Which of the following best explains the cause of phenotypic variation observed in butterflies?

Genome-Wide Evidence for Speciation with Gene Flow in Heliconius Butterflies

The closely related neotropical butterfly species Heliconius melpomene, Heliconius cydno, and Heliconius timareta are distasteful to predators and often exhibit Müllerian mimicry to more distantly related species (Martin et al., 2013). All three species include various wing pattern forms, or races, which can be viewed as incipient stages of speciation, as it seems that selection for Müllerian mimicry can lead to wing pattern divergence and color-based assortative mating with no geographic separation (Chamberlain et al., 2009). Heliconius cydno and H. timareta together form a clade that is sister to H. melpomene, but unlike H. cydno, several H. timareta races have H. melpomene-like patterns (Figure 2). Martin et al. (2013) provided evidence that 20–40% of the genome in H. melpomene shows a discordant phylogenetic pattern, consistent with admixture with H. cydno or H. timareta in sympatry. Estimates of admixture by means of the ABBA/BABA method increased with time period examined, implying continued gene flow during speciation as opposed to a recent burst. Additionally, linkage disequilibrium (LD) was strongest between derived alleles that were shared during the recent time period, indicating the existence of introgressed haplotype blocks that are yet to be broken down fully by recombination. Remarkably, blocks of shared sequence variation containing the B/D region, known to be responsible for red mimicry patterns between races of H. melpomene, were exchanged between postman H. timareta and postman H. melpomene (Heliconius Genome, 2012). Interspecific FST between sympatric species tended to be lower and more variable than between the corresponding allopatric populations, as expected under a model of admixture with variable selection against introgressing alleles. There was a significantly reduced signature of admixture on the Z chromosome compared with autosomes, which can be associated with Z-autosome incompatibilities known to cause female hybrid sterility in the system (Jiggins et al., 2001; Naisbit et al., 2003). Genomics thus provides empirical data for addressing the ongoing debate between recent proponents of sympatric speciation and the classical view of ubiquitous allopatric speciation.

MUTUALISM AND COEVOLUTION IN MÜLLERIAN MIMICRY

In contrast with the unilateral Batesian evolution in which mimics outrun their models, Müllerian mimicry was traditionally thought to involve mutual resemblance of the species involved, as if all had moved toward some halfway phenotype. Of course, Müller himself and others were quick to point out that the mutual benefits were not even, but lopsided, that is, typically the rarer or the less distasteful species would benefit more than the more common or better defended one (respectively). However mutualistic the relation is, coevolution has often been assumed in Müllerian associations, and the protagonists are usually called comimics just because it is difficult to know if one species is driving the association. Coevolution also predicts that geographic divergence and pattern changes should be parallel in both species of comimics, like in the mimetic pair H. erato and H. melpomene in tropical America, presumably leading to parallel phylogenies. However, DNA sequences from mitochondrial and nuclear genes show distinct phylogenetic topologies in these two species and distinctly nonparallel evolution.

In fact, there are a number of grounds on which to believe that the asymmetrical relationship leads to one-sided signal evolution even in Müllerian mimicry, one species being a mimic and the other a model. First, because of number dependence, mimetic change of a rarer species toward a commoner species will be retained, but the reverse is not true: by mimicry of a less common species, the commoner species would lose the protection of its own ancestral pattern, and a change toward a rarer pattern would be initially disadvantageous. The commoner species is therefore effectively locked in its pattern, and initial changes are only likely in the rarer species. Second, given the selection against nonmimetic intermediates, the mutants in the rarer species will have to be roughly mimetic of their new model to be selected, thus bringing the ultimate shared signal closer to that of the common species. Once this initial step is made by the mimic, there could be gradual “coevolution” to refine the resemblance, but the resulting change in color pattern will inevitably be more pronounced in the mimic, the model remaining more or less unchanged. Because Müllerian pairs are of a mimic–model nature, even with mutual benefits, the prediction for parallel evolution is therefore not likely to be valid. Indeed, in the mimetic pair H. erato/H. melpomene, the phylogeography suggests that H. melpomene has radiated onto preexisting H. erato color-pattern races, thus colonizing all color-pattern niches protected by H. erato in South America.

