Phylogeny, environment and sexual communication across the Drosophila genus

Throughout the 20th century, laboratory studies on the fruit fly Drosophila melanogaster have progressed our understanding of mechanisms surrounding development, genetic heredity, neuronal circuitry and social behaviour. This ever-increasing bounty of knowledge gained from D. melanogaster within the laboratory as a model organism can obscure interest in its evolutionary history. For example, teasing apart the neuronal circuits and genes mediating mating behaviour does not rely on appreciating that D. melanogaster originates in Africa and only increased its geographic range with human activity thousands of years ago (David and Capy, 1988). Despite our history together, it has been only within the past dozen years that we have begun to appreciate the social life of D. melanogaster. Various studies have demonstrated that D. melanogaster displays distinct courtship/mating behaviour, aggressive behaviour, social learning and emergent group-based behaviour (Billeter and Levine, 2013; Schneider et al., 2012a). These behaviours are thought to be mediated via sight (Agrawal et al., 2014; Kohatsu and Yamamoto, 2015), taste (Schneider et al., 2012b), touch (Schneider et al., 2012b; Ramdya et al., 2015) and olfaction (Duménil et al., 2016). Yet D. melanogaster is simply one species nested within the approximately 1500 classified species of the Drosophila genus (Markow and O'Grady, 2006). Upon considering the diversity of drosophilids, a critical question becomes evident: what is ‘Drosophila’ social communication? How much of it is shared, and how much of it is unique to a particular species within its niche?

There are many examples of striking behaviours observed contrasting those of D. melanogaster. The Hawaiian Drosophila species, with their conspicuously bright colouration and territoriality during mating, have evolved complex courtship rituals (Spieth, 1974). Cactophilic species also differ in their mating behaviour from D. melanogaster in that flies mate away from feeding sites (Markow, 1988), and females produce courtship song duets with males (O'Grady and Markow, 2012). Male flies of species within the obscura group have been observed to regurgitate a nutritious pre-nuptial gift for the female to consume during courtship (Steele, 1986). Across all the aforementioned taxonomic groups, many species have been reported to be unable to mate in the dark (Grossfield, 1971) and to vary in mating frequency (Markow, 2002). One possible explanation for such varied behaviours is the extreme abiotic factors that characterize some of these species' niches. For instance, cactophiles such as D. mojavensis and other repleta species endure temperatures ranging from below 5°C at night to over 40°C during the day (Gibbs et al., 2003). As a result, it is likely that these cactophilic species evolved physiological tolerance in response to these extreme abiotic factors. An example of a physical trait that confers resistance to these extreme conditions are the cuticular lipids [also known as cuticular hydrocarbons (CHCs)], which could also impact the behaviour of these species.

Although comparative studies of behaviour and social signalling are scarce, a wealth of comparative physiological studies across Drosophila species have characterized their responses to abiotic stresses, such as heat (Stratman and Markow, 1998; Krebs and Loeschcke, 1995; Loeschcke et al., 1994; Hoffmann et al., 2003; Kellermann et al., 2012b), cold (David et al., 1998; Gibert and Huey, 2001; Kellermann et al., 2012a; Andersen et al., 2015; Gibert et al., 2001; Bertoli et al., 2010) and desiccation (Gibbs and Matzkin, 2001; Kellermann et al., 2012a; Matzkin et al., 2009; Gibbs and Markow, 2001; Kalra et al., 2014; Hoffmann and Harshman, 1999). Not surprisingly, abiotic stress tolerances have been reported to be reliable indicators of a species’ geographic distribution (Kellermann et al., 2012b; Andersen et al., 2015). This geographic correlation to stress tolerance is strengthened through comparative studies that examined stress resistance across many species. Using phylogenetic comparative methods, these studies outlined the extent to which convergent evolution and common descent influences the evolution of stress resistance (Kellermann et al., 2012a,b). Another aspect that arises out of these comparative studies, and supported by established climate models (Peel et al., 2007), is an attractive generalization that ‘arid species’ possess high abiotic stress tolerance, ‘tropical species’ possess low abiotic stress tolerance and ‘temperate species’ are intermediate. The exceptions are species known to have world-wide distributions, which are often referred to as ‘cosmopolitan species’. Examples of these include D. melanogaster, D. simulans, D. hydei, D. funebris and D. immigrans (Markow, 2015). While these categories represent over-simplified generalizations, they serve as discussion points to highlight that environmental factors, such as climate and precipitation patterns, may constrain the evolution of social behaviour.

