Fruit aromas in mature fleshy fruits as signals of readiness for predation and seed dispersal

Abstract

Contents

	'Summary'	36
I.	'Introduction'	36
II.	'VOCs in plants'	37
III.	'Fruit VOCs and interactions with vertebrates'	40
IV.	'Fruit VOCs and interactions with insects'	41
V.	'Fruit VOCs and interactions with microbes'	41
VI.	'VOCs as mediators of indirect interactions'	42
VII.	'VOCs in fruit crops and agriculture'	43
VIII.	'Concluding remarks and future prospects'	44
	'Acknowledgements'	45
	References	45

Summary

The dispersal of seeds away from parent plants seems to be the underlying selective force in the evolution of fleshy fruits attractive to animals. Secondary metabolites, which are not essential compounds for plant survival, are involved in the interaction of fleshy fruits with seed dispersers and antagonists. Plant volatile organic compounds (VOCs) are secondary metabolites that play important roles in biotic interactions and in abiotic stress responses. They are usually accumulated at high levels in specific plant tissues and organs, such as fleshy fruits. The study of VOCs emitted during fruit development and after different biotic challenges may help to determine the interactions of fleshy fruits not only with legitimate vertebrate dispersers, but also with insects and microorganisms. A knowledge of fruit VOCs could be used in agriculture to generate attraction or repellency to pests and resistance to pathogens in fruits. This review provides an examination of specific fruit VOC blends as signals for either seed dispersal or predation through simple or complex trophic chains, which may also have consequences for an understanding of the importance of biodiversity in wild areas.

I. Introduction

‘… that a ripe strawberry or cherry is as pleasing to the eye as to the palate (…) will be admitted by every one. But this beauty serves merely as a guide to birds and beasts, in order that the fruit may be devoured and the matured seeds disseminated’

(Darwin, 1872).

In addition to the phylogenetic and physiological constraints influencing fruit traits (Eriksson & Ehrlén, 1998; Whitehead & Poveda, 2011), the dispersal of seeds away from parent plants seems to be an important selective force in the evolution of fleshy fruits attractive to vertebrates (Van der Pijl, 1969; Snow, 1971). There is still little empirical evidence on the primary function of secondary metabolites in fleshy fruits, but it is widely assumed that they are involved in the mediation of two main goals: the attraction of seed dispersal organisms and the avoidance of consumption by seed predators. It is thought that the primary function of these specialized metabolites in immature fruits is to defend them against all types of potential consumers (Cipollini & Levey, 1997; Mack, 2000). Other hypotheses, such as direct nutritional benefits, defense tradeoff, attraction/association, seed germination inhibition and influence on protein assimilation and gut retention time, have also been proposed (Cipollini, 2000). Changes in secondary metabolites occur during ripening, in combination with changes in size, texture, taste, aroma and color; however, their biological role and whether they have evolved under the selective pressures of frugivores are largely unknown.

Fruit traits are perceived by animal frugivores in a hierarchical manner. The aroma and color are probably the first cues for frugivore attraction at distance; once a frugivore contacts the fruit, it perceives morphological traits; finally, fruit chemistry determines taste and digestibility. Visual signals have been investigated extensively in recent times and special attention has been paid to the function of anthocyanins in the attraction of mutualists and/or deterrence of antagonists (Schaefer, 2011; Valido et al., 2011). However, the role of ripe fruit volatiles as olfactory signals directed to legitimate dispersers and predators has not been investigated in detail. Recently, it has been shown that the aroma and color in wild fig fruits (flower-bearing receptacles called syconia) in Papua New Guinea have evolved in concert and as predicted by differences in the behavior, physiology and morphology of their bird and bat dispersers, indicating that differences among vertebrate frugivores have shaped the evolution of fruit traits. This evidence provides experimental support, for the first time, for the existence of seed dispersal syndromes, at least for fruit aroma and color (Lomáscolo et al., 2010).