Evidence of selection

Several investigators have shown that reproductive isolation has evolved as a by-product of adaptive divergence, a process called ‘ecological speciation’. For example, the evolution of mimicry appears to have played an important role in speciation in the butterfly genus Heliconius. The recently split sister species H. melpomene and H. cydno have diverged to mimic the color patterns of different model taxa (Figure 3a), but strong assortative mating (preferential mating among individuals with similar phenotypes) based on the mimetic coloration results in substantial pre-zygotic isolation (Figure 3b). Rare hybridization events produce individuals with poorly adapted intermediate phenotypes, demonstrating that divergent patterns of mimicry also result in some post-zygotic isolation. Among plants, species in the genus Mimulus have evolved highly divergent floral morphologies that appear to be adaptations to different types of pollinators (either bees or hummingbirds) that in turn effectively reproductively isolate sympatric populations. Using controlled laboratory experiments, several studies have also generated strong assortative mating among replicate lines of houseflies and species of Drosophila by subjecting them to artificial divergent selection on behavioral, morphological, and physiological traits. Control populations show little behavioral isolation, indicating that divergent selection, rather than genetic drift, caused the incidental evolution of pre-zygotic isolation.

Sexual selection is also probably important in allopatric speciation given that divergence in sexually selected traits will necessarily diminish interbreeding. Many experiments have shown that among species where a female ‘chooses’ males to mate with, females make their choices based on sexually dimorphic traits in males, such as large body size, bright coloration, large antlers, and elongated tail feathers. Sexual selection is probably involved in the evolution of gametic incompatibilities and isolation in some marine organisms that release their gametes into the water column in mass spawnings, a scenario that results in intense sperm competition among males. Analysis of the underlying DNA sequences for some sperm surface proteins indicates that the evolution of these proteins is driven by selection. Rapid development of gametic incompatibility may explain how some populations evolve reproductive isolation during only brief periods of transient geographic isolation.

Mimicry in Heliconius Butterflies

Few organisms exist that attract the attention, admiration, and intrigue of scientists, naturalists, and laymen alike as do butterflies. They are conspicuous in nearly every biome. Undoubtedly, this contributed to the attention they received by Bates, Müller, and a host of other scientists. Yet their existence is obscured by the lack of information regarding their ecological significance, especially throughout the tropics. Perhaps this reflects their largely solitary existence or their minor role in pollinating food crops. Or it simply may reflect their preference toward a nomadic existence, thereby rendering them difficult to study. Whatever the case, little is known about the role they play in their environment. In a few instances, mutualistic relationships have been implicated as a result of close study. Such is the case with members of the genus Heliconius. Both Bates and Müller used members of this group as their models, and it is not difficult to see why. They are abundant throughout the neotropics, are conspicuous owing to their orange and black coloration/patterns, and are slow flyers as compared to most other species. Perhaps this is why they have been reasonably dealt with on more than a taxonomic level.

To the casual observer and expert alike, considerable similarity exists in coloration and patterns within Heliconius butterflies. In a classic example of Müllerian mimicry, Heliconius melpomene and Heliconius erato share nearly identical wing markings and coloration throughout a wide geographic area. In fact, they so closely resemble one another that there is disagreement as to which form acts as the model and which is the mimic. Apparently, this is complicated by their numerical dominance in different parts of their range. However, Mallet (2001) suggested that H. erato serves as the model because it is twice as abundant as H. melpomene throughout most of its range. Whatever the case, both species are unpalatable and, therefore, confirm Müllerian mimicry. Of greater interest is the degree of phenotypic variation noted throughout their range. Nearly a dozen different races are recognized throughout Central and South America and in almost every case, H. melpomene and H. erato possess a counterpart for each other (Turner 1975, 1981; Sheppard et al. 1985; Turner and Mallet 1996; Mallet 2001). These forms are so similar that some scientists speculate that they may have coevolved or codifferentiated from a common ancestral pattern. However, cladograms of races based on mitochondrial DNA sequences of H. erato and H. melpomene are not concordant and indicate that they do not share a common biogeographical history (Joron and Mallet 1998).

