Fruit Scent: Biochemistry, Ecological Function, and Evolution
Fruit scent plays an important role in human preference and has thus been studied primarily in the context of agricultural science. In wild species, fruit scent has long been speculated to play a role in mediating the mutualistic interaction between plants and fruit-eating animals that disperse their seeds. Yet until recently, empirical studies addressing this hypothesis have been all but absent. Studies in the past decade emphasized the ecological role of fruit scent as an animal attractant, as well as its evolution as a ripeness signal. But data are still limited and many questions remain open. This chapter summarizes recent developments in the study of the chemical ecology and evolution of wild fruit scent. It explores the chemistry and biochemistry of fruit scent, its use by various important seed dispersal vectors, its evolution, and other functions it may fulfill. We end with recommendation for future studies, in the hope that the next decade will be at least as fruitful as the previous one.
KeywordsCo-evolution Constraints Frugivory Mutualism Odor Olfaction Seed dispersal Sense of smell
Like all plant tissues, fruits are packed with secondary metabolites . This diversity of secondary metabolites has been suggested to fulfill a plethora of ecological functions, from attraction of seed-dispersing frugivores (fruit-eating animals), through regulation of their interaction with the seeds, to repellance of fruit antagonists [2, 3]. In contrast, it has been suggested that fruit secondary metabolites are primarily defensive and in most cases similar to those present in nonreproductive tissue or unripe fruits, thus making them an extension of the plant’s general line of defense [4, 5]. The assertion that fruit secondary metabolites are primarily the result of “leakage” from nonreproductive tissue has since been refuted [6, 7]. Nonetheless, among the myriad possible functions of secondary metabolites, defense against microbial, insect, and vertebrate antagonists has been in the focus of most studies [6, 7, 8, 9, 10, 11]. This is not surprising given that in other plant tissues, the most explored and probably also prominent function of plant secondary metabolites is defense .
The prominence of defense in the discussion regarding fruit secondary metabolites can probably be attributed to the fact that most studies have focused on the large, nonvolatile, secondary metabolites. Indeed, many large, often water-soluble, secondary metabolites such as cyanogenic glucosides, glucosides, and polyphenols are defensive [12, 13]. But much less attention has been given to volatile secondary compounds (or volatile organic compounds: henceforth VOCs) – lighter and often more hydrophobic compounds that constitute what we would colloquially recognize as “scent” or “odor” [14, 15]. VOCs are ubiquitous in fruits and some are likely to play a role in fruit defense [16, 17]. Yet as recognized by the much more developed study of commercial fruit production, they are responsible for the aroma of fruits and their attractivity to human consumers [14, 18, 19]. Thus, another reason explaining the focus on the defensive rather than attractive function of fruit secondary metabolites is likely to have originated from the fact that VOCs of wild fruits have until recently rarely been studied [14, 15].
Over the past decade, the interest in wild fruit VOCs has increased substantially, in particular in their role as an attractant of animal seed dispersers . This has been the result of a growing understanding that despite various constraints [21, 22], many fruit traits are likely to have evolved in response to animal seed disperser preferences [23, 24, 25, 26]. Along with the proliferation of methods that allow analysis of scent [27, 28], the decades-old hypothesis that ripe fruit scent – i.e., VOC profile – is an evolved trait whose function is to attract seed dispersers  began receiving support . In that, the study of fruit traits and seed dispersal has followed the much more mature field of pollination ecology, in which the role of floral scent as a pollinator attractant has been recognized for decades [20, 23, 30, 31].
Yet despite the growing body of knowledge and understanding of the role of fruit scent as an attractor of seed dispersers, the number of studies that conducted chemical characterizations of wild fruit scent is still low, and is based on a narrow taxonomic coverage from only a few geographical regions, exclusively in the tropics. Moreover, the focus on fruit scent as an animal attraction mechanism risks downplaying other ecological roles which fruit scent may fulfill . Thus, it is important to remember that the field is still in its initial stages and that there are many more open questions than definite answers .
This chapter will summarize the latest developments in the study of the ecology and evolution of wild fruit scent. It will first examine what chemicals tend to characterize fruit scent and describe the basic biochemical processes leading to their synthesis. It will then examine how fruit scent is used as a food detection and selection by various animals and present the supporting evidence for the adaptive hypothesis of fruit scent. It will end with a discussion of other factors determining fruit scent evolution, and conclude with recommendations for future studies.
2 Fruit Scent: Chemistry, Biochemistry, and Patterns of Emission
2.1 Chemistry and Biochemistry
Fruits of wild and cultivated species emit complex mixtures that can comprise dozens to hundreds of different VOCs . Most of these belong to three prominent chemical classes, terpenoids, fatty acid derivatives, and aromatic compounds [32, 33], although some fruits also contain rarer compounds such as amino acid derivatives [14, 15].
Terpenoids are synthesized via two separate biosynthetic pathways and are a construct of two or three basic C5 (isoprene) units, which are then modified to create the end product [35, 42]. The enormous diversity of plant terpenoids is a result of the latter stage, in which a narrow range of precursors are transformed into thousands of different end products by various members of the terpene synthase (TPS) family [42, 43 ]. TPSs are highly non-specific, and thus terpenoids are always emitted in diverse mixtures [34, 44]. Terpenoids are involved in plant defense, either through direct toxicity or in indirect defense systems, as recruiting signals for natural enemies of antagonists [13, 34, 45]. At the same time, their presence in ripe fruit scent has been demonstrated to attract bats  and primates .