Social communication in Drosophila involves a complex combination of visual, auditory, mechanosensory and chemosensory cues (Bontonou and Wicker-Thomas, 2014). The role of these sensory modalities and their neurological input and output mechanisms are best known in D. melanogaster because of the organism's well-characterized genome and abundance of genetic tools for neurological manipulation. For this Review, we are interested in elaborating on aspects of social communication beyond this single species. We are interested in elucidating the extent to which diverse Drosophila species vary in their social behaviours and how phylogeny and climatic factors influence these behavioural differences. What is a species-specific behaviour? Given their ecological range, what individual adaptations do these species have? Would such adaptations engender differences in their behaviour? How do these behavioural differences map onto the Drosophila phylogeny? Is there evidence of convergent behaviours? As we will discuss, these topics are under-represented in the Drosophila literature. Here, we review studies that have shed light on how visual and chemical communication displayed during courtship/mating varies across species. We choose to focus on these two sensory modalities because they play important roles within the larger context of social communication of D.melanogaster (Agrawal et al., 2014). We limit the context of social communication to courtship/mating because it is the only social behaviour studied across a wide variety of species. Throughout this Review, we will focus on a small subset of the overall diversity of the Drosophila genus and, whenever possible, connect behavioural differences to phylogeny and distribution patterns (i.e. tropical, temperate, arid and cosmopolitan; see Fig. 1).

Visual perception of D. melanogaster, brought by their compound eyes composed of photoreceptive ommatidia individually connected to their optic lobes, provides the fly with a wide range of vision essential for flight, but limits colour and object recognition (Borst, 2014). However, some studies have shown that the limited object recognition D. melanogaster possesses is sufficient for social mate choice learning (e.g. Mery et al., 2009). It has also been demonstrated that this species prefers to mate in the dark (Hardeland, 1972), suggesting that visual cues are not essential for mate communication. However, this is not the case throughout the Drosophila genus.

Since the 1940s, discrepancies were being documented between Drosophila species on their visual communication during mating. Drosophila subobscura, for example, will not mate in the absence of light, which contrasted with the behaviour of the closely related species D. pseudoobscura and D. persimilis (Wallace and Dobzhansky, 1946). Indeed, observations of the courtship sequence of D. subobscura involve a ‘wing-dance’ where the male faces the female and side-steps while spreading his wings (Pinsker and Doschek, 1979). Wing removal in D. subobscura has been demonstrated to have a detrimental effect on copulatory behaviours in males (Grossfield, 1968). This provides evidence of the wing-dance serving as a crucial visual cue during courtship, as D. subobscura has been reported to not produce courtship song through wing vibration (Ewing and Bennet-Clark, 1968).

Here, we compiled a list of 47 species and noted their ability to mate in the dark. Most of this list is attributed to Grossfield (1971), who categorized 42 species into three classes: (Class I) light independent, meaning the species can mate in the dark; (Class II) facultative dark mating, meaning darkness has a slight repressive effect on mating; and (Class III) dark repressed, meaning the species cannot mate in the dark (Grossfield, 1971). By displaying a phylogeny of species and mapping their ability to mate in the dark on the tree, we see that most species rely on the presence of light in order to successfully mate (Fig. 2). This suggests, based on our limited sample, that diurnal mating may be the ancestral trait of all Drosophila species. We see evidence of conserved traits across closely related species, as the tropical immigrans group species are all Class III species (Fig. 2). Also, the quinaria species mostly share the inability to mate in the dark (Fig. 2). On the other end of this behavioural spectrum, the ability to mate in the dark appears to have evolved multiple times, and we note that ‘cosmopolitan’ species such as D. melanogaster, D. hydei, D. mercatorum, D. immigrans, D. funebris and D. ananassae are not closely related, yet they freely mate in the dark. It is conceivable that these species, being human commensals (Markow, 2015; Markow and O'Grady, 2008), thrive in dark microhabitats created by humans and such environments might have selected for visually optional mate communication.

