids and flavonoids in either arils or seeds. However, morphs showed consistent differences in carotenoid profiles' and elemental content of arils, suggesting that selection by avian and insect seed dispersers, seed predators and herbivores should be investigated. These patterns indicate that both abiotic and biotic factors may contribute to selection on the A. ligulata polymorphism. This revised version was published online in August 2006 with corrections to the Cover Date.     

Density-dependent and frequency-dependent selection by bumblebees Bombus terrestris (L.) (Hymenoptera: Apidae)

The behaviour of bumblebee workers foraging on arrays of artificial flowers of two colour morphs was observed. Experiments were conducted on arrays of varying morph frequencies and at three different total flower densities. Bumblebees consistently showed a preference for the commonest colour morph, and this behaviour was not significantly affected by changing density. In contrast, frequency-independent preferences changed significantly with density. At low densities, there was a strong bias towards the more conspicuous colour, whereas at higher densities there was no overall colour bias. Flight distances between flowers decreased significantly at high density. Bumblebees also visited flowers of similar colours sequentially, but this behaviour was not density-dependent. It is suggested that as densities increase, there is an increased probability that bumblebees detect yellow flowers, which were probably less conspicuous compared with blue flowers, and that this might be caused by changes in flight speed with flight distance. Where there is a positive relationship between pollinator visitation and the relative fitness of a floral morph, the observed behaviour would induce positive frequency-dependent selection on a plant population with two corolla colour morphs on which the bumblebees were foraging, which would result in stabilizing selection for a single corolla colour, irrespective of density. There was no indication that rare colour morphs would be preferred at high density. The probability of different corolla colour morphs going to fixation would, however, be affected by density.     

Is the colour dimorphism inDactylorhiza sambucina maintained by differential seed viability instead of frequency-dependent selection?

The European rewardless, bee-pollinated orchidDactylorhiza sambucina commonly produces yellow-flowered and purple-flowered individuals in frequencies that range from balanced (per population) to very unbalanced, with parts of the species’ range entirely monochromatic. We studied male and female reproductive success of the two morphs in 22 populations in the Czech Republic, relating it to morph frequency, population size and density, and presence and abundance of yellow and purple co-flowering nectar-providing species visited by the same bee species. Cumulative abundances of yellow nectar-producing co-flowering species (of which, on average,Primula veris made up 56%) had a negative effect on male reproductive success of the yellow morph, and spectral analyses showed that to bumblebees the colours ofP. veris and yellowD. sambucina are different, permitting ready visual discrimination. The cumulative abundance of purple co-flowering species had no significant effect on morph reproductive success. Morph frequencies were unrelated to reproductive success and population size, and there was no evidence of frequency-dependent selection except in one highly unbalanced population. Density of flowering conspecifics was negatively correlated with male reproductive success of the purple morph. Seed mass, viability, and germination success depended on whether seeds resulted from outcrossed or selfed matings and on morph colour. Selfed seeds and seeds produced by the yellow morph from yellow × yellow and yellow × purple crosses had zero germination (after three months), providing the first hint that differential vegetative fitness, rather than differential reproductive fitness via pollinator selection, may explain morph frequencies inD. sambucina.     

Frequency-dependent selection by pollinators: mechanisms and consequences with regard to behaviour of bumblebees Bombus terrestris (L.) (Hymenoptera: Apidae)

Bombus terrestris , a typical pollinating insect species, was offered artificial flowers of two different corolla colours with the same sucrose solution reward in an array. Common colours were significantly preferred, and the strength of the frequency-dependent response increased as a result of learning. There were also frequency-independent biases towards blue flowers, probably because blue flowers appeared more conspicuous to bumblebees than yellow flowers, and the degree of preference for blue was greater when flowers had low nectar rewards. Flower-to-flower movements by individual bumblebees between flowers were non-random, were biased to movements within the same flower colour, and were also dependent on morph frequency. The mechanisms governing flower selection in bumblebees are discussed. Pollinators foraging similarly in a natural situation would induce positive frequency-dependent selection, assortative mating, and directional selection on different corolla colour morphs of the plant population being visited, resulting in stabilizing selection for a single flower colour.     