2.1.2 Fatty Acid Derivatives
Fatty acid derivatives such as saturated and unsaturated hydrocarbons and alcohols, esters, and aldehydes, ketones, and carboxylic acids are among the largest classes of volatile secondary metabolites in flowers . Volatile fatty acid derivatives are primarily synthesized by degrading C18 linoleic and linolenic acids into C12 and C6 alcohols (e.g., n-hexanol, 2- or 3-hexenol), aldehydes (e.g., n-hexanal, 2- or 3-hexenal), and carboxylic acids (e.g., 3-hexenyl acetate) [33, 48, 49, 50]. These compounds and their derivatives are very common in plant green tissue and are often collectively called “green leaf volatiles” .
Fatty acid derivatives are highly common in fruits of wild and cultivated species [14, 15, 39, 40, 51]. Some fatty acid derivatives such as various green leaf volatiles can be found in both ripe and unripe fruits and are possibly involved in fruit defense . In contrast, aliphatic esters tend to be more common in ripe [18, 51, 52, 53] and even more in overripe  fruits. Interestingly, while the synthesis of most plant VOCs is based on self-biosynthetic machinery and precursors, esters are at least partially synthesized by bacteria-produced precursors: Esters are synthesized by a condensation of a carboxylic acid and an alcohol, and alcohols are often the limiting factor in ester synthesis in fruits . Ethanol, a precursor of ethyl esters, is a product of sugar fermentation by microbes , and treatment of fruits with antibiotics leads to a substantial reduction in ester emission  (Fig.1, 4–5).
2.1.3 Aromatic Compounds
Aromatic compounds are those which contain at least one conjugated planar ring. Like terpenoids and fatty acid derivatives, they are very common in flowers across plant families [32, 47] and are involved in pollinator attraction  and leaf defense . Aromatic VOCs are common in ripe fruit [15, 38, 39, 40, 51] and constitute a significant portion of the scent emitted by wild fruits in Uganda and Madagascar . The vast majority of volatile plant aromatic compounds are synthesized by a complex biosynthetic process whose precursors are aromatic amino acids synthesized via the shikimate pathway [33, 57, 58]. VOC synthesis is the result of deamination of the amino acid L-phenylalanine and reduction to C9 compounds [33, 49, 57]. One volatile product of this process is trans-cinnamic acid, which was identified in several Malagasy fruit species in its methyl ester form  (Fig. 1, 6). Further reduction of trans-cinnamic acid by removal of a C2 unit is the basis for synthesis of many other aromatic VOCs . Other common aromatic compounds include methyl- and ethyl-salicylate, both esterized forms of the phytopheromone salicylic acid [59, 60]. While of lesser importance compared to other plant VOCs like terpenoids, aromatic compounds also play a role in both active and passive leaf defense [13, 58] and possibly play a role in fruit defense too .
2.1.4 Nitrogen- and Sulfur-Containing Compounds
While terpenoids, fatty acid derivatives, and aromatic compounds dominate the fruit scent profiles of most cultivated and wild species [14, 15, 51], fruits of some species emit less common compounds. Several wild fruits in Madagascar have been found to contain nitrogen- and sulfur-containing compounds [15, 51], although even in these species the relative contribution of these compounds to the scent profile was minor. However, compounds of these classes dominate the scent profiles of at least one wild species, the (in)famous durian (Durio sp.). Durian fruits, which are also cultivated in Southeast Asia, are known for their strong and distinct scent, although some cultivars are almost odorless . While there has been some debate over which specific VOCs are responsible for the foul scent, it is known that Durian scent contains, in addition to more typical compounds like alcohols and esters, primarily sulfur-containing compounds [61, 62, 63, 64]. Notably, as opposed to some past claims, the sulfur-containing compounds are synthesized by the fruit itself and not by bacteria inhabiting the flesh . Interestingly, nitrogen- and sulfur-containing compounds are a product of protein metabolism . This led to the speculation that their presence and amount in fruit scent could serve as an honest signal of protein content .
2.2 Patterns of Scent Emission
Although based on few cultivated model systems, it appears that at least in some species fruit VOCs are synthesized by specialized cells situated on the fruit’s skin . It is generally assumed that VOC release is predominantly passive through diffusion and that therefore it can only be regulated by up- or downregulating VOC synthesis . Indeed, VOC synthesis is strongly regulated by the presence and activity of the participating enzymes, i.e., regulated both by gene expression and the transcription level . However, diffusion of largely hydrophobic VOCs at published rates would require extremely high concentrations which are potentially harmful, and it has therefore been proposed that plant VOCs are actively emitted using transmembrane structures  which are yet to be identified. Either way, emission of plant VOCs is a controlled process which is adjusted to the developmental stage of a particular tissue or even on smaller scales such as the circadian rhythm [68, 69].
2.2.1 The Ripening Process
The chemicals described above were documented in the scent of wild and cultivated ripe fruits. However, the data available per species is almost exclusively based on snapshots – records taken in a single moment, often in an unstandardized moment in the fruit’s maturation process. However, fruit scent is not static. Many – but notably not all – fruits change their scent qualitatively and quantitatively once ripe [14, 37, 39, 40, 51]. In cultivated species, which have probably been artificially selected to increase their aroma, the amount of scent emitted by ripe fruits increases by a factor of up to 30 [14, 70]. In wild fruits the median increase in 19 species specializing on seed dispersal by primates was found to be 2.3 . Notably, the same study found no increase in the amount of scent emitted in fruits consumed by sympatric bird-dispersed species, highlighting the not surprising similarity between human artificial selection and natural selection by our closest living relatives. An increase in the amount of scent emitted by ripe lemur-consumed fruits was found in other sites in Madagascar  and in two fig species from Panama . Fruit scent also changes qualitatively upon ripeness, with increased emission of compounds that were present in low proportions or fully absent in unripe fruits [14, 37, 39, 40, 51].