Although some of the dark-repressed class III species, such as D. subobscura and the quinaria species, are distributed in temperate regions of the world, 11 out of 21 Class III species are distributed in tropical regions (Figs 1 and 2). This is consistent with the observation that many tropical species exhibit colourful morphologies and visually display stereotyped courtship rituals. The Hawaiian picture wing species are a classic example of Drosophila that depend on visual cues during courtship. These species show diversity in wing pattern colours, and display clear sexual dimorphism based on their morphology (Edwards et al., 2007). The diverse wing patterns on the male flies are thought to offer species-specific visual cues to the female during courtship (Spieth, 1974). Similarly, D. grimshawi, a Hawaiian representative, cannot mate in the dark and is also a Class III species (Grossfield, 1971). This complements accounts of Hawaiian courtship involving males securing a territorial lek and facing females during a complex visual-based courtship ritual (Spieth, 1974).

Yet wing patterns in Drosophila are not traits restricted to the tropical Hawaiian clade, as similar characteristics are observed in other distantly related tropical species. A comparative study investigating sexual characters in a variety of melanogaster species found throughout Australasia (referred to as ‘Oriental’ species; Fig. 1) has revealed that the dark spots found on the wings of D. eugracilis, D. suzukii, D. pseudotakahashii, D. fuyamai, D. biarmipes and D. elegans have independent evolutionary origins (Kopp and True, 2002). It is thought that these wing spots serve as a crucial visual cue during the courtship of these species (Fuyama, 1979). Given that the Hawaiian and some of the ‘Oriental’ species are tropical, and that they have evolved melanised wing spots independently, one hypothesis is that the common abiotic conditions of tropical environments in different areas of the world play a role in shaping these convergent traits. Perhaps tropical conditions with stable climates relax naturally selective pressures on Drosophila and enable rapid evolutionary changes to behaviour through sexually selective mechanisms, resulting in convergence. Another possibility is that the relative density and variety of species in tropical regions require mate coordination that can communicate precisely at short distances without being influenced by surrounding messages (as could be the case with chemical signalling). However, out of the ‘Oriental’ species mentioned with the wing spots, there is only evidence of D. suzukii displaying courtship repression in the dark (Grossfield, 1971). To gather support for such hypotheses, testing the courtship repression of a larger sample of tropical ‘Oriental’ species with and without melanised wing spots would be a good start.

Unlike the tropical species with sexually dimorphic wing spots, D. mojavensis males and females look alike and during courtship the males mostly position themselves behind the female (Markow and O'Grady, 2005; O'Grady and Markow, 2012; Krebs and Bean, 1991). This lack of sexual dimorphism is seen in other virilis and repleta species, such as D. virilis (temperate) and D. hydei (cosmopolitan), which are examples of two species that freely mate in the dark (Grossfield, 1971). Although the arid cactophiles – D. mojavensis, D. arizonae and D. buzzatii – are well studied, we are not aware of any experiments that directly tested the ability of these species to mate in the dark, although the close relatives of these species, D. mulleri (arid) and D. mercatorum (cosmopolitan), are Class II and Class I species, respectively (Grossfield, 1971; Ikeda, 1976). In addition, locomotor activity experiments revealed that, in a laboratory setting, D. mojavensis displays peak activity during dawn and dusk (Hardeland and Stange, 1973). Its activity cycle may have evolved in response to the stress of nocturnal cold temperatures, which may inhibit nocturnal activity and mating, combined with the aridity and heat of the day. These two factors are avoided by scheduling activity during transition times, as is seen in D.mojavensis, which mates predominately after sunrise (Krebs and Bean, 1991). This could also be true in the Sonoran Desert endemic species, D. nigrospiracula and D. pachea, that, like D.mojavensis, tolerate desiccation stress and temperatures that may exceed 40°C during the day and between 0 and 5°C at night (Gibbs et al., 2003). Experiments that screen a large sample of arid species for their ability to copulate in the dark would offer insight into our speculation that the arid climates influenced the evolution of mating activity during dawn, when temperatures in these desert environments are neither too hot nor too cold. Consequently, this may have relinquished visual signalling during courtship, and perhaps other social behaviours because, at this time of day, there may be lower sun exposure on cactus necroses.