Even more interesting are the similarities noted among distinct families of butterflies resembling Heliconids. Members of the Ithomiidae, Danaidae, and Pieridae, and even in a few tailless species of swallowtail butterflies (Papillionidae) possess mimics of Heliconids. One of the more stunning examples occurs in the family Pieridae. Members of this family typically exercise sporadic flight patterns and possess solid yellow, white, or orange wings with few markings. They are fast flyers and are not shy about puddling on riverbanks or in open areas. But Dismorphia amphione, an unusual member of the Pieridae, is phenotypically aligned to the well-patterned orange, black, and yellow, slender bodied, elongate winged Heliconius butterfly, and is a presumed Batesian mimic. The disposition of this butterfly also favors the slow-moving heliconians that gently waft about the neotropical canopy.

Often, large groups of butterflies may simultaneously occupy a mimicry ring, a group of sym-patric species sharing a common warning pattern. The extent of this phenomenon is worldwide. One well-known ring complex exists in the Ecuadorian Amazon. Here, ithomiine butterflies belong to some eight distinct mimicry rings harboring more than 120 distantly related species of butterflies, and in some instances day-flying moths (Joron and Mallet 1998). These relationships are by no means static, as species may move in and out of specific rings. There is considerable experimental evidence that butterflies and moths switch rings through mutations that control key color patterns and distribution features (Mallet and Gilbert 1995). These are usually a result of single gene mutations that result in phenotypic variation; this is thought to have driven the substantial racial variation among H. erato and H. melpomene (Mallet 1986; Turner 1988) throughout their extended range.

On top of being protected via mimicry, Heliconius butterflies possess unusual behavioral characteristics. Interestingly, they are the only butterflies known to eat pollen. They are long lived (with longevity extending for more than a year), often forage along the same routes day after day, and roost communally. Most intriguing are their efforts to avoid areas where they have been netted or harassed even days after the incident (personal observation). In studies assessing their ability to associate flower colors with a reward, it has been shown that they learn colors as quickly as many bees (Oliveira 1998). It is obvious that these butterflies are equipped to deal with a host of pressures not observed in other butterfly families.

Even the larval stages of Heliconius butterflies are protected. A primary strategy lies in the choice of foliage they ingest from the passion vine (Passifloraceae). This is a large plant family in the neotropics and is, therefore, easily accessible to larvae of Heliconius butterflies. A well-known property of this group is that the foliage synthesizes cyanogenic glycosides and cyanohydrins that are considered toxic. Since Heliconius eggs are deposited by the adult butterfly on the underside of the passion vine leaves, larvae hatch with the luxury of not having to seek out a food source. In the process of ingesting the foliage, larvae requisition toxins that render them unpalatable. Actually, the situation is a bit more twisted than described. The story really centers on the ability of the larvae to detoxify cyanogenic glycosides. Apparently, adult butterflies are able to distinguish between the many different species of passion vine (Passiflora) and choose to lay eggs only on those species that produce glycosides that their larvae can detoxify (Spencer 1984). This would seem not only to help protect their offspring but also to reduce competition between like family members that share interest in the same food source. Additionally, by laying eggs on specific vines, larvae may acquire permutations of the same basic toxin that may further help to confuse predators. In this respect, it is the adult that assures larvae will contain a distinct type of unpalatable property. Larvae feeding upon passion vine have the necessary enzymes to detoxify only certain species within this plant group. Therefore, this also tends to eliminate competition by adult female selection. In addition, the production of cyanogens among different species of passion vine may revolve around the degree and extent of attack between butterfly and plant over extended periods. Correspondingly, the ease at which butterflies detoxify components from these plants has undoubtedly helped to refine the relationship.

Many passion vine species have adapted to being parasitized by Heliconius butterflies. It is interesting to observe the variation in leaf morphology encountered within one passion vine. More than a dozen different-looking leaves within the same vine is not uncommon. This seems to be one way that the plant has adapted to tricking the butterfly to look elsewhere. However, if chemical cues direct homing to passion vines, a change in morphology would seem not to be of any significance.

Of related importance, some passionflower vines release attractants from tiny extrafloral nectaries found on the underside of the leaves. These structures produce important amino acids that are sought after and harvested by ants, bees, and wasps (Horn et al. 1984). The presence of these aggressive insects serves to discourage herbivory from invading caterpillars and other small insects. In addition, the nectaries look strikingly similar to butterfly eggs and may serve to trick gravid heliconid butterflies to look for other leaves without eggs (Gilbert 1971). This behavior appears to be important in reducing both sibling and intraspecies competition among larvae and is practiced by many insects. Gravid butterflies of many species often examine the underside of leaves thoroughly before depositing their eggs.

