The Mechanism for Mimicry: Instant Biosemiotic Selection or Gradual Darwinian Fine-Tuning Selection?
- 187 Downloads
Biological mimicry is regarded by many as a textbook illustration of Darwin’s idea of evolution by random mutation followed by differential selection of reproductively fit specimens, resulting in gradual phenotypic change in a population. In this paper, I argue that some cases of so-called mimicry are probably merely look-a-likes and do not gain an advantage due to their similarity in appearance to something else. In cases where a similar appearance does provide a benefit, I argue that it is possible that these forms of mimicry were created in a single generation. An interpretive response to an appearance as a sign can make a new structure perform drastically differently in an environment. In such cases, Darwin’s natural selection mechanism only helps to explain gradual the spread of these new forms, not the creation of them. I argue that biosemiosis should be regarded as a much more powerful mechanism for affecting evolutionary trajectories than the gradualist view allows. I focus on two cases of butterfly mimicry: the Viceroy (Nymphalidae: Limenitis archippus) and Monarch (Nymphalidae: Danaus plexippus) butterflies, supposed Müllerian mimics, and deadleaf mimic butterflies (Kallima).
KeywordsSaltationism Turing patterns Mimicry Pattern formation Genus Kallima Mimicry skepticism H. F. Nijhout
As a recent study by Ioannidis (2005) observes, academic journals tend to publish confirmations of theories, and when a researcher conducts an experiment and fails to confirm a theory, the study is less likely to be published. Even though there was an overwhelming number of published papers confirming a particular mimicry theory, this may not give a realistic picture of the actual outcomes of experiments that were tried. My interest in the question of the Monarch-Viceroy relation is inspired by the world’s most famous lepidopterist, novelist Vladimir Nabokov, who tasted both the Monarch and Viceroy himself in the 1940s and found they were both bitter (Boyd 2000, 535). He hypothesized, therefore, that the Viceroy might not be a Batesian mimic of the Monarch. Despite his more direct empirical approach, Nabokov’s controversial view of the status of this supposed mimic was condemned by his peers and he lost interest in trying to publish.
After Ritland and Brower’s findings were disemminated, instead of accepting that similarly bad tasting Viceroys and Monarch may not have a mimicry relation at all and may merely look alike, researchers changed the mimicry status of the Viceroy from Batesian mimic (a species looks like another that has an advantage against predation) to Müllerian mimic (two species have an advantage against predation and, being similar in appearance, share the cost of educating would-be predators). The Müllerian theory asserts that the Monarch and Viceroy share a conspicuous warning pattern because natural selection has shaped and retains both patterns in the populations.1
If we find that indeed the similarity does serves the function of signaling bad taste, the mimicry researcher must still examine other assumptions before attributing similarity to the patient process of natural selection. For example, firstly, it is commonly assumed that the near-identical appearance of two different species is too unlikely a phenomenon to occur without the pressure of selection. This may not be the case. How do we determine which patterns are more likely to occur in the absence of selection? A second assumption of the gradualist view is that there are transitional forms were significant enough to be fine-tuned by selection. How do we test such an assumption? A third assumption is that the similar species are not hybrids or the result of lateral gene transfer. Has genetic sequencing been conducted? A fourth assumption is that the mimic species’ reproductive strategy, which, if profligate rather than parsimonious, does not make Darwinian selection ineffective. A fifth assumption is that selection pressure is as active on the similarity as on some other reproductive advantages. All assumptions must be valid to ascribe the similarity to a gradual process of natural selection.
I argue here that mimicry does not lend itself to a gradual selection process. A resemblance has to be good enough to fool predators or prey, and if the first instance were but a slight resemblance, it would have not offered enough of an advantage for selection to work on it. Darwinists have long conceded that natural selection could only perform fine-tuning for mimicry once a good look-a-like form has appeared by chance (Poulton 1913; for a biosemiotic discussion, see Maran 2015). But even fine-tuning may be a task ill-suited for natural selection, because the process cannot make forms better if the improvement does not confer a significant additional reproductive advantage. For example, as Nabokov pointed out (Boyd 2000, 86), natural selection could not paint faux fungus spots on a butterfly wing that already looks like a dead leaf—because fungus spots would add a degree of realism unnecessary to better fool predators.