Out of the 47 species we illustrate in Fig. 2, the majority of species possess either a slight repression of courtship activity (Class II) or complete repression of courtship activity (Class III) when maintained in the absence of light. The majority of cosmopolitan species are Class I (5 out of 8), the majority of temperate species are at least slightly repressed in the dark (9 out of 20 are Class II, 7 out of 20 are Class III), tropical species are mostly of Class III (11 out of 16; Fig. 2) and while the three arid species are mostly Class III (2 out of 3), although only one of these species is nested within the repleta group. Even if the phylogeny and ecology are sometimes correlated (such as the tropical nasuta cluster), we see that light-dependent courtship is a complex trait that appears to have been shaped by both phylogenetic and ecological pressures. Species within similar ecologies share mating habits with regards to the light, which may outline some aspect of environmental influence on the dependence of visual cues for social communication. Unfortunately, this sample of 47 species does not offer an accurate reflection of the approximately 1500 Drosophila species. Any patterns regarding the distribution, phylogeny and emphasis of visual cues in these species could result from sampling bias. To complicate matters, the timing of mating events varies across species (Hardeland, 1972) and across populations of the same species (Billeter et al., 2012). With this in mind, it would be beneficial to examine whether populations of the same species sampled across the globe are all classified together. This would enable one to directly test associations between environment and visual-dependent courtship communication. One hypothesis is that a species broadly categorized as dark-repressed (Class III), but collected from a more arid and warm region, could exhibit mating behaviours more similar to Class II species in its attempts to thermo-regulate and avoid activity during peak temperatures. It also seems worthwhile to identify trends in the phylogeny regarding light-dependent mating to highlight exceptions – these exceptional cases tentatively hint at ecological niches that have spurned adaptation. One such example is D. willistoni and D. nebulosa. These are two related tropical species, and one can mate in the dark (D.willistoni) while the other cannot (D.nebulosa; see Fig. 2). The life history that caused this divergence is an open question.

In all Drosophila species, volatile pheromones and non-volatile cuticular lipids are involved in many aspects of social interactions. The synthesis of these chemical compounds is quite plastic and depends on temperature (Savarit and Ferveur, 2002), humidity (Frentiu and Chenoweth, 2010), photoperiod and social environment (Krupp et al., 2008). It has been observed across many species that during courtship males tap the abdomen of females (Pinsker and Doschek, 1979; Spieth, 1974), and this behaviour has also been observed in groups of flies of the same sex (Schneider et al., 2012b; Ramdya et al., 2015; Sexton and Stalker, 1961). This tapping behaviour is assumed to involve the fly tasting chemical secretions in gustatory receptors on their forelegs (Vosshall and Stocker, 2007), though purely tactile information can also be conveyed (Ramdya et al., 2015). The chemical compounds that are thought to be exchanged during these tapping behaviours are the CHCs that are synthesized in the oenocytes (Billeter et al., 2009). Extracting the CHCs of a fly may purify a blend of compounds that vary quantitatively and qualitatively based on the length of the molecule in carbons, the saturation and branching.