Plant volatile organic compounds (VOCs) comprise a wide diversity of low-molecular-weight secondary metabolites, with an appreciable vapor pressure under ambient conditions. Although some VOCs are probably common to almost all plants, others are specific to only one or a few related taxa. To the first type belong the so-called ‘green leaf’ volatiles (GLVs) because of their ‘fresh green’ odor. This group comprises short chain (C6) acyclic aldehydes, alcohols and their esters produced by plants from most taxa as a wound response via the enzymatic metabolism of polyunsaturated fatty acids. However, species- or genus-specific VOCs have been described in some species, such as the sulfur-containing VOCs of Alliaceae and Brassicaceae (Qualley & Dudareva, 2001). To understand the functional significance of VOCs in ripe fruits, it is necessary to know their biosynthesis and developmental regulation, their quantitative and qualitative accumulation and the responses triggered by VOCs on organisms interacting with the fruit, including vertebrates, insects and microorganisms. In this review, we attempt to update and integrate all relevant references pertaining to this issue to obtain a clearer picture of the VOC biosynthetic patterns in fleshy fruits and of the putative roles of VOCs in the attraction or deterrence of seed dispersers and/or predators.

II. VOCs in plants

It is assumed that VOCs were originally antimicrobial compounds that later also served to combat pests, thus providing plants with a kind of immune system (Turlings & Tumlinson, 1992). In vegetative tissues, VOC patterns have coevolved with phytophagous insects, and their chemical diversity has escalated, probably to gain improved defenses (Becerra & Venable, 1999). It has been proposed that the different VOCs could act synergistically, as in conifer resins, for simultaneous protection against pests and pathogens (Phillips & Croteau, 1999). Recent data have demonstrated that VOCs serve as signals for communication between plants and between distal parts within the same plant (Qualley & Dudareva, 2001). They are also involved in protecting the plant against abiotic stress, defending the plant against pests and pathogens, and attracting herbivore predators and pollinators (Gershenzon & Dudareva, 2007; Kessler et al., 2008). It is well documented through genetic engineering experiments that specific terpenoid compounds emitted by leaves can intoxicate, repel or deter herbivores (Aharoni et al., 2003), or they may attract natural predators and parasitoids of damaging herbivores, thus protecting plants from further damage (Kappers et al., 2005). It has also been demonstrated that specific volatile compounds emitted by flowers greatly contribute to the plant's reproductive success and survival in natural ecosystems (Kessler et al., 2008).

Our knowledge regarding VOC synthesis and accumulation in fruits is much less extensive than that related to flowers and leaves. There are few references that have considered specific VOCs or VOC blends in mature fleshy fruits for the attraction of legitimate disperser organisms (Lomáscolo et al., 2010) and no references considering VOCs in interactions with putative predators, probably as a result of the difficulties and complexities involved in the measurement and analysis of VOC contents and emission from fruits under different developmental and environmental conditions in ecological contexts. In contrast, the importance of the interaction of fruit VOCs with specific insects or microorganisms in agricultural contexts has been the subject of extensive research because of its economic impact (Bruce et al., 2005), although there are few works on the interactions of fruit VOCs with vertebrates in crops (Borges et al., 2011).

VOCs in fleshy fruits

In general, flowers and fruits release the widest variety of VOCs, with emission rates peaking before pollination and at ripening, respectively (Dudareva et al., 2004). In addition, flowers, leaves and fruits often show different VOC profiles, suggesting that their functions in different tissues or organs may also be different (Fig. 1; A. Rodríguez et al., unpublished). For example, mono- and sesquiterpenes are major compounds of mango leaves and fruits, although specific VOCs can be ascribed to each tissue, such as esters that are not detected in leaves (Lalel et al., 2003; Silva et al., 2012). Similarly, important scent VOCs in ripe peach (linalool and C10 lactones) are absent from leaves (Horvat & Chapman, 1990). Specific fruit and leaf VOCs have also been reported in citrus (Dugo & Di Giacomo, 2002) and in the wild Schinus molle (Maffei & Chialva, 1990). Based on principal component analysis, Oliveira et al. (2010) showed that the peel, pulp and leaves from different fig cultivars can be distinguished by their distinct abundance of monoterpenes, sesquiterpenes and aldehydes.

Figure 1. Terpene volatile profile of different citrus tissues and organs: flower, peel from immature fruit, peel from mature fruit and leaf. Relative amounts of sesquiterpene- and monoterpene-derived volatiles are presented as a percentage of each class with respect to the total.