Thus, before considering that a similarity between two animals might be created and retained in a population only by selection for reproductive fitness, we must first rule out the two most likely hypotheses for the pattern’s origin and subsequent persistence: first, physiological constraints on development (Cf. Kleisner 2010) and second, genetic mixing. Either process would tend to produce multiple individuals having the same feature, whereas the Darwinian mechanism, which depends on random mutation, requires many singular occurrences selected over time. I propose that in the majority of cases, a look-a-like appears in a single-generation by chance, like Richard Goldschmidt’s (1940) “hopeful monster,” due to constraints on development or hybridization. If the constraints are powerful enough or genetic mixing frequent enough, the new form might persist even if it confers no function.
Once the look-a-like form is in existence, it might be selected for function or it might not. In either case, the hopeful monster origin hypothesis has the advantage of being much more elegant than the gradualism hypothesis as it involves fewer steps, one significant coincidence instead of many relatively less significant coincidences.
The hopeful monster hypothesis also highlights the role of interpretation, or rather misinterpretation. Biosemiotics helps us understand how radically new mimics may be instantly created in nature via a predator’s misinterpretation of physical reality as a sign.2 It is not competition that spurs innovation, creates new functions, or that determines radically new evolutionary trajectories by gradually refining forms. It is the misinterpretation of things as signs that creates new functions, innovations that are not merely a little better than before, but radically different. This biosemiotic theory of mimicry is saltational, not gradual.
This paper focuses on two cases: 1.) the Viceroy (Nymphalidae: Limenitis archippus) and Monarch (Nymphalidae: Danaus plexippus), supposed Müllerian mimics that, I suggest, may not be examples of mimicry at all but simply a chance match, which is neutral with regard to fitness; and 2.) various examples of deadleaf butterflies (of the genus Kallima) that, I argue, may present a remarkable, saltational case of mimicry due to biosemiosis alone, not gradual selection.
Monarch and Viceroy: Hybridization
Monarch butterflies have somewhat unusual mating practices compared to other butterflies; they have at least partially abandoned the use of pheromones to attract members of their own species, and subsequently male Monarchs have a tendency to attempt, aggressively, to mate with other species (Rothschild 1978). Although butterflies in different genera are unlikely be able to produce viable hybrid offspring, the relatively higher frequency, compared to other butterflies, of wild encounters between aggressive and non-discriminating Monarchs and their cousins Limenitis (butterflies in the Viceroy’s genus) would increase the, however remote, possibility of viable hybrid male offspring (following Haldane’s rule for hybridization, insect females tend to be sterile) and subsequent assortative mating or backcrossing. Hybridization between individuals from different genera may be rare, but it does occur in nature occasionally.
The Viceroy’s Monarch-like appearance is unique among its own genus, members of which are more typically dark all over with a broad white stripe, and therefore the Viceroy’s appearance would be coming from the father’s side of the family, if the similarity is due to interbreeding. Interbreeding should be conclusively ruled out before accepting that the pressure of reproductive fitness has gradually fine-tuned the pattern of the Viceroy to make it look more like the Monarch while simultaneously stabilizing the aposematic Monarch pattern.
Genetic sequencing has been performed on Monarchs (Zhan et al. 2014) and on Viceroys (Mullen 2006). Steven Reppert, whose lab did the Monarch sequencing, reports (pers. com. June 27, 2017) that, to his knowledge, so far a genetic comparison between Monarch and Viceroy has not been made. Although interbreeding may not be behind the similar appearance in the case of this butterfly pair, it may be behind other instances of supposed mimicry in the animal kingdom and is almost certainly the cause of some similarities in the plant kingdom where hybridization is much more common.
If interbreeding can be ruled out as a cause of similar appearance, we still might consider the much more radical type of hybridization process, the lateral gene transfer of interspecies DNA through bacterial and viral infection, as a possible cause of similar forms with overlapping habitat. The discoveries in the last few decades of the extent to which interspecies gene swapping can occur via micro-organisms makes it more plausible that Monarchs and Viceroys may have come to share genes even without interbreeding. Butterfly family trees may have some branches that fuse together. Monarch genes contain parasitic wasp genes (Gasmi et al. 2015) and the same wasps might prey on Viceroys—or wasps carrying infections might act as vectors for swapping virus or butterfly genes. Such transfers of genetic material might result in rapid phenotypic change and produce a good number of look-a-likes whose similarity serves no reproductive function. Again genetic sequencing should be performed before hybridization is ruled out.