Across species, CHCs are known to vary in quantity. For instance, the arid virilis–repleta species tend to synthesize higher quantities than Sophophora and other tropical species. Drosophila mojavensis has been reported to contain two orders of magnitude more CHC extract in its cuticle compared with D. melanogaster (Ferveur, 2005). The high quantity of CHCs in species adapted to drier environments and the latitudinal variation of CHCs within D. melanogaster (Frentiu and Chenoweth, 2010) suggest that CHCs also play a role in desiccation resistance. Contrary to CHC quantity, the length of CHC chains shows more conservation across species groups: Hawaiian and Sophophora species produce the shortest CHCs (between 23 and 29 carbons in most species), the virilis species produce CHCs that are slightly longer (between 22 and 31 carbons), and the repleta species synthesize the longest (between 28 and 40 carbons; Bontonou and Wicker-Thomas, 2014). We compiled a list of known CHC compounds synthesized in 36 species and computed the average length of CHCs (based on the number of carbons) to help visualize any climatic correlation of CHC differences across Drosophila (Fig. 3). When incorporating a variety of climatic variables (see Fig. 3), we see that environment predicts average CHC length to a reasonable degree across all environments (Fig. 3A), as well as within environments (Fig. 3B–D). Temperate and tropical species overlap in the CHCs they produce, whereas arid species cluster on their own (Fig. 3). Two conclusions of our preliminary regressions are worth mentioning as they deserve further investigation. One is that the temperature of the warmest month has a negative effect on CHC length (at least in a combined model or temperate species on their own). If (and potentially why) large CHCs are detrimental in temperate species that experience extreme warm months is an interesting question. Another general conclusion is that precipitation can have different effects on tropical (positive) or arid (negative) species. This supports the notion that it is not simply the measured regional abiotic factors, but potentially interactions with complex microhabitats that shape Drosophila's physiology. The long hydrocarbon chains observed in arid repleta species likely correlates with their increased thermal and desiccation resistance (Gibbs and Matzkin, 2001; Kellermann et al., 2012a; Matzkin et al., 2009). Both thermal and desiccation tolerance have shown phylogenetic correlation (Kellermann et al., 2012a,b), and the long CHC chain length exclusive to repleta species are intuitively very similar in pattern. Perhaps the common ancestor of the repleta radiation acquired the ability to polymerize elongated cuticular hydrocarbons and this, in turn, enabled these species to thrive in arid climates. This speculation is an example of how traits that are physiological adaptations to abiotic stresses can also play a role in social communication. Whether specific CHCs in these arid repleta species influence mating behaviour, much like how 7,11-heptacosadiene stimulates courtship behaviour in male D. melanogaster (Billeter et al., 2009), is currently not well understood.