Several genera of Heliconius butterflies have larvae that are spiny. On top of being unpalatable, the spines further help to discourage predators from partaking of an easy meal. The voracious appetite of the larvae, especially in a collective sense, can inflict considerable damage to the host plant. In periods when larvae are not feeding, they sometimes congregate or huddle into a vertical mass that resembles a spiny brush. Individuals cling to a central twig or stripped petiole and bend acutely. This arrangement gives the collective appearance of a botanical structure, possibly that of a wilted or dried spiny inflorescence (see Figure 22-4). This is most likely an example of abstract mimicry (broad generalized features of a plant model in this instance).

V.E. Ecological Consequences of Genetic Structure

Whereas the genetic basis of a selection response might be in part determined by the selective regime under which it arose, a given genetic structure can subsequently influence ecological processes. A famous example in this context is the analysis of the mimetic colorations of Heliconius butterflies. Both H. melpomene and H. erato exist in a number of distinct color morphs or races that occur in nonoverlapping regions of their respective species range. Both taxa are unpalatable to avian predators. In any one location, birds learn to avoid the local wing pattern that is shared between co-occurring Heliconius species form a so-called mimicry ring. The maintenance of local color morphs is due to frequency dependent selection: the locally frequent wing morph is well protected against predation because birds have learned to avoid it, whereas any rare morph that might appear suffers predation from naïve birds that have yet to learn about its distastefulness.

Through the analysis of numerous hybrid crosses between the local races, the total phenotypic variation of the different color morphs within each of the two species could be attributed to 22 and 17 loci respectively. Some of these have large effects on the formation of pattern components such as wing bands and spots, whereas others modify some detail of an existing pattern. There are complex dominance relationships as well as interactions among loci, such that the effect of a given allele at a particular locus depends on the individual genetic background in which the allele is placed (epistasis, cf. Box 1). For any one pair of geographically adjacent mimetic forms within a species, one finds genetic differences at no more than four major loci plus some small number of modifier loci. Nevertheless, this number of loci is sufficient in principle to produce a large number of recombinants upon interbreeding in areas where the racial distribution ranges adjoin. These animals with recombined wing patterns should suffer greatly increased predation, as any of the local birds would not recognize them as distasteful.

In contrast, there is another mimetic system that involves the swallowtail butterfly Papilio memnon. This butterfly is edible, but females gain protection from predation because they mimic other distasteful species. In this case, one finds several different female morphs within populations, which are due to allelic variation at a so-called supergene consisting of several tightly linked loci. As a consequence, only a limited number of phenotypes is produced, each of which may be mimicking different models. The high fitness of these morphs is in part a consequence of the underlying genetic structure. The existence of several unlinked loci as in the case of Heliconius would preclude intrapopulational variation in color morphs, because too many of the possible recombinants would not be adaptive.

4 Beyond Knockouts of Candidate Genes

CRISPR/Cas9-mediated KOs have been a major step forward for those wishing to elucidate the functional role of a number of candidate genes. However, KOs are only the beginning, and a number of research groups are developing CRISPR/Cas9-based techniques to go beyond KOs. For example, as discussed elsewhere in this chapter, mosaic phenotypes may not always be easily interpretable and/or informative of function and in such cases it is desirable to generate stable transgenic lines. A potential positive side effect of generating stable transgenic lines from CRISPR/CAS9-induced mosaic founders is the ability to recover an allelic series of the targeted locus. Mosaicism in germline of the injected founders can lead to transmission of an array of mutant alleles in the next generation (Wei et al., 2014; Yen et al., 2014). In butterflies, the successful transmission of an allelic series of mutants was first demonstrated in D. plexippus using ZFNs (Merlin et al., 2013). Germline transformation in general has also been achieved with Pieris rapae (Stoehr et al., 2015) and B. anynana, using a PiggyBac construct (Chen et al., 2011). The complexity of such germline mutations could be further increased by coinjecting a cocktail of several different sgRNAs targeting the same locus. The utility of an allelic series of protein-coding genes may not be intuitive; however, it could provide a crude but efficient tool to probe the function of noncoding regulatory loci. This approach has been used to functionally validate upstream regulatory elements of the tyrosinase gene in mice (Seruggia et al., 2015).