If the similar appearance is not the result of some such hybridization event, the shared pattern may be the more or less common result of the reaction-diffusion processes underlying the development of the wing pattern, processes that tend to converge upon the same pattern from different initial conditions. The physical constraints on pattern formation should be regarded as the most likely hypothesis for the tendency of similar patterns to arise. As Turner (1984) has argued, because butterflies share a common “toolbox” for forming patterns, one or few mutations could lead to a large change in appearance, bringing one species reasonably close to another.
Monarch and Viceroy: Determining the Probability of Similar Appearance without Natural Selection
Neo-Darwinists have assumed (Fisher 1930; Pfennig et al. 2001) there is an extremely low probability that one species might mutate into a match of another species without the help of natural selection. A method of determining how many possible distinct butterfly wing patterns exist within a butterfly family, and which of these may be more probable than others, would need to be established before one could objectively determine how hard selection would have to work, if at all, to find a match.
Because there is so much diversity in butterfly families, there are many possibilities for similarities. Indeed we do find that the Monarch and the Viceroy are not the only striking look-a-likes between members of their respective butterfly sub-families. Danaus eresimus montezuma looks like Limenitis archippus obsoleta. The Cuban King, Danaini Anetia cubana, from a genus related to the Monarch, looks like Limenitis weidemeyerii weidemeyerii. Among Monarchs and Viceroys there are variations of dark orange to pale yellow, with differing fore and hindwings. A variation of Danaus plexippus plexippus looks like Limenitis archippus lahontani. These three apparently radically different pattern types may actually be closely related in terms of the types of underlying diffusion processes that help determine them.
Butterfly Wing Pattern Formation Model: Self-Organization by Autocatalysis and Lateral Inhibition
The Appropriateness of Reaction-Diffusion Models for Biology
The use of reaction-diffusion equations to model various oscillating patterns in animal morphology has been long established (Bard 1977; Bard 1981; Meinhardt 1982; Oster 1988; Murray 1989). Such models are accurate to the extent that the genes just produce materials (and in some cases the rate of production) and the laws of physics and chemistry determine global spatial pattern. In Nijhout’s model the development of the pattern begins in an undifferentiated field of cells. Some genes within some cells, triggered by an instability, begin to produce materials (morphogens) and these materials react and diffuse away from the cell creating spatial patterns. We may argue a priori that cells are not triggered to produce patterns by internal information about their position within a field. Cells could not have longitudinal and latitudinal information about their location vis-a-vis a global map; they can only have access to information about their local conditions, but this is sufficient for differential global patterns to emerge via sign action. We may say, local conditions indicate (are a sign of) global conditions that the individual cell interprets (responds to) and thereby helps create. It is true of all complex interacting systems that simple local rules determine complex emergent patterns. To reiterate, it is argued that the genes themselves do not specify the global patterns; the gene-produced materials simply self-organize spontaneously due to the properties of the materials and the physical constraints on the reaction.
Reaction-diffusion processes may be regarded a main mechanism for differentiation and regulation, providing the spatial and/or temporal differences that differentially trigger genes, determining different cell fates. Because butterfly wing pattern development initiates in a simple undifferentiated two-dimensional structure (the wing membrane), the diffusion patterns are no different from the types of diffusion patterns that might occur in a similarly configured field of chemicals in a petri dish or a similarly configured pixel field on a computer screen. If Nijhout’s model can only approximate the butterfly wing groundplan, where it fails to produce reasonably good representations of existing species may provide an indication of further genetic effect on the pattern that is as yet not understood by researchers.
While most pattern formation researchers attempt to reverse engineer the particular pattern in which they are interested, find the equation that creates their pattern, I would like to see a computer model of the groundplan run through hundreds of thousands of various parameters just to see what turns up. I believe it is important to mimicry research to try to discover which universal reaction-diffusion patterns are created the most often, and within the widest of parameter ranges, in the absence of natural selection.