Species differences in CHCs are not only correlated to abiotic selection pressures. Even the particular blends of cuticular hydrocarbons appear to be a trait that expresses species identification signals that have higher variation between closely related species, suggesting a link to reinforcement barriers. A classic example is D. melanogaster and D. simulans, two sibling species that often coexist in sympatry (Markow, 2015). Drosophila melanogaster females produce unsaturated CHCs with two double bonds in the carbon chain (dienes) and the males do not, making this species sexually dimorphic in terms of its CHC blends. However, D. simulans does not produce dienes, and dienes exhibit anti-aphrodisiac qualities to D. simulans males (Billeter et al., 2009). Considering the CHC profile of other closely related species reveals that D. sechellia and D. erecta females produce dienes like D. melanogaster, but D. yakuba females do not (Ferveur, 2005; Cobb and Jallon, 1990). Drosophila yakuba and D. erecta are sibling species that are sympatric in western Africa (Lachaise et al., 1988), and this parallels the relationship between D. melanogaster and D. simulans, suggesting character displacement. Although the size and quantity of CHCs a species synthesizes could depend on environmental pressures, such as heat and aridity, the qualitative aspects of CHC profiles (i.e. the range of compounds synthesized) appear sensitive to reproductive isolation mechanisms. Interestingly, an experimental evolution study utilizing D. serrata and D. birchii found that exposing the two species to novel diets (rice and corn) caused a divergence in both CHCs and female mating preference (Rundle et al., 2005). Consequently, the distribution of individual CHC compounds may not hold any phylogenetic correlation and may evolve frequently based on reinforcement, mate selection and shifts in host resources. Accordingly, by compiling a list of species that are known to lack dimorphism in their CHC profile and considering the phylogenetic relationships of these species, we see that this is a trait that appears to have independent origins (Fig. 4). Interestingly, species that show courtship repression in the dark, such as D. grimshawi, D. suzukii, D. simulans and D. mauritiana, are examples of species without dimorphic CHCs (Fig. 4). The lack of dimorphism in their CHCs may partially explain why these species depend on visual cues for courtship and mating. Of special note is that the arid species sampled are exclusively dimorphic in CHCs. This adds credence to our earlier hypothesis of arid repleta species underemphasizing visual signals during courtship. One hypothesis that follows from this observation is that the environmental pressures on arid species are so high that dimorphic CHC profiles became essential for social communication either in dark and cool microhabitats, or at a time of day when sun exposure is reduced.

Cuticular hydrocarbons serve as chemical signals at short range and are mainly detected through physical contact. In D. melanogaster, longer range olfactory communication is known to be carried out through cis-vaccenyl acetate (cVA), which is produced in the ejaculatory bulb of male flies (Butterworth, 1969). This compound is thought to play multiple roles: aggregation (Bartelt et al., 1985), aggression (Vander Meer et al., 1986; Wang and Anderson, 2010) and courtship inhibition (Vander Meer et al., 1986). Males transfer cVA to females during copulation, which then acts as an ‘anti-aphrodisiac’ to other males, and while cVA is in the female reproductive tract it coats the eggs as they are oviposited. Drosophila melanogaster females tend to oviposit communally and this significantly affects the survival of larvae because large aggregates of larvae can outcompete harmful microbes, enhancing their survival (Wertheim, 2001; Rohlfs, 2005). However, cVA is restricted only to melanogaster species and distantly related immigrans–tripunctata species (Symonds and Wertheim, 2005). Across other species, pheromone compounds that facilitate aggregation are generally volatile esters, ketones or unsaturated hydrocarbons (Hedlund et al., 1996). Examples of these include ethyl hexanoate, found in immigrans–tripunctata species (Symonds and Wertheim, 2005), 10-heptadecanone, found in repleta species (Bartelt et al., 1989; Schaner and Jackson, 1992), and heneicosene, found in virilis species (Bartelt et al., 1986). Symonds and Wertheim (2005) mapped the known aggregation pheromones in 28 Drosophila species and found a phylogenetic pattern to the distribution of these pheromones across species. While the authors suggested that pheromonal evolution occurred gradually in Drosophila (Symonds and Wertheim, 2005), we wonder how many examples of convergent evolution (e.g. cVA in D. immigrans and D. melanogaster) correlate with the abiotic conditions of the environment they are distributed in.

A recent study demonstrated that species within the Drosophila subgenus tend to be sexually dimorphic for a family of triglycerides (TAGs). Interestingly, these TAGs are synthesized in the male ejaculatory bulb of these species and appear to function as anti-aphrodisiacs, much like cVA in D. melanogaster (Chin et al., 2014). Expression of these compounds also correlated with the age of reproductive maturity in D. mojavensis and D. arizonae, suggesting that these compounds are necessary for courtship (Chin et al., 2014). These compounds are highly conserved in the virilis–repleta species and completely absent in species within the Sophophora subgenus. One hypothesis is that the utilization of TAGs for pheromonal communication may be related to the adaptations of arid and cactophilic virilis–repleta species; that is, perhaps they are more stable under high temperatures and low humidity. As a result, TAGs, instead of cVA, may have evolved in these species as a result of being more effective for chemical communication. Another recent study described a novel pheromone called CH503, which is only expressed in the melanogaster subgroup species, yet appears to act as another anti-aphrodisiac pheromone in a variety of species (Ng et al., 2014). In fact, species in the Drosophila subgenus that do not express this compound, such as D. virilis and D. mojavensis, displayed a stronger courtship suppression response to females perfumed with this compound than did species that express this compound (Ng et al., 2014). This suggests that there is conservation in the response to pheromones across species in the subgenera Drosophila and Sophophora. The synthesis of pheromone profiles may vary across species, but their receptors may display conservation across species.