A hypothesis to explain accuracy of wasp resemblances

Mimicry is one of the oldest concepts in biology, but it still presents many puzzles and continues to be widely debated. Simulation of wasps with a yellow‐black abdominal pattern by other insects (commonly called “wasp mimicry”) is traditionally considered a case of resemblance of unprofitable by profitable prey causing educated predators to avoid models and mimics to the advantage of both (Figure 1a). However, as wasps themselves are predators of insects, wasp mimicry can also be seen as a case of resemblance to one's own potential antagonist. We here propose an additional hypothesis to Batesian and Müllerian mimicry (both typically involving selection by learning vertebrate predators; cf. Table 1) that reflects another possible scenario for the evolution of multifold and in particular very accurate resemblances to wasps: an innate, visual inhibition of aggression among look‐alike wasps, based on their social organization and high abundance. We argue that wasp species resembling each other need not only be Müllerian mutualists and that other insects resembling wasps need not only be Batesian mimics, but an innate ability of wasps to recognize each other during hunting is the driver in the evolution of a distinct kind of masquerade, in which model, mimic, and selecting agent belong to one or several species (Figure  1b). Wasp mimics resemble wasps not (only) to be mistaken by educated predators but rather, or in addition, to escape attack from their wasp models. Within a given ecosystem, there will be selection pressures leading to masquerade driven by wasps and/or to mimicry driven by other predators that have to learn to avoid them. Different pressures by guilds of these two types of selective agents could explain the widely differing fidelity with respect to the models in assemblages of yellow jackets and yellow jacket look‐alikes.     

Multitrait aposematic signal in Batesian mimicry

Batesian mimics can parasitize Müllerian mimicry rings mimicking the warning color signal. The evolutionary success of Batesian mimics can increase adding complexity to the signal by behavioral and locomotor mimicry. We investigated three fundamental morphological and locomotor traits in a Neotropical mimicry ring based on Ithomiini butterflies and parasitized by Polythoridae damselflies: wing color, wing shape, and flight style. The study species have wings with a subapical white patch, considered the aposematic signal, and a more apical black patch. The main predators are VS‐birds, visually more sensitive to violet than to ultraviolet wavelengths (UVS‐birds). The white patches, compared to the black patches, were closer in the bird color space, with higher overlap for VS‐birds than for UVS‐birds. Using a discriminability index for bird vision, the white patches were more similar between the mimics and the model than the black patches. The wing shape of the mimics was closer to the model in the morphospace, compared to other outgroup damselflies. The wing‐beat frequency was similar among mimics and the model, and different from another outgroup damselfly. Multitrait aposematic signals involving morphology and locomotion may favor the evolution of mimicry rings and the success of Batesian mimics by improving signal effectiveness toward predators.     

NEGATIVE FREQUENCY-DEPENDENT SELECTION BY POLLINATORS ON ARTIFICIAL FLOWERS WITHOUT REWARDS

Many species of nonmodel deceptively pollinated orchids are polymorphic for corolla color. These species are pollinated by naive insects searching for nectar, and are not mimics. It has been suggested that the foraging behavior of insect pollinators during the avoidance learning process results in these stable corolla color polymorphisms; for this to occur pollinators must induce negative frequency-dependent selection on corolla color. Therefore the hypothesis that pollinator behavior results in a preference for rare color morphs of deceptive species was tested experimentally. Bumblebees (Bombus terrestris) foraged in the laboratory on arrays of artificial flowers with different corolla color morphs. Morphs were varied in frequency, and bumblebee preferences were recorded on arrays where morphs did and did not contain sucrose solution rewards. Bumblebees preferred the most common color morph when flowers contained sucrose solution rewards, but overvisited rare morphs when sampling flowers that contained no rewards. Bumblebees also tended to move between unlike color morphs when these were unrewarding, suggesting that a probabilistic sampling strategy was adopted. Thus experiments demonstrated that pollinator behavior could result in a selective advantage for rare color morphs of plant species that are pollinated by deception without mimicry, which would induce negative frequency-dependent selection on corolla color. The observed pollinator behavior could allow stable corolla color polymorphisms to be maintained by selection in nonmodel deceptively pollinated species.     