Studies conducted in the wild [37, 39, 40, 51, 71] have aimed to record dichotomous behavior among animals while interacting with ripe and unripe fruits. This choice led in most cases to a comparison of two snapshots of fruit scent – one before the onset of ripening and the other around its peak. Thus, studies of wild species are not informative with regard to the process of change in fruit scent. A single exception is a study by Sánchez et al. , who reported a decrease in the amounts of ethanol and acetaldehyde in rusty figs (Ficus rubiginosa).
In contrast, several studies on cultivated fruits have tracked the changes in aroma compounds, along with changes in fruit quality and seed development [14, 73]. With regard to the ripening process, fruits can be roughly divided into two groups: climacteric and non-climacteric. Climacteric fruits are those whose final stage of development is characterized by increased respiration and ethylene production, and they tend to exhibit a rapid ripening process, while non-climacteric fruits mature more gradually [73, 74]. In climacteric cultivated species, the shift to ripe fruit scent is rapid. For example, in peaches, the major shift in fruit scent occurs abruptly, and once the seeds approach their final weight, as emission of three GLVs characteristic to unripe fruits decreases, while the emission of compounds typical to ripe fruits increases . Similarly, another study of apricot and plum X apricot hybrids compared the volatile profiles of fruits in three late developmental stages: mature green, commercial ripe, and tree ripe. In most cases, the latter two were similar, indicating that the shift in scent occurs abruptly and only when the seeds are mature . In snake fruits (Salacca edulis), the prominent acids and alcohols increase more gradually during maturation, but ester emission skyrockets abruptly around the time the fruits become softer and mature . Taken together, this is exactly the pattern expected in cases where fruit scent is used by animals to find or identify ripe fruits [37, 38, 46, 78], as plants are expected to be selected to begin attracting them only after the seeds are viable. This is similar to the patterns of floral scent emission, which tend to peak when the flowers become ready for pollination . Yet while many wild fruits exhibit a rapid maturation process (Nevo, personal observation), it is rarely known whether wild fruits are climacteric.
2.2.2 Circadian Rhythm
In flowers, emission of VOCs can be constant or change rhythmically over the 24 h cycle, often the case in plants pollinated by nocturnal animals [30, 49]. Diel variation in fruit scent is far less investigated, and data are available for only two fig species from India. Mature syconia of Ficus benghalensis, a species dispersed by both diurnal birds and nocturnal bats, change their scent over the 24-h cycle: day VOC emissions are dominated by sesquiterpenes and fatty acid derivatives, while night emissions are substantially poorer in sesquiterpenes and contain more aliphatic esters and aromatic compounds . In contrast, mature Ficus racemosa syconia do not show day-night differences in their scent profiles , even though they do show diel cycle variance in earlier stages of their development .
It is unknown whether fruits or mature syconia of other species alter their scent over the 24-h cycle. Yet it is likely to be common given the prevalence of this phenomenon in flowers [30, 49] and the fact that fruits tend to be more generalist than flower, i.e., they interact with a wider range of animal mutualists  and thus more likely to interact with diurnal, cathemeral, and nocturnal frugivores. For example, species like Ficus maxima are dispersed by both bats  and diurnal primates . Bats and primates use their sense of smell differently: bats can rely on olfactory cues to detect and locate fruits , while primates do so only for selection over short distance . Therefore, this and other similar species are excellent candidates to examine whether fruits1 are selected to emit different olfactory signals at different times of the day.
3 Fruit Scent and Seed Disperser Attraction
The role of fruit scent as an attractant of vertebrate seed dispersal vectors has been the main focus of most work on fruit VOCs in recent years , although the idea that fruit scent is used by olfactorily oriented frugivores is decades old. Early works have integrated it into the framework of the Dispersal Syndrome Hypothesis, according to which fruit characteristics have evolved in response to the traits of their primary seed disperser [29, 83, 84]. Yet empirical tests of the hypothesis that fruit scent has evolved as a signal for seed dispersers have until recently been absent, possibly due to the predominance of the view that fruit traits are not strongly selected by frugivores in the last 15 years of the last century [21, 22, 85, 86] – a timeframe in which the understanding of floral scent evolution has exploded [30, 32, 47]. Another factor has been that chemical communication between fruits and frugivores is more common in tropical regions, in which chemical sampling and analysis tend to be more challenging. Yet the growing support for the dispersal syndrome hypothesis [24, 26, 87] and increasing availability of techniques allowing chemical sampling and analysis [27, 28] have led to a renewed interest in the question. We first examine this question from the animal side, asking how and whether animals may use fruit scent to find and identify ripe fruits, i.e., whether fruit scent is a useful cue for frugivores. We then move on to examine whether in cases in which it is used by animals, fruit scent can be considered an evolved signal which is selected to fulfill this function.