We have come across several open issues relating to CHCs across species. Although CHC quantity appears somewhat correlated with aridity, climate seems to predict CHC length and arid species strongly cluster together (Fig. 3). Untangling reproductive barriers, ecology and phylogeny in CHC blends is an ongoing field of study. There are also several tantalizing aspects that appear mysterious, including cVA as a convergent aggregation pheromone, and the general response to the species-specific hydrocarbon CH503. More studies dedicated to pheromone compounds and their behavioural effects in other Drosophila species might lead to a broader understanding of the evolution of social behaviour in Drosophila. One issue that keeps returning is the lack of a comprehensive database in which to catalogue collection sites and hydrocarbon profiles. This is sorely needed, as the blend of CHCs that a species produces can vary with latitude (Frentiu and Chenoweth, 2010). This variation across populations of the same species can reveal greater insights into the diversity and correlations between CHC blends and the environment. It is likely that pheromone evolution across Drosophila species is influenced by a combination of life history factors (such as host resource preference, temperature tolerance, etc.) and phylogenetic history. Disentangling each factor promises to be an exciting and rewarding field of study.

Compared with D. melanogaster, the sensory modalities of social communication are less thoroughly studied in other species. Studies that do focus on the social behaviour of other Drosophila species tend to focus on those with striking ecologies, such as the Hawaiian species (Spieth, 1974; Edwards et al., 2007; Widemo and Johansson, 2006) and the cactophilic species (Markow, 1988; Krebs and Markow, 1989; Krebs and Bean, 1991; Chin et al., 2014), or on species that coexist in sympatry to study reinforcement mechanisms (Noor et al., 2001; Noor and Coyne, 1996; Markow, 1981; Civetta and Cantor, 2003; Giglio and Dyer, 2013; Gleason et al., 2012). While valuable, these studies often only examine select aspects of certain species, which creates gaps between comparative data. To visualize these gaps, in this Review, we compiled qualitative visual and chemical communication data of 75 species (Table 1). Only nine of these species overlap in having been tested for mating activity in the dark and having their CHC profiles examined (Table 1).

Generally speaking, unlike D. melanogaster, most species appear to require light and visual cues for successful copulation (Fig. 2). Many of the species that freely copulate in the dark are cosmopolitan (Figs 1,2). Most tropical species, irrespective of their placement in the Drosophila phylogeny, appear to depend on visual cues, and this could explain why species found in tropical regions, such as the Hawaiian species, are sexually dimorphic in their abdomen and wing colour morphology (Edwards et al., 2007). Also, some tropical species display stereotyped visual displays during their courtship ritual that involve directly facing the female (Markow and O'Grady, 2005; Spieth, 1974). These visual-based courtship rituals may result from relaxed abiotic selective pressures on tropical species, but enhanced sexually selective pressures. We wonder whether the cooler temperatures and higher precipitation of tropical environments influences the effects of sexual selection and reproductive isolation mechanisms on CHC modification because these seemingly species-rich environments require species boundaries. Very few arid species have been tested for their ability to mate in the dark (Fig. 2, Table 1). Based on the lack of visual communication that has been observed in arid species from field studies (O'Grady and Markow, 2012), perhaps non-visual communication is more efficient as arid Drosophila may rely on temperature-defined timing in activity in order to escape the extreme conditions. Further comparative investigations of CHC profiles and disentangling the phylogenetic and ecological relevance of CHCs and pheromones across many Drosophila species could offer insight into such hypotheses. Arid species show the most differences in the size of their CHC chains (Table 1, Fig. 3) compared with tropical and temperate species. This, in addition to their conserved desiccation tolerance (Kellermann et al., 2012a), suggests genetic adaptations acting on CHC metabolism may have occurred that enabled these species to colonize arid habitats.