Flower Color Polymorphism in Rhododendron cyanocarpum (Ericaceae), an Endangered Alpine Species Endemic to NW Yunnan,China

Rhododendron cyanocarpum is a narrow endemic species with pink and white floral color. In the present study, to investigate the significance of petal color morphs, we examined color morph frequencies, petal color reflectance and other associated floral characters, effective pollinators, visitation frequencies, and fruit production in the field. In all surveyed known populations, plants with pink color morph dominanted and comprised 77%-100% of individuals. Two peaks at 430 nm and 650 nm were found in the petal color reflectance of pink flowers, and only one peak at 430 nm was found in the reflectance spectrum of white flowers. In addition, color morphs were also associated with colors of style and stigma, lengths of corolla, calyx and pedicel, the closest distance between stigma and stamen, but not with style length and nectar production. Moreover, higher visitation frequency of their shared pollinators (bumblebees) and fruit production were observed of pink flowers than white flowers. Despite a briefly temporal and spacial study, we suggest that color morph frequencies, visit frequencies of bumblebees and fruit production, all favor to be stablilizing selection for the pink color morph.     

滇西北特有植物蓝果杜鹃的花色多态性研究(英文)

蓝果杜鹃(Rhododendron cyanocarpum)为大理苍山特有的濒危植物,有粉色和白色两种花冠类型。为了探讨该物种花色多态性的意义,本研究调查了粉色花和白色花植株在已知的各居群的分布频率、花冠的反射光谱及其它的花部特征、有效传粉者及其访花频率与结实情况。结果表明:粉色花植株在所有调查的居群中占优势(77%~100%)。粉色花的花冠反射光谱在430 nm和650 nm有两个峰,而白色花只在430 nm有一个反射峰。同时,花特征如:花柱与柱头颜色、花冠长度、花萼长度、花梗长度以及雌雄蕊最短距离,两种花冠存在显著差异。另外,尽管熊蜂作为这两种花冠的主要传粉者,但粉红花的访花频率以及自然条件下的结实情况显著高于白色花。本研究结果推测粉红色花可能受到了稳定性选择的作用。     

Nymphal colour/pattern polymorphism in the leafhoppers Eupteryx urticae (F.) and E. cyclops Matsumura (Hemiptera: Auchenorrhyncha): spatial and temporal variation in morph frequencies

Variation in colour/pattern morph frequencies in Eupteryx urticae and E. cyclops is described for various field populations. Eupteryx urticae populations in S Wales exhibit a steep morph-ratio cline, such that black morph frequencies are positively correlated with altitude. High melanic frequencies at high-altitude sites, and the absence of the two darker morphs in lowland populations, suggest a similar trend in E. cyclops , but the data are insufficient to confirm this statistically. No differences in morph frequencies were detected on different parts of the primary host plant or on alternative host species. Similarly, there were no consistent trends within or between the two annual generations of either species, although melanic morph frequencies in one E. urticae population were heterogeneous over 10 generations. It is suggested that the polymorphism in E. urticae is strongly influenced by climate selection, darker morphs being at an advantage in cooler environments where their coloration enhances absorption of solar radiation. The advantage gained through thermal melanism is probably balanced by visual selection against black morphs by entomophagous parasitoids.     