3.1 Fruits Scent as a Cue
With 1200 known species, bats are the second-most diverse group of mammals . Frugivory has evolved independently in the Old and New Worlds, and in both systems bats are important seed dispersers, contributing to early succession and, in the Old World, recruitment of canopy species . As nocturnal animals, bats can rely on their vision less than diurnal species, although many retain dichromatic vision and may rely on vision more than previously considered . Some bats lineages have evolved to use sonar  and can echolocate flowers and fruits which have presumably evolved specialized structures that reflect back their calls [20, 92, 93, 94, 95]. But echolocation is not present in most Old World frugivorous bats , and thus a major sensory trajectory for frugivorous bats is olfaction . Olfaction also plays a role in bat pollination, a relationship which is primarily facilitated by chemical communication  or a combination of olfaction and echolocation .
Reliance on olfaction has been demonstrated in behavioral tests that focused primarily on New World frugivorous bats. New World bats have been reported to be attracted to fig scent [98, 99], and early experiments showed attraction to the scent of bananas, which are consumed by local bats but are not wild and thus possibly not representative . Thies et al.  showed that New World Carollia perspicillata and C. castanea use the scent of Piper fruits to identify ripe fruits. In a series of experiments, they showed that unripe fruits are rejected and that artificial fruits are approached only when impregnated with the scent of ripe fruits. C. perspicillata were also shown to possess high olfactory sensitivity to a series of esters, alcohols, and carboxylic acids common in fruits . Similarly, in an experimental setting, New World Artibeus watsoni and Vampyressa pusilla showed clear preference to ripe over unripe fruits, were attracted to experimental devices that emitted the scent of ripe fruits, and rejected dry-frozen fruits that retained the morphological features of ripe fruits but did not emit scent . In contrast, it should be noted that in some bat-plant interactions (Phyllostomus hastatus feeding on Gurania spinulosa), fruit scent does not seem to play a role .
Experiments with Old World frugivorous bats have been rarer. Some tried but could not record reliance on olfaction in fruit foraging [72, 103]. However, these experiments focused primarily on ethanol and methanol, which are not typical plant secondary metabolites. In another study short-nosed fruit bats (Cynopterus brachyotis) were shown to prefer ripe fruits over unripe fruits of two fig species and to rely primarily on scent to find and identify ripe fruits in an experimental setting . A follow-up study tested the response of the same Old World bat species and New World Jamaican fruit bats (Artibeus jamaicensis) to the scent of fig species from both habitats . Both species were attracted to scent of figs from their respective habitats, but only the Neotropical species were attracted to the scent of unknown Paleotropical figs . Since Old World frugivorous bats are older , show similar patterns of olfactory receptor evolution , and in most cases cannot echolocate , it is very likely that their ability and tendency to use their sense of smell for food detection and selection are comparable to that of New World bats.
All studies which tested the attraction of bats to fruit scent used intact fruits, fruit extracts, or synthetic mixtures. It is thus unknown whether any individual compound is particularly attractive to them. While bat-pollinated flowers tend to emit uncommon sulfur-containing compounds , bat-consumed fruits tend to emit common VOCs [37, 38, 40]. Monoterpenes are particularly common in the scent of both Paleotropical and Neotropical bat-consumed figs and have been proposed to play an important role in attracting them .
Along with bats and birds, primates are one of the biggest groups of seed dispersers in tropical systems . Primate seed dispersal plays a pivotal role in a complex web of interactions between plants, primary and secondary seed dispersers [105, 106]. The fact that most sympatric frugivorous primates overlap in their diets but vary in their body size, movement patterns, and group size renders them, as a group, highly effective seed dispersers . For example, since many primate species are large and arboreal, many species are unlikely to visit early-phase secondary forests in which trees are still too small to support them. But small-bodied primates like tamarins (Saguinus spp.) do venture into regenerating forests and thus fulfill a function similar to that of birds and bats by effectively dispersing seeds into secondary forests .
As opposed to bats, most primates are diurnal and, relative to other mammals, possess excellent color vision [109, 110]. As a result, visual cues play a role in the process of ripe fruit detection and selection [111, 112, 113], and primate color vision is likely to have exerted non-negligible selection pressures on the evolution of fruit color [114, 115].
However, primates are now recognized to possess an excellent sense of smell that is often on par with that of mammals like dogs and rodents , and it becomes increasingly clear that olfaction plays a major role in the feeding ecology of primates, primarily for discrimination between ripe and unripe fruits [15, 78]. Until recently, most studies that demonstrated reliance on olfaction for fruit selection did not consider the chemical properties of fruits . Nevo et al.  showed that spider monkeys (Ateles geoffroyi) can discriminate between synthetic mixtures mimicking the scent of ripe and unripe fruits of two Neotropical fruit species. Notably, the monkeys can discriminate between the scents of ripe and unripe fruits even when individual compounds in the scent of unripe fruits are manipulated to match the concentration in ripe fruit scent. This indicates that identification of ripe fruits is not based on individual compounds and thus less sensitive to within-species variance in fruit VOC content, which has been found in studies that sampled multiple fruits per species [39, 51].