VOCs in fruits are diverse, consisting of different chemical products comprising only 10⁻⁷–10⁻⁴ of the fresh fruit weight (Jiang & Song, 2010). Hundreds of VOCs are identified in most fruits, and this diversity is partially responsible for the unique scent found in different fruit species and cultivars. The aroma properties of fruits depend on the combination of VOCs produced and on the concentration and odor threshold of each in the blend. Most can be divided into four major classes according to their metabolic origin (Negre-Zakharov et al., 2009): terpenoids (e.g. mono- and sesquiterpenes and apocarotenoids), phenylpropanoids/benzenoids (e.g. eugenol, benzaldehyde), fatty acid derivatives (e.g. hexenal, hexenol) and amino acid derivatives (e.g. thiazole, 2- and 3-methylbutanal). Among them, terpenoids and lipid derivatives are probably the most abundant and expensive to produce in terms of energy and nutrients (Gershenzon & Dudareva, 2007). From a chemical view, these VOCs can be classified as esters, alcohols, aldehydes, ketones, lactones and terpenoids.

To attract seed disseminators and thus to ensure reproductive and evolutionary success, many plants release diverse blends of VOCs from their fruits. With their huge number of compounds and high structural diversity, terpenoids are one of the largest classes of VOCs in fruits, especially monoterpenes, sesquiterpenes and irregular terpenes of low molecular weight. In lulo, myrtle, coriander, mango and citrus mature fruits, monoterpenes are the most representative type of volatiles, their content varying between 50% to almost 100% of the total VOCs (Table 1). Esters are also key contributors to the fruity aroma. For example, the ester fraction has been described as the determinant for the characteristic varietal aroma in apple cultivars and Lambrusco and hybrid grapes (Jiang & Song, 2010; Yang et al., 2011). In Ficus racemosa, esters dominate volatile profiles (86–94% of total), as also occurs in Ficus benghalensis fruits during the night (Borges et al., 2011). Some other groups of minor volatiles are also important for fruit scent in terms of concentration, for example, apocarotenoids, also called norisoprenoids, derived from carotenoids by oxidative cleavage. Studies in tomato, melon, peach and watermelon indicate that the carotenoid profile has a clear impact on aroma via the determination of the suite of synthesized apocarotenoids (Lewinsohn et al., 2005; Rodrigo et al., 2012). Other compounds, such as sulfur volatiles, mainly arising as degradation products of cysteine, cystine, methionine, glutathione and some vitamins, are also characterized by their extremely low aroma thresholds (Du et al., 2011). VOCs derived from amino acids are important flavor constituents of many ripe fruits, such as strawberries, tomato, melon and apples (Goff & Klee, 2006; Gonda et al., 2010).

Table 1. Changes in the total amount and selected groups of volatile organic compounds during the development and ripening of different fleshy fruits