Nijhout’s Model and Monarch-Viceroy Similarity
According to Nijhout’s model, a Monarch-Viceroy type dark vein pattern would be laid down in the first stage of the reaction, when the activator first increases along the veins and wing edge. Although Nijhout does not offer a justification for initiating activation along the veins, we can assume activation self-enhancement may start along the edges, without being specifically programed to do so, because these areas do not get the rapidly diffussing inhibitor from all neighboring regions as other areas do. This instability could be the trigger that initiates the process of differentiation, as per Waddington (1940), Turing (1952)3 and Meinhardt and Gierer (2000). We need not look for an inducer, a specific chemical agent, that turns the process of differentiation on. If the trigger is a direct result of the physical limits of the wing cell field, then the trigger should always be identical in all butterflies whose wing veins similarly constrain chemical communication among cells. If this is the case, the dark vein pattern and dark submarginal band would be a kind of default pattern if the activation diffusion period is brief and melanin is later produced. According to Nijhout’s model if there are strong venous stripes formed by melanin, that is, the dark pigment activator does not diffuse very far away from the wing veins, no additional pattern elements will be expressed in melanin. The Viceroy has a hint of a central symmetry system showing up in the conspicuous line across its hindwing: more melanin has diffused away from the veins in this species than it has in the Monarch. In the submarginal band, which has edges on three sides constraining clashing waves of inhibitor followed by activator, we observe the evidence of a chemical threshold switching on and off, producing black bands and white dots, traces of activation and inhibition.
Two pigments types are likely involved in the basic Monarch and Viceroy wing patterns: melanins for black and dark brown to reddish brown and ommochromes for red, orange, and brown (Kayser 1985). Ommochromes rendering the orange color activate first, followed by melanin (ffrench-Constant and Koch 2003). I have not been able to determine whether or not researchers believe the white dots on the submarginal bands are due to another type of pigment or to the absence of pigmentation. Vane-Wright has supposed that the White Monarch variant is due to a recessive allele that renders the insect unable to produce the orange pigment (Vane-Wright 1986). In any case, Nijhout’s model is mainly concerned with the shape of the pattern irrespective of color. Because the shape of the wing pattern is determined by the activator/inhibitor diffusion process prior to pigmentation, the type of pigments involved (e.g. of differing molecular sizes) may have little effect on the pattern.
Because the Monarch model has only three pattern elements expressed—dark wing veins, dark submarginal bands and solid orange background—the probably of matching these characters would, of course, be higher than the probably of matching five expressed pattern elements. Given the constraints underlying the production of this simple pattern, the similarity between the Viceroy and the Monarch does not seem unlikely enough to require an additional causal explanation. Not everything that exists in nature has been selected for its reproductive fitness. Some non-utilitarian forms are just likely to appear fairly frequently and persist.
The Deadleaf Butterfly as Mimic
Deadleaf butterflies are categorized as “plant-part” mimics, resembling a specific plant in the environment. The insect is not camouflaged; it is supposed that predators do not merely overlook it; they see it as a leaf. In biosemiotic terms, its wings function, for the insect, as a sign of a leaf to the would-be predator. There are about ten different species within the genus Kallima with underside wing patterns resembling dead leaves and upperside wing patterns with bright and bold colors. Kallimas are found in the tropical forests of East and Southeast Asia, and when they rest in trees or are in flight, they are conspicuous amid the bright green foliage. They slightly resemble Asian evergreen oak leaves, such as Quercus lanata, hence they are also known as “oakleaf” butterflies. When deadleaf butterflies spend time on the forest floor among dried oak leaves and other brown debris, consuming decayed fruits and dung, as they are wont to do (Igarashi and Fukuda 2000), they are well disguised—but only if they keep their wings closed, if not, they are quite conspicuous. Of nine amateur videos found available online showing various species of Kallima eating fruit, six show the insects beating their wings, calling attention to themselves while occupied with dinner.4 The three videos that show butterflies with wings sensibly closed were less than a minute in length.