In the field, two recent studies quantified the relationship between fruit olfactory conspicuousness, defined as the difference between ripe and unripe fruit scent, and the tendency of primates to sniff fruits before ingesting or rejecting them. Red-bellied lemurs (Eulemur rubriventer) are more likely to sniff fruits of species which increase the amount of scent upon ripeness or change the chemical composition of ripe fruits . In the neotropics, white-faced capuchins (Cebus capucinus imitator) increase the rate of sniffing when feeding on fruits of species in which the amount of scent emitted by ripe fruits is larger . Another study found no relationship between sniffing behavior in brown lemurs (Eulemur fulvus) and the overall amount of scent emitted by ripe fruits , indicating that the determining factor is not scent per se but the olfactory conspicuousness of the fruit, i.e., how different it is from conspecific unripe fruits [15, 39]. However, scent in this study  was corrected for the surface area of the fruit. Thus, the variable analyzed was not the amount of scent available for the lemurs but the amount emitted by a unit of surface area. This procedure is meaningful when studying, for example, the costs of scent emission. But from an ecological perspective, in fruit selection, the animal is exposed to the scent emitted by a single fruit, which is therefore a more appropriate measurement.
Primate-consumed fruits tend to emit common VOCs: terpenoids, aromatic compounds, and fatty acid derivatives [15, 39, 51]. Lemur-consumed fruits in Madagascar – especially those which attract more olfactory investigation by lemurs – tend to be rich in aliphatic esters . Some fruits also emit nitrogen- and sulfur-containing compounds [15, 16, 51].
Frugivorous birds are important seed dispersers across the tropics and are probably the most important animal seed disperser in temperate regions . The most dominant group of frugivorous birds are the passerines, although frugivory is also present in non-negligible numbers among woodpeckers, parrots, and pigeons . Birds possess an excellent color vision. Most species are tetrachromatic, i.e., possess one more pigment type than the best color-discriminating primates, among them humans [109, 121, 122]. As a result, they tend to rely strongly on fruit color for detection and selection [24, 26, 123] and have exerted substantial selection pressures on fruit color [24, 26, 124].
It is a long-held notion that frugivorous birds tend not to rely on olfaction as strongly as mammals [29, 39, 40, 51, 125]. Evidence supporting this notion has been rather circumstantial and was based primarily on the relatively simple olfactory anatomy of many birds, especially passerines [126, 127, 128], and the reports that, unlike mammals, frugivorous birds do not sniff fruits before ingesting . Bird-consumed fruits tend to emit lower amounts of VOCs  and change their scent profiles in ripeness less than mammal-consumed sympatric species [39, 40, 51]. This parallels the pattern observed in the bird pollination syndrome: in contrast to insect- and bat-pollinated flowers, bird-pollinated flowers tend to emit only scant amounts of scent . In addition to the high reliance on vision, it is thus often assumed that frugivorous tend not to rely strongly on olfactory cues.
However, these notions should be taken with cause since it is possible that bird reliance on olfaction is simply less visible to human observers. Several studies demonstrated the ability of passerines to use chemical cues in various situations [128, 129, 130, 131]. Performance in olfactory sensitivity and discrimination capacity tests is difficult to compare directly to other frugivores. The olfactory sensitivity of passerines is somewhat low but, in the range, relative to other birds and primates [128, 132]. In conditioning tests, blue tits (Cyanistes caeruleus) – a non-frugivorous passerine – learn to identify lavender oil, but their performance is rather low compared to, for example, primates [127, 133, 134].
Thus, the absence of evidence for reliance on fruit scent, along with the high visual capacities of birds and the fact that they tend to feed on less olfactory conspicuous fruits, indicates that frugivorous birds tend to rely less on fruit scent for food detection or selection. Yet given the many functions olfaction plays in the lives of many birds, including passerines , and the fact that non-frugivorous birds possess excellent olfaction , it is well likely that future studies would demonstrate that this notion is oversimplified.
3.1.4 Other Animals
As the primary agents of seed dispersal in most systems, bats, birds, and primates have received most of the focus in the study of the interaction between frugivory, sensory ecology, and seed dispersal. But other animals consume fruits and may use fruit scent to detect or identify ripe fruits. Elephants possess an excellent olfactory system [136, 137] which can be employed to find plant material , mate choice , and even identify human ethnic groups which hunt them . Balanites wilsoniana, a species growing in continental Africa, is dispersed by elephants [141, 142] and emits a strong scent rich in aliphatic esters (Nevo, Valenta, Chapman, unpublished data), which the elephants are likely to use to find and select fruit.
Although birds are the most important animal seed disperser in temperate regions, some plant species receive dispersal services from mammals, many of them nocturnal . While less is known about the sensory ecology of less studied animals like hedgehogs, they possess very large main olfactory bulbs  and are very likely to rely on fruit scent when possible.
Finally, seed dispersal by invertebrates is fairly common. The most prominent invertebrate seed disperser is ants , which are attracted by fatty acids  present on the elaiosome – a lipid-rich appendage which serves as a reward and is functionally similar to fleshy fruits. Another invertebrate which occasionally provides seed dispersal services is slugs [147, 148], although neither their relative importance nor their reliance on scent has been investigated.
3.2 Fruit Scent: A Cue or an Evolved Signal?
While the previous section considered only the animal side of the interaction, i.e., how different animals may use fruit scent to find and identify ripe fruits, there is strong evidence that olfactory conspicuousness has evolved in some species to promote seed dispersal. The first is convergent evolution across taxa which share a dispersal vector. Comparing bat- and bird-dispersed figs, Hodgkison et al.  found that bat-dispersed figs have converged to emit monoterpenes, although, as they acknowledge, monoterpenes are not the dominant VOC class in a few other bat-dispersed figs [40, 79]. More qualitatively, Nevo et al.  found that ripe lemur-dispersed fruits are much more likely to emit aliphatic esters than do sympatric bird-dispersed species. In both studies, the fact that species which share a disperser at least partially converged in their ripe fruit scent chemistry is an indication that there is some selective pressure exerted by these seed dispersers.