Female Colour Polymorphism: Gender and the Eye of the Beholder in Damselflies

Damselflies provide a classic example of female colour polymorphism. Usually, one female morph resembles the blue male colour (andromorph) while one, or more, female morphs are seen as typically female (gynomorph). Damselfly species fall in two distinct groups with respect to recent developments in mimicry theory: in some species females are perfect, they match male colouration and black patterning, and in other species they are supposed to be imperfect mimics, only matching male colouration. However, the underlying assumption of one female morph looking male-like is mostly based on human vision. Therefore we investigated the black patterning and colour of the three female morphs in Coenagrion puella, an imperfect mimic, using image analysis. In C. puella the blue female morph is perceived as male-like. We found that the black patterning of such females cannot be distinguished from the other female morphs, and is clearly different from males. Furthermore, the blue colour of andromorph females differs from the blue colour of males. Intriguingly, however, the red content did not differ between blue males and females.     

Dietary conservatism may facilitate the initial evolution of aposematism

It has generally been assumed that warningly coloured organisms pay a cost associated with their increased visibility, because naïve predators notice and eat them. This cost is offset by their enhanced protection from educated predators who associate the colour pattern with unprofitability. However, some studies have suggested that avoidance of novel prey by avian predators ("dietary conservatism") can actually place novel colour morphs at a selective advantage over familiar ones, even when they are highly conspicuous. To test this idea, we experimentally simulated the appearance of a single novel-coloured mutant in small populations (20 individuals) of palatable artificial prey. The colour morph frequencies in each "generation" were determined by the relative survival of the previous generation under predation by birds. We used wild-caught European robins Erithacus rubecula foraging on pastry "prey" of different colours. The aim was to test whether prey selection by predators prevented or facilitated the novel colour morph persisting in the prey population over successive generations. We found that the novel colour morph quickly increased to fixation in 14/40 prey "populations", and at least once each in 8 of the 10 birds tested. Novel mutants of the classic aposematic colours (red and yellow) reached fixation most frequently, but even the green and blue novel morphs both increased to fixation in 2/40 trials. Novel colours reached fixation significantly faster than could be accounted for by drift, indicating active avoidance by the birds. These results suggest that a novel colour morph arising in a prey population can persist and increase under the selective pressure imposed by predators, even to the local exclusion of the original morph, despite being fully palatable. The consequences of this finding are discussed in relation to receiver psychology, the evolution of aposematism and the existence of polymorphism in Müllerian mimics.     

No evidence for differential survival or predation between sympatric color morphs of an aposematic poison frog

Because variation in warning signals slows down the predator education process, aposematic theory predicts that animal warning signals should be monomorphic. Yet, warning color polytypisms are not uncommon in aposematic species. In cases where warning signal variants are separated geographically, adaptation to local predators could explain this variation. However, this cannot explain the persistence of sympatric polymorphisms in aposematic taxa. The strawberry poison frog (Oophaga pumilio) exhibits both allopatric and sympatric warning color variation in and around the Bocas del Toro archipelago of Panama. One explanation that has been proposed for the rapid diversification of O. pumilio coloration in this archipelago is low predation; if island populations have few predators, stabilizing selection would be relaxed opening the door for diversification via selection or genetic drift. Using a combination of mark-recapture and clay model studies, we tested for differences in survival and predation among sympatric red and yellow color morphs of O. pumilio from Bastimentos Island. We found no evidence for differential survival or predation in this population, despite the fact that one morph (red) is more common and widely distributed than the other (yellow). Even in an area of the island where the yellow morph is not found, predator attack rates were similar among morphs. Visual modeling suggests that yellow and red morphs are distinguishable and conspicuous against a variety of backgrounds and by viewers with different visual systems. Our results suggest that general avoidance by predators of red and yellow, both of which are typical warning colors used throughout the animal kingdom, may be contributing to the apparent stability of this polymorphism.     