Alfred Russel Wallace famously claims (1870) that he observed deadleaf butterflies concealing themselves by resting on the forest floor or on dead twigs, but naturalist Bashford Dean, mimicry skeptic, contradicts (1902) Wallace’s observations, noting that, in his opinion, the butterflies did not put their excellent disguise to very good use at all, tending to rest on broad green leaves. But Wallace’s account anticipates such skepticism, claiming that when deadleaf butterflies are being chased by birds, they dive to the ground and disappear among the debris by closing their wings. But this scenario requires an active use of the leaf disguise—the butterfly sensing pursuit and taking action to avoid it. The most effective type of disguise would be one that is passively functional at all times. Wallace also notes that the underside pattern varies a great deal from specimen to specimen. This, of course, suggests that the pattern has not been fixed by selection. Nevertheless, to Wallace, this is the “more extraordinary part of the imitation” that natural selection has created in a single species, “representations of leaves in every stage of decay, variously blotched and mildewed and pierced with holes.” I wonder if Wallace and other mimicry researchers do not try a little too hard to imagine how the fantastically disguised deadleaf may have evolved in mimicry.
Probably an indistinct brown color on both sides of the wings would have been excellent camouflage for an insect on the forest floor among the dung and debris. We may consider that the insect’s incredibly accurate elm leaf disguise is a bit superfluous.
The Likelihood of the Deadleaf Groundplan Form Appearing in the Absence of Natural Selection
The first deadleaf forms could have been produced by trauma disrupting development. The insect is sensitive to environmental factors during development and has a wet season and a dry season form. Temperature-shock studies reveal that insects exposed to extreme conditions during development tend to be duller in color and have smaller eyespots than their parents (Nijhout 1991, 122). If trauma foreshortened the wings such that they became more symmetrical and the wing apices pinched (like a leaf), the distorted frame of the reaction-diffusion process may have been sufficient to increase the probability of displacing the hindwing central symmetry system in a single generation. We can observe this happening, in fact, in variations in a related species. In the wet-season form of Precis almana, both wings are in the shape of the typical Nymphalid groundplan. In the dry-season form of Precis almana, the hindwing is foreshortened, an apex forms that looks like a leaf stem and the leaf midrib-like line appears.
Although temperature-shocked specimens still retain the genotype that produces the parents’ wing shape and pattern, through a process of assortative mating, the new phenotype may become fixed in the genome. As some butterflies are known to segregate by color (Chamberlain et al. 2009) aberration-prone individuals may reproductively isolate themselves. If temperature shocked individuals tend to mate with each other, the possibility for genetic assimilation to occur greatly increases (see Alexander 2001; Kull 2014, 2016). If so, natural selection would not be necessary even to explain the proliferation of this remarkable insect; reproductive isolation alone may be enough to account for the aberrations not being backcrossed into the population and their novel form outnumbered in the reproduction competition.
If the leaf appearance was created in a single generation by the deforming action of temperature shock, and would-be predators actually mistook it for a leaf and did not just overlook it, then the deadleaf butterfly would exemplify saltational biosemiotic selection for mimicry. Temperature-shocked hopeful monsters may have functioned instantly as icons of fallen dead leaves to predators. If so, Darwinian gradual selection could help only explain this insect’s gradual proliferation, it would not be useful to explain the creation of its form.
If the researcher has confirmed that a plant or animal, such as the deadleaf butterfly, has been selected for its similarity to something else in nature, then biosemiotics can add valuable insights considering the perceptual and interpretative abilities of predators, prey, or mates that may be foolded by the disguise. That type of analysis, however, it out of the scope of this paper, which has focused on the process of distinguishing between mere look-a-likes and true mimics.
A Profligate Reproductive Strategy Does Not Encourage Natural Selection
Why are butterfly wing patterns so various? Is it because natural selection has created so many differentially fit patterns? Or is this great variety an indication of relaxed or non-existent fitness selection? Most insects use a lottery strategy of reproduction, having many more offspring than will survive to reproduce: this means there will be a lucky winner that may have no better fitness than the rest. Butterflies lay up to 500 eggs; profligate reproduction ensures that at least a couple of individuals will survive by pure chance to reproduce. Insofar as butterfly populations remain stable, we can guess that not very many of the 500 eggs laid survived long enough to breed. We cannot say that natural selection can be very effective on such a population. When selection pressure is thus absent, forms will be constrained only by existing physiological constraints. The fullest range of morphological space would be most clearly visible in organisms that are not subject to severe selection. This is why Theodor Eimer (1897) used butterfly wing patterns to develop his theory of orthogenesis to define the underlying tendencies of evolution in the absence of fitness selection.