A second line of evidence supporting the hypothesis that scent is an evolved signal to seed dispersal comes from a handful of studies which looked at the patterns of change in fruit scent upon ripeness and compared them between sympatric species that rely on mammal and bird seed dispersers. Studying three fig (Ficus spp.) species, Borges et al.  showed that at the dispersal stage, only bat-dispersed figs had a unique scent which probably drives attraction of seed-dispersing bats. Bat-dispersed fruits also tend to emit stronger scents than sympatric bird-dispersed fruits . Taking a similar approach but going beyond the Ficus model system, Nevo et al.  showed that in a system of four Neotropical species, primate-dispersed fruits change their scent upon ripeness, while bird-dispersed species do not. This pattern was replicated on a much larger model system in Madagascar, where it was shown that species which specialize on lemur seed dispersal change their scent – qualitatively and quantitatively – significantly more than sympatric bird-dispersed species . The pattern that emerges from these studies is that even though the ripening process is accompanied by much biochemical activity which may affect fruit scent, the change in fruit scent upon ripeness is much greater in species which interact with olfactorily oriented mammals.
3.3 Multimodality: Color and Scent
Animals rarely rely on a single sensory modality, and it would therefore be naive to think that any of the animals discussed above relies solely on scent to find fruits. Using different senses, animals respond to a combination of signals and cues which can be either complementary or redundant [20, 149]. In the former, different cues provide different information, while in the latter the information is the same and the function of the redundant signals is either to provide backup or to ensure that a wide range of frugivores, which emphasize different senses and receive the message.
In fruit foraging and selection, olfaction is often used along with vision or echolocation in bats. Thies et al.  found that olfaction and echolocation play complementary roles, as the former is used for longer-distance detection and the latter for more fine-scale final localization of fruits. In contrast, in primates, olfaction probably plays an important role in close-range selection within a patch [46, 51, 78], while visual cues have been hypothesized to be most relevant for identification of fruit patches over longer distances . However, olfactory and visual cues can also be redundant. This can be demonstrated in polymorphous primate species, in which some individuals possess full trichromatic vision while the rest are dichromats, i.e., red-green color blind [109, 110]. While feeding in the same patch, dichromatic white-faced capuchin monkeys (Cebus capucinus imitator) sniff fruits significantly more than trichromatic group members [117, 150]. This indicates that dichromatic individuals compensate for the lesser access to visual cues by acquiring more information through olfaction. Interestingly, ripe fruit ingestion rates are higher in trichromats , possibly indicating that either visual cues are more accurate or, more realistically, that their acquisition is more time-efficient.
4 Other Factors Affecting Fruit Scent Evolution
While plants are under selection to advertise ripeness through fruit scent, other factors are likely to drive the evolution of fruit scent. These include other adaptive functions scent may fulfill, trade-offs, and various constraints.
4.1 Fruit Defense
Like other plant parts, ripe fleshy fruits are subjected to attack by vertebrate, invertebrate, and microbial antagonists . VOCs are routinely involved in leaf defense  but are rarely considered in fruit defense. Nonetheless, since a major fraction of fruit VOCs such as terpenes, green leaf volatiles, and phenolics play some defensive role in other plant parts, they are likely to be involved in fruit defense too . Thus, plants may be selected to alter the emission rates of individual or multiple VOCs, and hence change their scent, in response to antagonists. At the same time, even compounds which are defensive in other tissues or play a defensive role in fruits may primarily be selected due to their secondary function in attracting seed dispersers. This would parallel the evolutionary pathway of floral scent, which has in many cases evolved secondarily out of VOC-based defense mechanisms . For example, limonene, a monoterpene which dominates the scent of oranges and has been considered to play a defensive role, is in fact an attractant of vertebrate and invertebrate antagonists . This may apply to many compounds that are considered to be defensive solely based on their broad chemical characteristics.
4.2 Developmental and Phylogenetic Constraints
Fruit scent composition may be affected by both phylogenetic and developmental constraints. Often defined in different ways, we refer to developmental constraints as the tendency of fruits to emit VOCs that are present in unripe fruits or other plant parts, not because of their fitness benefit in the fruits but because it is developmentally impossible, or too costly, for the plant to change the VOC profile of only ripe fruits. We refer to phylogenetic constraints as the tendency of a species to possess a trait not because of its fitness benefits to its own ecological circumstances but because it was inherited from an ancestor. The two are inherently connected, as, for example, developmental constraints would slow down adaptive change and hence render closely related taxa more similar.
4.2.1 Developmental Constraints
While hardly studied, the role of developmental constraints in determining ripe fruit scent is probably marginal. Nonvolatile ripe fruit secondary metabolites are independent of other plant parts and robust to changes in the abiotic environment, thus indicating that their presence is adaptive rather than a by-product of regulatory or biochemical processes originating outside the fruit [6, 7].