	Increase during ripening	Aldehydes		Esters		Monoterpenes		Sesquiterpenes		References
		Change	Total	Change	Total	Change	Total	Change	Total
^a Slight increase from the onset of ripening. ^b Specific to ripening and/or mature fruits. ·/·, Reduction in accumulation. x, Increase in accumulation. nd, non-detailed.
Solanum vestissimum	> ×30	a	1.4	> ×6.0	75.0	> ×0.5	1.2	nd	nd	Suárez & Duque (1992)
Schinus molle	·/· 1.5	nd	nd	nd	nd	×1.0	85.9	×1.0	10.8	Hosni et al. (2011)
Myrtus communis	×3.0	nd	nd	·/· 3.3	0.1	×0.9	71.6	·/· 4.5	3.06	Aidi Wannes et al. (2009)
Coriandrum sativum	×30.7	·/· 2.1	1.2	·/· 14.7	2.4	×1.6	90.4	·/· 1.9	2.0	Msaada et al. (2009)
Psidium guajava	×2.9	·/· 12.5	2.3	×8.8	54.0	b	10.0	b	33.5	Soares et al. (2007)
Psidium salutare	·/· 2.2	nd	nd	·/· 1.2	2.5	·/· 2.2	58.5	·/· 2.6	20.0	Pino & Queris (2008)
Mangifera indica	nd	nd	nd	×44.6-b	7.7–38.4	·/· 1.2–1.6	53.6–78.3	·/· 1.9	7.9–10.8	Lalel et al. (2003)
Fragaria ananassa	×4.0–19.0	·/· 60.3	0.5–4.2	×1.3	78.0–91.0	×3.3	nd	nd	nd	Menager et al. (2004); Azodanlou et al. (2004)
Actinidia deliciosa	×5.6	·/· 1.7–2.0	9.3–15.8	×330–409	71.6–82.8	×1.0–1.1	0.0–0.1	nd	nd	Garcia et al. (2012); Wang et al. (2011)
Actinidia chinensis	×60–117	·/· 1.1–×1.6	0.3–0.7	×1116–2381	76.0–83.0	×1.0–4.4	0.0–0.9	nd	nd	Wang et al. (2011)
Malus × domestica	×1–30	×1.7–18.3	3.3–54.3	×9–515	11.5–97.0	nd	nd	nd	nd	Villatoro et al. (2008); Ortiz et al. (2011)
Prunus persica	·/· 1.0–4.6	·/· 3.1–×1.2	0.3	nd	nd	·/· 1.7- b	0.3–40	nd	nd	Engel et al. (1988); Aubert et al. (2003)
Carica papaya	×3.3	nd	nd	×1.6	50.8–95.1	b	1.6	nd	nd	Almora et al. (2004); Fuggate et al. (2010)
Capsicum annuum	·/· 1.2–1.3	×1.3	13.8	·/· 1.7	52.8	×1.9	18.1	·/· 1.4	10.4	Forero et al. (2009)
Solanum lycopersicum	×1.9–1077.8	·/· 2.7-×5.5	13–82.0	nd	nd	×1.3–1.6	<0.1	nd	nd	Birtic et al. (2009); Ortiz-Serrano & Gil (2010)
Ficus scortechinii	×18.0	nd	nd	×3.1	61.1	nd	nd	·/· 17.3	0.1	Hodgkison et al. (2007)
Ficus hispida	×30	nd	nd	×2.6	46.4	nd	nd	·/· 6.9	0.1	Hodgkison et al. (2007)
Ficus benghalensis
Diurnal	nd	nd		nd	13.34	nd	12.16	nd	35.86	Borges et al. (2011)
Nocturnal	nd	nd	4.6	nd	46.9	nd	10.61	nd	6.67
Ficus racemosa	nd	nd	0.5–0.6	nd	85.6–93.2	nd	3.1–3.6	nd	nd	Borges et al. (2011)
Citrus sinensis	×4.2	·/· 48	0.48	b	0.01	×1.2	96	·/· 8	0.34	Rodríguez et al. (2011)

VOC changes during ripening

The following examples, without being an extensive review of the published literature, illustrate how VOC profiles change during fruit ripening. VOC production increases between 1 and > 1000 times during the maturation of most fruits (Table 1). Concomitantly, qualitative changes in the VOC profile take place during ripening (Table 1). For example, short-chained aldehydes, which provide the ‘fresh green’ odor, are abundant in numerous unripe green fruits, and their concentration decrease with ripening in fruits, such as nectarines, guavas, apples, coriander, strawberries and kiwis. In other fruits, such as neutral grapes, few volatiles other than C6 compounds accumulate (Yang et al., 2011). In this case, VOCs with a green flavor increase until the period of ripening and then decrease. The monoterpene profile also changes during the ripening of lulo, myrtle, coriander and citrus fruits (Table 1). For example, in oranges, there is a reduction in the linalool content as maturation progresses, whereas the limonene content increases from 30- to 100-fold between the green and color break stages (Dugo & Di Giacomo, 2002; Rodríguez et al., 2011). In white guava, mono- and sesquiterpenes, which are absent in unripe fruit tissues, accumulate and increase during ripening (Soares et al., 2007).