Such an approach might be feasible to begin dissecting the cis-regulatory elements associated with the reported red pigment patterns in Heliconius species (Nadeau et al., 2012, 2013; Reed et al., 2011; Wallbank et al., 2016). Genomic studies in H. melpomene and its relatives have further narrowed the location of these noncoding regulatory elements down to a few kilobases (Wallbank et al., 2016), and chromatin accessibility profiles and methods are now available in Heliconius to narrow down sites that are differentially active between tissues and stages (Lewis et al., 2016). Targeting these regions simultaneously with several sgRNAs, spaced out across the length of the locus, could potentially generate butterflies carrying regulatory alleles of various lengths (including complete deletions) (Fig. 5). Phenotypic analysis of this panel of alleles in heterozygous (F1 generation) and/or in homozygous (F2 generation) state could aid in identifying the functional elements of these regulatory regions and further refining the location of the causative regions underlying differences in red patterns. Such an approach could also be applied to putative regulatory regions of WntA (Martin et al., 2012) (Fig. 5). Moreover, targeting regulatory regions of the genes, instead of the protein-coding regions, potentially alleviates issues of lethality associated with complete functional KO of pleiotropic and early embryonic genes (Huang et al., 2016). However, it should be borne in mind that genetic variation and enhancer dominance may limit to an extent obtaining quick and easy-to-interpret results.

More precise manipulations such as integrating inducible expression cassettes; for conditional expression (Chen et al., 2011), insertion of small protein tags (e.g. V5 epitope), or larger fluorescent proteins for expression assays, and direct allelic replacement (to identify causative nucleotide(s)) would also benefit from the generation of stable transgenic lines. Such applications rely on using the CRISPR/CAS9 system alongside the cells’ innate HDR machinery which has shown to be conducive for successfully generating knock-ins in butterflies (Zhang and Reed, 2017). These initial efficiencies of HDR-mediated knock-ins appear to be low but could potentially be increased by in vivo linearisation of donor DNA (Irion et al., 2014), the use of single-stranded oligodeoxynucleotides (ssODNs are particularly useful for introducing small changes (Ran et al., 2013), including the Easi-CRISPR technique (Quadros et al., 2017)), suppression of the innate NHEJ machinery (NHEJ and HDR machinery compete to repair DSBs; Singh et al., 2015) and perhaps even avoiding the HDR machinery altogether (Auer et al., 2014). It is likely that finetuning of these and other parameters are likely to lead to a further increase in knock-in efficiency in butterflies. In the future, the application of such experimental strategies may largely be limited by generation times, mating strategies, maintenance of inbred lines, and the ability to rear large numbers in the lab—factors that will vary by species.

The technique of CRISPR/Cas9 has proved a breakthrough for the study of butterfly wing pattern and eyespot development, and their evolution. Exciting new results are being produced at an accelerated rate and further advances of the technique, coupled with other techniques such as transcriptomics to identify novel relevant genes, will rapidly provide us with detailed insights into how the wide variety of beautiful butterfly wings develop and evolve.

Cortex

A second major locus is responsible for switching on and off most white and yellow colour pattern elements in H. erato, H. melpomene and H. cydno (Figure 2) [21,23]. Interestingly this locus also overlaps with two inversions present in certain morphs of H. numata, which control quite different colour patterns of black, orange and yellow spots [31]. H. numata differs from most other Heliconius species in that multiple colour patterns are usually present within a single population and that all colour pattern variation is controlled by multiple alleles at single genetic locus with a strict dominance hierarchy between these alleles [32]. The gene cortex appears to be, at least partially, responsible for these colour pattern variants, with population genomics approaches mapping colour pattern variation within H. erato, H. melpomene and H. numata to within or near this gene and H. melpomene and H. numata showing colour-pattern-associated expression differences of cortex [33••].