The strongest argument against a significant role of developmental constraints on ripe fruit scent is the significant change, qualitatively and quantitatively, in the scent of fruits as they mature [37, 39, 40, 51]. Since ripe fruits develop from unripe fruits, the fact that a countless number of species changes their scent profile drastically indicates that selection can effectively alter ripe fruit scent. Nonetheless, it should be remembered that different scents – i.e., different unique mixtures of VOCs – can originate from the same biochemical pathways . For example, species such as Micronychia macrophylla, a lemur-dispersed species from Madagascar, change the scent of ripe fruits qualitatively and quantitatively, but do so primarily using terpenoids , which tend to originate from few biochemical pathways . The tendency to emit chemically similar odorants may be the result of some constraints. But this is not universal: in the same system, Ficus tiliifolia, another lemur-dispersed species, changes its scent profile from a terpene-dominated VOC bouquet in unripe fruits to an aliphatic acid-dominated scent in ripe fruits .
Finally, another factor that could constrain the amount of scent emitted by a fruit is simply its size. Fruit size is one of the factors most malleable to selection by frugivores: in bird-dispersed species, there is an upper cap determined by bird gape width [154, 155], and in species that rely on larger and more energy-demanding frugivores, fruits tend to be bigger . This could have a strong effect on the potential of effective chemical signaling in fruits, as all else being equal, larger fruits would emit stronger scents. In cases where the olfactory signal is selected to allow animals to identify that an individual fruit is ripe [46, 78], an individual fruit needs to emit an amount of scent strong enough to be detected. As a consequence, the costs of olfactory signaling in small fruits would be much higher and might drive them not to signal ripeness through scent.
4.2.2 Phylogenetic Constraints
Phylogenetic constraints are hard to detect since they are a function of the time since speciation, selection pressures, and other constraints. Yet a common method to approximate them is to observe to what extent closely related taxa are similar. Although fruit scent is harder to compare between species due to its multidimensionality (a scent bouquet is composed of dozens, if not hundreds, of VOCs), a few studies addressed the question whether ripe fruits of closely related taxa tend to emit similar scents.
In a community of 30 species from Madagascar, Nevo et al.  found no effect of phylogeny on ripe fruit scent. In a cluster analysis, closely related taxa were found to be more similar to other species than to congeneric or confamilial species. These results have been replicated on a sample of 49 species from Uganda (Nevo et al., unpublished data) and South Germany . Within the fig clade (Ficus spp.), a similar trend was found as on a global scale, as far-related taxa with similar ecology (bat dispersal) were found to be more similar to one another than they are to more closely related bird-dispersed species . However, at a more local level, closely related taxa that are dispersed by bats did show clustering, indicating some phylogenetic conservatism in fruit scent .
An important point is that all these studies dealt with the different VOCs in fruit scent profiles as independent variables. In other words, they assume that a switch from compound A to B is equally likely to a switch from A to C. This assumption is problematic because some compounds are synthesized through the same pathways, and thus switches between them are more likely . A recent study offers a method to address this issue by integrating the biochemical pathways to statistical analyses . However, this approach is not easily implemented in large datasets in which many compounds are not fully identified, and hence not all biochemical pathways are known.
5 Conclusions and Future Directions
The understanding of the evolution and ecological functions of fruit scent has evolved tremendously in the past decade. The VOC profiles of ripe, and sometimes unripe, fruits of dozens of species from the neotropics, continental Africa, Madagascar, and Southeast Asia have been published and used to address various evolutionary and ecological questions. In combination with some behavioral studies, they have shown that fruits emit a tremendous diversity of scents which are used by animals to detect and identify them and that animal behavior has in turn exerted selection pressures on some fruits to become olfactorily conspicuous. At the same time, much more is required to fully understand the selection pressures and constraints which shape the diversity of wild fruit aroma.
An aspect which clearly lags behind chemical characterization, but is equally important to answer both ecological and evolutionary questions, is behavioral essays. As discussed above, many studies examined either scent-based fruit foraging and selection or fruit scent chemistry, but only a handful [37, 38, 39, 46, 51] did both. Yet even as it becomes clear that several groups of animals rely on fruit scent, many questions remain open – most of them can only be addressed through systematic behavioral essays of the kind that has become ubiquitous in insect chemical ecology : To what aspects of fruit scent do they respond? What information they seek? On which odorants they rely? Behavioral studies are also paramount to answering the question whether and to what extent frugivorous birds use fruit scent. It is a common assumption that they do not, or at least do so substantially less than mammals [39, 40, 51]. While there is evidence suggesting that this assumption is to some extent true, it should be verified and rigorously tested in behavioral tests.
Behavioral tests with vertebrates are particularly challenging: wild population densities are low, many animals avoid interactions with humans, and their intelligence and ability to learn complicate many experimental designs which have worked well with invertebrates. Some of these challenges can be met in more controlled experiments with captive animals, which are in turn hindered by the fact that captive animals are in some cases not good representative of wild behavior. Thus, a combination of wild and captive approaches is probably necessary to address many of the questions regarding the use of scent by animal seed dispersers.
Comparative studies of fruit scent would benefit greatly from an increased standardization in the field, which could allow syntheses based on the results of studies conducted by different groups and in different locations. At the moment, the use of different techniques renders most comparison between studies very unreliable . Yet a global comparative approach is crucial to pinpoint the multiple selection pressures and constraints which resulted in contemporary patterns of scent released by fruits. At the same time, it is also important not to forgo higher-resolution studies of individual species or narrow lineages, which are more suitable for integrating factors like the biochemistry of fruit scent and other functions it may fulfill. The contrast between higher-scale lack of phylogenetic signal [38, 51] and its absence in lower scales  and the fact that both should be studied in the context of the biochemical pathways behind fruit scent  demonstrates this point. We are in hope that further integration in the field would take these steps so that the next decade will be at least as fruitful as the previous one.