Selection Pressure Should Be Signifiant on the Wing Pattern Similarity for it to Be Considered Mimicry
The most significant selection pressure on butterflies occurs when the insect is in the caterpillar and chrysalis stages. Many eggs do not hatch, many caterpillars are eaten, many chrysalis are damaged by fungus or extreme weather, and most do not survive to the butterfly stage. The key function of the butterfly stage is mating, thus selection pressure at this stage is more likely directed at mating. Vladimir Nabokov, who passionately argued against gradual selection as the cause of mimicry (Alexander 2001, 2003, 2016), found the most reliable way to distinguish between butterfly species is by looking at reproductive organs, not the wings. Using this technique, he correctly understood the complicated migration history of and familial relationships between a group of Lycaenid butterflies with the common name “Blues” (Vila et al. 2011).
If butterfly mimicry exists, we can expect it to appear more often in the early stages. Indeed we find that caterpillars of Blues are parasitic on ants and secrete mimic ant pheromones to attract them. This mimic secretion serves a well-documented purpose, and thus this appears to be a case of true mimicry with a clearly defined model (See short summary in Maran 2017, 132–133). A biosemiotic approach to this mimicry case would focus on the instant appearance and misinterpretation by the ants of the novel pheromone. It would not depend upon gradual change of the pheromone to better attract the ants.
The Consequences of Saltational Mimicry Theory on Other Issues
In some important sense, all of biosemiosis involves mimicry, the interpretation of (use of) similarities as meaningful/functional. Even a cell’s responses to enzymes and chemical/hormonal signals are based on the likeness of the shape of the sign vis-a-vis the receptor. When a random mutation results in a new biological structure or form, it must function as a sign to an already existing sign-reading system; most likely this occurs when the new form mimics (with an important difference) a previously existing one. Thus, mimicry may be the sine qua non of biosemiosis at all levels, within cells and between organisms.
I conclude by reflecting upon the larger social implications of this saltational mimicry hypothesis. Darwin’s theory has lent support to the idea that competition is the driver of innovation, a de facto creator, incrementally shaping new forms and providing direction to evolutionary trajectories. In general, however, competition and selection actually tend to reduce variety and therefore stifle innovation. When selection pressure is not active, mutations can freely explore new structures that are irrelevant to survival and reproductive success. This may be why we see so many different species of butterflies with so many different kinds of patterns: natural selection is not constraining them.
An increase in diversity gives chance a larger pool from which to choose and leads to the increase in the possibility for novel functionality to be discovered through a misinterpretation of iconicity. I argue that a combination of the luck of a hopeful monster and biosemiotic interpretative process probably drives the most significant number of innovations in evolution. If true, this should shape our approaches to all those disciplines which are misinformed by the competition theory of innovation, such as economics, education and other areas of political life.
Such mimicry comes in various forms, Müllerian Convergence, Batesian-Müllerian spectrum, quasi-Batesian mimicry, arithmetic mimicry or auto mimicry, and may involve an edibility spectrum from palatable to highly noxious, and palatability may be affected by food sources. See Maran (2015).
The misinterpretation of a mimic as a sign may also be made by potential mate, as in the case of wasp-mimicking orchids, or by potential prey, as in the case of the live bait-mimicking angler fish.
Whether or not Turing patterns and the equations used to describe them were equivalent to the equations that could describe biological patterns was a subject of debate and doubt for many years. Sheth et al. (2012) has finally provided evidence of a hypothetical Turing system involving Hox genes, and Raspopovic et al. (2014) confirms Sheth et al. (2012) revealing which signaling molecules act as the Turing system.