Cortex belongs to a family of cell cycle regulators [34], which includes two genes that are highly conserved in all eukaryotes, CDC20/fzy and cdh2/fzr, and have a fundamental role in cell cycle progression [35]. Cortex itself appears to be insect specific and to have a much higher evolutionary rate [33••]. It seems likely that it could control scale cell colour through control of scale developmental rate, as melanic scales are known to develop at a slower rate than scales of other colours across a diversity of lepidoptera [36]. Indeed, the cortex gene also appears to regulate melanic pigmentation in the peppered moth, with the insertion of a transposable element in this gene producing the melanic form that proliferated during the industrial revolution [37•]. Therefore, it seems likely that cortex has a role in scale cell development and pigmentation across all lepidoptera.

Again, the precise functional variants of cortex causing differences in pigmentation patterning are unknown, but appear to be cis-regulatory rather than coding. Cortex has several 5′ untranslated exons (5′ UTRs) spanning a region of over 100 kb, suggesting a complex of dispersed regulatory elements [33••]. In addition to splicing variation of these 5′ UTRs, there are also alternative coding isoforms, some of which show associations with colour pattern. Further work is needed to understand if this splicing variation affects scale pigmentation.

Insect groups involved in patterns of cospeciation

Insects are one of the most successful groups of animals on Earth. They are involved in a multitude of interactions with other organisms such as bacteria, microsporidia, fungi, plants, nematodes, and vertebrates, which make them great models for large-scale cophylogenetic studies (Box 1) and, more generally, coevolution. Patterns of cospeciation have been observed in multiple systems that include insect lineages. Generally, cospeciation is more frequently expected when the life histories of the two interacting lineages are tightly linked such as for vertically transmitted symbionts and their insect hosts, or when interactions are species specific.

Box 1

Glossary

When starting to compare the evolutionary history of interacting lineages, one may stumble on the following terms especially since they are frequently used interchangeably: codivergence, coevolution, cophylogeny, cospeciation. This box attempts to give what we believe are consensual definitions of these terms and others, that may appear confusing.

Coadaptation: Microevolution of two or more interacting species in response to reciprocal selection between them [49].

Codivergence: The parallel divergence of interacting lineages.

Codiversification: Correlative diversification of two or more interactive lineages or organisms. Speciation events in one lineage are correlated with speciation events in a second lineage [30•].

Coevolution: Reciprocal natural selection occurring during reciprocal evolutionary interaction between two or more organisms.

Cophylogenetics: field of research that focuses on the macro scale coevolutionary associations formed between the phylogenies of interacting lineages [50].

Cophylogeny = cophylogenetic analysis explores the relationships between the phylogenetic trees of interacting lineages. In most analyses, the goals are to determine whether the match/congruence between the two (or more) trees is significant and to find the best explanation for the differences between the trees [20].

Cospeciation: The matching of speciation events and their co-occurrence upon the time between two or more interacting lineages.

Phylogenetic tracking: A pattern in which cladogenesis occurs in parallel in two interacting lineages of organisms, but the speciation events are not synchronous (i.e. one lineage speciates first and is followed by speciation in the other).

Mutualistic interactions between insects and their endosymbiotic bacteria are ubiquitous and occur in many insect groups (e.g. [1]). These interactions facilitate the use by the insects of nutritional resources from various difficult to digest sources, such as sap, wood, etc. For Hemipteran insects, endosymbionts provide nutrients to their insect hosts that feed on sap, a resource that miss essential amino acids. In these obligate interactions, congruence among phylogenies has been demonstrated (aphids and Buchnera [2], leafhoppers and Sulcia/Baumannia [3], stinkbugs (Plataspidae) and a specific gut bacterium (γ-Proteobacteria) that is vertically transmitted from the mother to her developing eggs [4]. Strict cospeciation also occurs between cockroaches or termites and their obligate endosymbionts (Blattabacterium) and/or gut microbiomes that exhibit a cellulolytic and diazotrophic activity [5,6]. Another case of mutualistic interaction in which cospeciation may be expected is mimicry (e.g. between species of Heliconius butterflies H. erato and H. melpomene). For this example, results remain unclear. The first phylogenetic analyses found contrasting histories with topological and temporal incongruence that argued against codivergence [7•]. However, using coalescent based methods and cutting-edge cophylogenetic methods, Cuthill and Charleston [8] concluded that the evolutionary history of H. erato and H. melpomene was compatible with a number of temporally congruent codivergence events.