Body size influences differently the detectabilities of colour morphs of cryptic prey

Body size and coloration may contribute to variation in performance and fitness among individuals; for example, by influencing vulnerability to predators. Yet, the combined effect of size and colour pattern on susceptibility to visual predators has received little attention, particularly in camouflaged prey. In the colour polymorphic pygmy grasshopper Tetrix subulata (Linnaeus, 1758), females are larger than males, although there is a size overlap between sexes. In the present study, we investigated how body size and colour morph influenced detection of these grasshoppers, and whether differences in protective value among morphs change with size. We conducted a computer‐based experiment and compared how human ‘predators’ detected images of large, intermediate or small grasshoppers belonging to black, grey or striped colour morphs when embedded in photographs of natural grasshopper habitats. We found that time to detection increased with decreasing size, that differences in time to detection of the black, grey and striped morphs depended differently on body size, and that no single morph provided superior or inferior protection in all three size classes. By comparing morph frequencies in samples of male and female grasshoppers from natural populations, we also examined whether the joint effects of size and colour morph on detection could explain evolutionary dynamics in the wild. Morph frequency differences between sexes were largely in accordance with expectations from the results of the detection experiment. The results of the present study demonstrate that body size and colour morph can interactively influence detection of camouflaged prey. This may contribute to the morph frequency differences between male and female pygmy grasshoppers in the wild. Such interactive effects may also influence the dynamics of colour polymorphisms, and contribute to the evolution of ontogenetic colour change and sexual dichromatism. © 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 113 , 112–122.     

Intensity of male‐male competition predicts morph diversity in a color polymorphic lizard

Sexual selection is one of the main processes involved in the emergence and maintenance of heritable color polymorphisms in a variety of taxa. Here, we test whether the intensity of sexual selection, estimated from population sex ratio, predicts morph diversity in Podarcis muralis, a color polymorphic lizard with discrete white, yellow, orange, white‐orange, and yellow‐orange male and female phenotypes (i.e., morphs). In a sample of 116 Pyrenean populations and 5421 lizards, sex ratios (m/f) vary from 0.29 to 2.5, with the number of morphs for each sex ranging from 2 to 5. Male‐biased sex ratios are associated with increased morph diversity as measured with Shannon's diversity index. The main factor accounting for this relationship is male morph richness (i.e., the number of morphs). In contrast, female morph diversity is not related to sex ratio. These results suggest a relationship between the intensity of male intrasexual competition and male morph diversity. While other selective forces may interact with sexual selection in maintaining the color polymorphisms in P. muralis, this evidence suggests a complex evolutionary scenario possibly involving frequency‐dependent selection of alternative reproductive tactics and/or complex balancing selection.     

Morph‐specific selection on floral traits in a polymorphic plant

Correlations between phenotypic traits are common in many organisms, but the relative importance of nonadaptive mechanisms and selection for the evolution and maintenance of such correlations are poorly understood. In polymorphic species, morphs may evolve quantitative differences in additional characters as a result of morph‐specific selection. The perennial rosette herb Primula farinosa is polymorphic for scape length. The short‐scaped morph is less damaged by grazers and seed predators but is more strongly pollen limited than the long‐scaped morph. We examined whether morph‐specific differences in biotic interactions are associated with differences in selection on two other traits affecting floral display (number of flowers and petal size) and on one trait likely to affect pollination efficiency (corolla tube width) in three P. farinosa populations. Differences in selection between morphs were detected in one population. In this population, selection for more flowers and larger petals was stronger in the short‐scaped than in the long‐scaped morph, and although there was selection for narrower corolla tubes in the short‐scaped morph, no statistically significant selection on corolla tube width could be detected in the long‐scaped morph. In the study populations, the short‐scaped morph produced more and larger flowers and wider corolla tubes. Current morph‐specific selection was thus only partly consistent with trait differences between morphs. The results provide evidence of morph‐specific selection on traits associated with floral display and pollination efficiency, respectively.     