https://www.youtube.com/watch?v=AdpXvAc7GTI30 seconds (same butterfly as above)
https://www.youtube.com/watch?v=VCGnaas5XdM&t=189s4 min 39 seconds
- Alexander, V. (2001). Neutral evolution and aesthetics: Vladimir Nabokov and insect mimicry. Working Papers Series 01-10-057. Santa Fe: Santa Fe Institute.Google Scholar
- Alexander, V. (2016). Chance, nature’s practical jokes, and the “non-utilitarian delights” of butterfly mimicry. In S. Blackwell & K. Johnson (Eds.), Fine lines: Vladimir Nabokov’s scientific art (pp. 225–234). New Haven: Yale University Press.Google Scholar
- Bard, J. (1977). A unity underlying the different zebra striping patterns. Journal of Zoology, 183(4), 527–539.Google Scholar
- Boyd, B. (2000). Nabokov’s butterflies: Unpublished and uncollected writings. New York: Beacon Press.Google Scholar
- Eimer, T. (1897). Orthogenesis der Schmetterlinge [Orthogenesis of Butterflies: A Proof of Specifically Directed Development and Impotence of Natural Selection in Species Formation; At the same time a reply to August Weismann]. Leipzig: Englemann.Google Scholar
- Ffrench-Constant, R., & Koch, P. (2003). Mimicry and melanism in swallowtail butterflies: Towards a molecular understanding. In C. Boggs, W. Watt, & P. Ehrlich (Eds.), Ecology and Evolution Taking Flight: Butterflies as Model Systems (pp. 259–279). Chicago: University of Chicago Press.Google Scholar
- Fisher, R. (1925). Statistical methods for research workers. Edinburgh: Oliver and Boyd.Google Scholar
- Gasmi, L., Boulain, H., Gauthier, J., Hua-Van, A., Musset, K., Jakubowska, A. K., Aury, J.-M., Volkoff, A.-N., Huguet, E., Herrero, S., & Drezen, J.-M. (2015). Recurrent domestication by lepidoptera of genes from their parasites mediated by bracoviruses. PLoS Genetics, 11(9), e1005470.CrossRefPubMedPubMedCentralGoogle Scholar
- Goldschmidt, R. (1940). The material basis of evolution. New Haven: Yale University Press.Google Scholar
- Igarashi, S., & Fukuda, H. (2000). The Life Histories of Asian Butterflies 2. Tokyo: Tokai University Press.Google Scholar
- Jordaan, F. (2009). Photo under creative commons. Available at https://www.flickr.com/photos/fjordaan/4209283247/in/set-72157622989071863/. Accessed 18 February 2019
- Kayser, H. (1985). Pigments. In G. A. Kerkut & L. I. Gilbert (Eds.), Comprehensive insect physiology, biochemistry, and pharmacology (Vol. 10, pp. 367–415). New York: Pergamon Press.Google Scholar
- Meinhardt, H. (1982). Models of biological pattern formation. London: Academic Press.Google Scholar
- Meinhardt, M., & Gierer, A. (1974). Application of a theory of biological pattern formation based on lateral inhibition. Journal of Cell Science, 15(2), 321–346.Google Scholar
- Nijhout, H. F. (1991). Development and Evolution of Butterfly Wing Patterns. Washington D.C.: Smithsonian.Google Scholar
- Poulton, E. (1913). Mimicry, mutation and Mendelism. Bedrock: A Quarterly Review of Scientific Thought, 2(3), 42–56.Google Scholar
- Rothschild, M. (1978). Hell’s Angels. Antenna, 2(2), 38–39.Google Scholar
- Schwanwitsch, B. N. (1924). On the groundplan of the wing-pattern in nymphalids and certain other families of rhopalocerous Lepidopetra. Proceedings of the Zoological Society of London B, 34, 509–528.Google Scholar
- Süffert, F. (1927). Zur vergleichende Analyse der Schmetterlingszeichnung. Biologisches Zentralblatt, 47, 385–413.Google Scholar
- Turner, J. R. G. (1984). Mimicry: The palatability spectrum and its consequences. In R. I. Vane-Wright & P. R. Ackery (Eds.), Biology of Butterflies (pp. 141–161). Londo: Academic Press.Google Scholar
- Vane-Wright, R. I. (1986). White monarchs. Antenna., 10(3), 117–118.Google Scholar
- Vila, R., Bell, C., Macniven, R., Goldman-Huertas, B., Ree, R., Marshall, C., Bálint, Z., Johnson, K., Benyamini, D., & Pierce, N. (2011). Phylogeny and palaeoecology of Polyommatus blue butterflies show Beringia was a climate-regulated gateway to the New World. Proceedings of the Royal Society B, 278(1719), 1–8.CrossRefGoogle Scholar
- Waddington, C. H. (1940). Organisers and Genes. Cambridge: Cambridge University Press.Google Scholar
- Wallace, A. (1870). Mimicry, and other protective resemblances among animals. Contributions to the Theory of Natural Selection: A series of essays. London: Macmillan & Co. 45–129.Google Scholar
- Wickler, W. (1968). Mimicry in plants and animals. New York: McGraw-Hill.Google Scholar