The concentration of esters and lactones, responsible for the spicy floral and fruity scent of many fruits, increases extraordinarily during ripening. In apple, mango, strawberry, kiwi, papaya, guava and lulo fruits, ester production increases by a factor ranging from 1.3 to > 2300 during maturation and, at the ripe stage, esters can account for up to 97% of the total VOCs (Table 1). It is interesting to remark that, in the case of strawberries, apples and lulos, a burst in ester production has been associated with the onset of ripening (Suárez & Duque, 1992; Menager et al., 2004; Villatoro et al., 2008; Table 1). In nectarines, lactones are characteristic of ripe fruits, their concentration increasing during maturation to reach up to 45.7% of the total VOCs (Engel et al., 1988).

The accumulation of low-strength ripeness-specific fruity aromas, such as apocarotenoids, sulfur volatiles and furan-related compounds, also increases during fruit ripening. In most carotenogenic fruits, the pigment profile changes during maturation, thus changing the apocarotenoid profile as well. In tomato, apocarotenoid VOCs are not emitted until relatively late in fruit ripening, and, during this process, the amount of apocarotenoids increases by a factor of 40 (Mathieu et al., 2009). In peaches, the apocarotenoid level also increases during fruit ripening, representing c. 40–60% of the total VOCs at the ripe stage (Aubert et al., 2003; Brandi et al., 2011). In strawberries, most sulfur volatiles increase by as much as 100% with maturity (Du et al., 2011). Methyl sulfanyl compounds increase considerably in kiwi with ripening, dimethyl sulfide being one of the key components that differentiates the aroma of yellow and green cultivars (Garcia et al., 2012). Finally, furanone-derived compounds also increase in concentration with the maturation of some fruits. In peaches, furan-related VOC accumulation starts just before color change, and its concentration reaches a maximum at the ripe stage (Brandi et al., 2011). In strawberries, furan-derived compounds are not detected until the fruit reaches a red color, and their content increases by c. 100-fold during maturation (Menager et al., 2004).

In summary, the influence of the ripening stage on fruit scent is clearly evident, and it is well documented that VOC composition changes both quantitatively and qualitatively during maturation. Indeed, the analysis of principal components has been successfully applied to discriminate between ripening stages depending on the presence/absence of some VOCs in many fruits, such as apples (Villatoro et al., 2008), grapes (Yang et al., 2011), mangos (Lebrun et al., 2008), strawberries (Azodanlou et al., 2004), figs (Hodgkison et al., 2007) peaches and nectarines (Lavilla et al., 2002). Some of these VOC modulations have been related to aroma chemical changes associated with ripening. For example, sugars, the concentration of which increases with ripening, are precursors of furanones, and, in tomatoes, a direct relationship has been established between sucrose and VOC production (Zanor et al., 2009). Fatty acids are quantitatively the major precursors responsible for the synthesis of esters, aldehydes, alcohols and acids found in fleshy fruits. Because lipid biosynthesis and membrane fluidity increase during ripening, a wider assortment of lipid-derived precursors of aroma-contributing VOCs is found in the tissues of fully ripe fruits (Sanz et al., 1997).

In addition, recent molecular findings support the idea that the de novo synthesis of VOCs is induced at ripening. Transcriptional regulation has been described for terpene-, carotenoid-, fatty acid- and phenylpropanoid-derived VOCs, and, in most cases, gene expression is induced on ripening concomitantly with the production of important flavor compounds (Rodrigo et al., 2012). Moreover, some genes have been shown to display a fruit-specific expression, such as those involved in different steps of alcohol and ester biosynthesis in melon (Yahyaoui et al., 2002; Manríquez et al., 2006). The expression of genes involved in the biosynthesis of amino acid-derived VOCs is also much higher in ripe fruits than in vegetative and unripe fruits (Gonda et al., 2010). Similarly, sesquiterpene synthase activity is evident in the rind from ripe melon, whereas it is not found in the flesh and unripe rind, where no sesquiterpenes accumulate (Portnoy et al., 2008). Moreover, sesquiterpene synthase genes are found to be transcriptionally regulated during fruit development and are likely to be associated with the VOC differences responsible for the unique aroma of different melon varieties (Portnoy et al., 2008). It is common in VOCs that a single enzyme catalyzes the synthesis of multiple products from different substrates (Pichersky & Gang, 2000). Therefore, it has been proposed that this broad substrate specificity is the result of convergent evolution, in which new enzymes with the same function have evolved independently in separate plant lineages from a shared pool of related enzymes with similar, but not identical, functions, providing an extraordinary versatility to VOC blend production patterns in specific plant tissues (Pichersky & Gang, 2000).