Change and stability in a steep morph‐frequency cline in the snail Cepaea nemoralis (L.) over 43 years

Populations of the polymorphic land snail Cepaea nemoralis (L.) from Deepdale, Derbyshire, UK, sampled in 1965–67, showed a pattern of area effects, with steep clines among groups of populations differing in shell colour and banding morph frequencies. In 2010, most of these populations were resampled. In particular, a continuous transect made in 1967 of 42 quadrats (18.34 × 18.34 m) across a steep cline in several morph frequencies was completely resampled. In the dale as a whole, yellow shells had increased in frequency. In the transect, the frequencies of banding morphs showed no significant changes, although colour morphs showed some changes. Pink shells had increased in frequency in a section in which scrub had developed, and brown shells had increased in frequency in the area in which they had originally been at the highest frequency. In each case, the selection coefficients were of the order of 4%. Yellow had increased elsewhere. Nevertheless, both in the dale as a whole and in the transect, the pattern of geographical change in morph frequencies had remained essentially the same. Estimates of migration based on previous studies of marked snails and on modelling of the effect of drift and migration suggest that, regardless of whether the cline is a product of differential selection or of the gradual merging of previously separate founding populations, it has been in existence for a long time, and that migration occurs over greater distances than estimated from direct observation on marked snails. Although we can demonstrate that selection has occurred, the origin and maintenance of the cline and others similar to it remain in doubt; the development and maintenance of polymorphism in this species may require consideration of several processes operating on different time scales. © 2012 The Linnean Society of London     

Color Polymorphism in Spiny Spiders (Gasteracantha fornicata): Testing the Adaptive Significance of a Geographically Clinal Lure

Many sit‐and‐wait predators use conspicuous displays of color to attract prey. These displays sometimes express discrete polymorphisms; however, the adaptive drivers of such variation are not well understood. Here, we explore a previously unknown color polymorphism in the orb‐web spider Gasteracantha fornicata. We discovered that in North Queensland, Australia, female spiders exhibit dorsal bands appearing (to the human observer) either white or yellow and characterized by sigmoidal spectral curves centered on approx. 447 and 496 nm, respectively. Based on sensory drive theory, we hypothesized that morphs may be alternatively favored by the switch in ambient viewing conditions engendered by sunny vs. cloudy skies. We addressed this hypothesis indirectly by studying morph frequencies across a approx. 200 km geographic gradient of solar exposure (a surrogate for cloudiness), and by investigating the phenotypic signature of catch success via a resource stress experiment and in wild spiders. Our data indicate substantial geographic variation in morph frequency, with white morphs dominating at more cloudy northern locations. Rather than a gradual cline, morph frequency inverted mid‐way along this range and was closely fit by a logistic relationship with latitude. Experimentally restricted access to larger prey caused spiders to lose more mass than size and to exhibit less bright dorsal markings. Wild‐sampled spiders from two localities of divergent morph frequency indicated no differences in residual mass, but intriguingly, the white morph was larger and heavier (than the yellow morph) where it was relatively rare. Our data hint at negative frequency dependence, but remain broadly consistent with a sensory drive explanation based on cloudiness, and we suggest these as worthy avenues for closer investigation.     

Biased predation could promote convergence yet maintain diversity within Müllerian mimicry rings of Oreina leaf beetles

Müllerian mimicry is a classic example of adaptation, yet Müller's original theory does not account for the diversity often observed in mimicry rings. Here, we aimed to assess how well classical Müllerian mimicry can account for the colour polymorphism found in chemically defended Oreina leaf beetles by using field data and laboratory assays of predator behaviour. We also evaluated the hypothesis that thermoregulation can explain diversity between Oreina mimicry rings. We found that frequencies of each colour morph were positively correlated among species, a critical prediction of Müllerian mimicry. Predators learned to associate colour with chemical defences. Learned avoidance of the green morph of one species protected green morphs of another species. Avoidance of blue morphs was completely generalized to green morphs, but surprisingly, avoidance of green morphs was less generalized to blue morphs. This asymmetrical generalization should favour green morphs: indeed, green morphs persist in blue communities, whereas blue morphs are entirely excluded from green communities. We did not find a correlation between elevation and coloration, rejecting thermoregulation as an explanation for diversity between mimicry rings. Biased predation could explain within‐community diversity in warning coloration, providing a solution to a long‐standing puzzle. We propose testable hypotheses for why asymmetric generalization occurs, and how predators maintain the predominance of blue morphs in a community, despite asymmetric generalization.