Including mature fig syconia, which are functionally equivalent.
ON was funded by the German Science Foundation (Deutsche Forschungsgemeinschaft; grant nr. 2156/1-1) while working on this chapter. Dr. Kim Valenta and Prof. Colin A. Chapman were heavily involved in data collection in Uganda, which was used for some of the unpublished results cited here.
- 1.Crozier A, Yokota T, Jaganath IB, Marks SC, Saltmarsh M, Clifford MN (2006) Secondary metabolites in fruits, vegetables, beverages and other plant based dietary components. In: Crozier A, Clifford MN, Ashihara H (eds) Plant secondary metabolites. Blackwell, Oxford, UK, pp 208–302CrossRefGoogle Scholar
- 28.Kalko EKV, Ayasse M (2009) Study and analysis of odor involved in the behavioral ecology of bats. In: Kunz TH, Parsons S (eds) Ecological and behavioral methods for the study of bat, 2nd edn. The Johns Hopkins University Press, Baltimore, pp 491–499Google Scholar
- 29.Howe HF, Westley LC (1986) Ecology of pollination and seed dispersal. In: Crawley MJ (ed) Plant ecology. Blackwell Scientific Publications, London, pp 185–215Google Scholar
- 37.Hodgkison R, Ayasse M, Kalko EKV, Häberlein C, Schulz S, Mustapha WAW, Zubaid A, Kunz TH (2007) Chemical ecology of fruit bat foraging behavior in relation to the fruit odors of two species of Paleotropical bat-dispersed figs (Ficus hispida and Ficus scortechinii). J Chem Ecol 33:2097–2110PubMedCrossRefPubMedCentralGoogle Scholar
- 61.Brown MJ (1997) Durio, a bibliographic review. Bioversity International, New DelhiGoogle Scholar
- 73.Borges RM (2015) Fruit and seed volatiles: multiple stage settings, actors and props in an evolutionary play. J Indian Inst Sci 95:93–104Google Scholar
- 96.Hayden S, Bekaert M, Goodbla A, Murphy WJ, Dávalos LM, Teeling EC (2014) A cluster of olfactory receptor genes linked to frugivory in bats. Mol Biol Evol 4:1–11Google Scholar
- 99.Kalko EKV, Condon M (1993) Bat-plant interactions: how frugivorous leaf-nosed bats find their food. Bat Res News 35:28Google Scholar
- 104.Chapman CA, Russo SE (2007) Linking behavioral ecology with forest community structure. In: Campbell CJ, Fuentes A, KC MK, Panger M, Bearder SK (eds) Primates in perspective. Oxford University Press, New York, pp 510–525Google Scholar
- 107.Nevo O (2016) The chemical ecology of primate seed dispersal. PhD thesis, Georg-August-Universität GöttingenGoogle Scholar
- 117.Melin AD, Nevo O, Shirasu M, Williamson R, Garrett E, Endo M, Sakurai K, Matsushita Y, Rothman J, Touhara K, Kawamura S. Accepted. Fruit scent and observer color vision shape food-selection strategies by wild capuchin monkeys. Nat CommGoogle Scholar
- 124.Valenta K, Kalbitzer U, Razafimandimby D, Omeja P, Ayasse M, Chapman CA, Nevo O (2018) The evolution of fruit colour: phylogeny, abiotic factors and the role of mutualists. Sci Rep. https://doi.org/10.1038/s41598-018-32604-x
- 128.Clark L, Hagelin J, Werner S (2014) The chemical senses in birds. In: Scanes CG (ed) Sturkie’s avian physiology, 6th edn. Academic Press, New York, pp 89–111Google Scholar
- 132.Laska M, Hernandez Salazar LT (2015) Olfaction in nonhuman primates. In: Doty RL (ed) Handbook of olfaction and gustation. Wiley, New York, pp 607–623Google Scholar
- 145.Penn HJ, Crist TO (2018) From dispersal to predation: a global synthesis of ant-seed interactions. Ecol Evol 210:291Google Scholar
- 149.Junker RR, Parachnowitsch AL (2015) Working towards a holistic view on flower traits-how floral scents mediate plant-animal interactions in concert with other floral characters. J Indian Inst Sci 95:43–67Google Scholar
- 152.Rodríguez A, San Andrés V, Cervera M, Redondo A, Alquézar B, Shimada T, Gadea J, Rodrigo MJ, Zacarías L, Palou L, López MM, Castañera P, Peña L (2011) Terpene down-regulation in orange reveals the role of fruit aromas in mediating interactions with insect herbivores and pathogens. Plant Physiol 156:793–802PubMedPubMedCentralCrossRefGoogle Scholar
- 154.Galetti M, Guevara R, Côrtes MC, Fadini R, Von Matter S, Leite AB, Labecca F, Ribeiro T, Carvalho CS, Collevatti RG, Pires MM, Guimarães PR, Brancalion PH, Ribeiro MC, Jordano P (2013) Functional extinction of birds drives rapid evolutionary changes in seed size. Science 340:1086–1090PubMedCrossRefPubMedCentralGoogle Scholar
- 156.Kleiner A (2018) Olfactory and visual signals of animal dispersed fruits in the temperate climate of South Germany. MSc thesis, University of UlmGoogle Scholar
- 157.Haynes KF, Millar JG (2012) Methods in chemical ecology volume 2: bioassay methods. Springer Science & Business Media, NorwellGoogle Scholar