Plant-herbivore or host-parasite coevolution can also result in patterns of cospeciation, though much less frequently. Mutualistic interactions involving insects are diverse and ecologically important [9]. In these highly specialized relationships, the two interacting species mutually benefit from their interactions. A few plant genera (Ficus, Yucca, and Glochidion) are exclusively pollinated by obligate seed-parasitic insects (Agaonidae wasps, Prodoxidae and Epicephala moths respectively). Insects pollinate the flowers and oviposit in the plant ovaries where larvae subsequently feed on a subset of the developing seeds. In these nursery pollination mutualisms, only a few studies have investigated the level of cocladogenesis between sparsely sampled phylogenies of the mutualistic partners. Cophylogenetic analyses of yuccas and their pollinator moths showed congruence between the phylogenies, though this pattern may be better explained by biogeographic factors than by coevolution (as within a lineage, yucca species and their hosts mostly occur in allopatry) [10]. Both cospeciation and host shifts have played an important role throughout the evolutionary history of the Glochidion and Epicephala moth system [11]. Finally, the largest cophylogenetic study published so far that focussed on the fig–fig wasps mutualism [12] highlighted long-term cospeciation between the partners (Box 2).

Box 2

Cospeciation in figs and fig wasps

Co-speciation may exist in a few obligate mutualistic partnerships. Among groups that exhibit intimate interactions, figs and their pollinating fig wasps may represent the first demonstrated case of long-term (∼75 myr) codivergence in an insect–plant association. The fig–fig wasp mutualism is an important focal association for the study of co-speciation and co-diversification [51,52]. To accurately determine the extent of shared evolutionary history between figs and fig pollinating wasps, phylogenies of both partners must be reconstructed reliably.

However, until now, testing cospeciation of figs and fig wasps has been hampered by the poor resolution of the deep nodes of both phylogenies. In all phylogenetic studies of the partners, the relationships among the fig sections and the agaonid genera (i.e. the backbone topology) were difficult to resolve [53,54]. Using ca 200 pairs of interacting fig and fig wasp species and about 5.5 kb DNA sequence for each species, Cruaud et al. [12] tested cospeciation between figs and fig wasps. However, the analysis was hampered by the poor resolution of the phylogenies of the mutualists and possibly also long-branch attraction artifacts for the fig trees. The phylogeny of figs was globally congruent with previous hypotheses, but relationships of deeper nodes were poorly supported. Similarly, the phylogeny of the fig pollinators (belonging to the chalcid family Agaonidae) was not resolved and the relationships among the major clades were unclear. Further studies are thus needed to end up with a clear scenario of what happened during the evolutionary history of this unique association.

Expanded views of different parts of the host (Ficus) and pollinator (Agaonidae) trees published by Cruaud et al. [12] showing different degrees of congruence/divergence. Top: Congruence between Ficus from the section Conosycea (Asia) and their three genera of pollinators (Eupristina, Deilagaon, Waterstoniella) is higher than between (Bottom) Ficus from the section Americana (Neotropics) and the genus Pegoscapus, though all these fig trees belong to the subgenus Urostigma (big stranglers). Patterns must be nevertheless considered with caution, as some nodes are poorly resolved (Black squares: BP > 70% and PPMrBayes or PPBEAST > 0.95; white squares: BP > 70% or PPMrBayes or PPBEAST > 0.95).

Regarding parasites, studies have been published on ectoparasites of animals such as chewing or sucking lice that develop on the body of birds (ducks, doves, flamingos, pelecans, penguins, pigeons, seabirds or toucans) or mammals (primates, rodents). Cospeciation patterns have been demonstrated between sucking lice and heteromyid rodents [13], chewing lice and marine birds [14] and body louse and New World doves [15]. However, for most groups of lice and their vertebrate hosts, phylogenetic congruence is not the rule. Cophylogeny between parasitoids and their insect hosts has been rarely investigated (e.g. [16]), though such studies may help to set up effective biocontrol programs (e.g. reduce unintended effects). On a more general note, most cophylogenetic studies are conducted on two-lineage systems and only a few focused on more complex systems (e.g. moth/parasitoids/plants [17]), though this may help to better understand dynamics among multiple trophic layers in an ecosystem or specialized interaction.