Throughout this Review, we have made broad generalizations across species through ‘climatic categories’. These categories are based on literature annotations, and each encompasses a wide range of localities spanning the globe (see Fig. 1 and Table 1). Although modelling abiotic conditions, such as temperature and precipitation, should be regional enough to make conclusions valid, we have not considered confounds from species that could have been sampled from microclimates. For instance, arid species, such as D. buzzatii, have been collected from arid corridors of otherwise tropical areas of Brazil (Oliveira et al., 2011); thus pressures concerning water and precipitation will be different than in arid areas adjacent to temperate regions. This highlights the difficulties in making generalizations, even within a species, and underscores the need to record where and under what conditions the samples are acquired in any study. We are similarly aware that the adaptability of certain types (especially cosmopolitan) can generate species that do not conform to the general conclusions and ‘categorizations’ of tropical, temperate or arid. Rather, these categories are fluid based on the collection sites, and how long these species are bred under constant conditions within the laboratory. We see the identification and study of species within these categories as a way of understanding how adaptations to the environment can have far-reaching implications, and intra-specific variation across habitats is an incredibly exciting area of study.

The currently accepted Drosophila phylogeny is quite comprehensive, with over 180 species (van der Linde et al., 2010). Future comparative studies need to sample species based on phylogenetic relationships, much like Kellermann and colleagues (Kellermann et al., 2012a,b). Such studies, whether they are behavioural, physiological or genetic, may dispel generalizations we have made, such as all repleta species being arid in distribution with high stress resistance and long CHCs. In fact, D. ellisoni, D. pavani and D. gibberosa are such exceptions as they are distributed in tropical regions, yet are not distantly related to D. mojavensis. In fact, D. ellisoni has low desiccation and cold tolerance, unlike its close relatives (Kellermann et al., 2012a). Does this species behave similarly to D. mojavensis at the social level? Or does it behave more like other tropical species? Does it mate in the dark? Does it have a similar hydrocarbon profile? Such questions can easily be answered and may offer valuable insight into how visual, chemical and other sensory modalities in Drosophila evolve based on environmental selective pressures.

Untangling these possibilities to understand the causes of variation between the species will ultimately rely on identifying the gene pathways and mechanisms responsible for the adaptation to the ecological and social pressures. Although we are at the stage where identifying gene pathways associated with behavioural evolution is more difficult in species other than D. melanogaster, studies have been conducted that were dedicated to untangling the phenotypic aspects of social communication during courtship. Such studies involved physically blinding flies (Gleason et al., 2012; Grossfield, 1971; Giglio and Dyer, 2013; Wallace and Dobzhansky, 1946), removing antennae to crudely ablate olfaction (Gleason et al., 2012; Giglio and Dyer, 2013) and removing wings to remove auditory and visual cues (Fuyama, 1979; Giglio and Dyer, 2013; Gleason et al., 2012; Grossfield, 1968). Recent publications have conducted all of the above sensory disruptions on D. willistoni and D. nebulosa (Gleason et al., 2012) and on D. subquinaria and D. recens (Giglio and Dyer, 2013) to untangle the influence of each sensory modality on courtship, and each study used two species that are closely related. More studies similar to these are desperately needed. This may lead to the identification of evolutionary mechanisms at play in the adaptation of Drosophila behaviour to the social group and other features of their environment.

We thank the anonymous reviewers for their constructive criticism that helped this manuscript grow from its previous state.