III. Fruit VOCs and interactions with vertebrates

Vertebrate seed dispersers of fleshy fruits are primarily birds and mammals, although fish and some reptiles have also been described as minor seed dispersal agents (Fleming & Kress, 2011). It is generally assumed that 140 million yr ago, when angiosperms probably originated, seeds were small and had very few dispersal attributes, indicating that dispersal was probably unassisted (Tiffney, 2004). Around the Tertiary (65 million yr ago), plant and fruit sizes became larger, strongly affecting the evolution of biotic dispersal via the production of fleshy fruits. The radiation of mammals and birds in the early Tertiary that mediated more efficient dispersal of larger seeds probably began at this stage (Fleming & Kress, 2011).

It is widely assumed that birds use primarily visual stimuli for the detection of fleshy fruits because the smell sense is less developed in avian dispersers (Schaefer, 2011). Obviously, this is not the case for nocturnal birds, which have well-developed olfactory bulbs (Corlett, 2011). Moreover, recent works have shown that at least some birds are able to detect VOCs (Mardon et al., 2010), and they use VOCs as cues to detect insect-infested trees (Mäntylä et al., 2008) or to recognize the fleshy fruits of figs (Borges et al., 2008, 2011). The VOC profiles in ripe fruits of different fig species are quite variable, and different VOC profiles have been observed in bat-dispersed vs bird-dispersed figs (Borges et al., 2008). Interestingly, in the case of F. benghalensis fruits, night VOCs, when seeds are dispersed by bats, are dominated by esters, whereas diurnal VOCs, when figs are consumed by birds, have a greater representation of terpenes (Borges et al., 2011). Without underestimating color cues, this evidence suggests that VOCs emitted by fig fruits are olfactory cues for either birds or bats.

Bat attraction by fruit VOCs has been investigated in further detail. Different bat species are able to distinguish by smell and clearly prefer ripe over unripe fruits of different plant genera (Luft et al., 2003; Schlumpberger et al., 2006; Hodgkison et al., 2007). Other bat species are attracted by artificial fruits impregnated with essential oils of Piper gaudichaudianum and Ficus insipida (Bianconi et al., 2007). Among the diverse aromatic profiles of mature fruits from Ficus species (Table 1), bats use smell, as well as color, as one of the components of the bat syndrome dispersion (Lomáscolo et al., 2010). This evidence has been used to estimate the occurrence of bat vs bird dispersal syndromes in 42 co-occurring Ficus species in Papua New Guinea. The results indicated that odor (and color) in figs have evolved as predicted by the selective pressures of their frugivores (Lomáscolo et al., 2010). In this study, VOC production was considered quantitatively and not qualitatively and, on average, the number and total peak area of VOCs were lower in bird-dispersed than in bat-dispersed figs, although the differences between these values were not statistically significant. It would be worth testing whether specific VOCs or VOC mixtures are actually preferred by either bird or bat dispersers.

Most frugivorous mammals rely on olfactory stimuli to detect ripe fruits. Dominy et al. (2001) proposed that early primates that were insectivorous and nocturnal were also frugivores, eating dull-colored and smelly fruits. Although most primates later acquired trichromatic vision, which permitted them to become more efficient in selecting ripe fruits, some primates still show extremely acute sensitivity to odors associated with fleshy fruits (Laska et al., 2006). Dichromatic white-faced capuchins rely more strongly on olfaction than do trichromatic individuals to detect fig fruits (Melin et al., 2009). Night monkeys (Aotus) could detect banana fruits by smell alone in laboratory trials, but diurnal monkeys could not (Bicca-Marques & Garber, 2004). Therefore, it is possible that odor cues remain important in primates (especially in nocturnal primates) to detect fleshy fruits, but it is reasonable to believe that other senses, such as sight or touch, are used almost simultaneously to decide whether to eat a fruit (Dominy et al., 2001). Mammals other than primates with an extraordinary sense of smell, including rodents, also use VOCs to recognize fleshy fruits (Corlett, 2011).

IV. Fruit VOCs and interactions with insects

Seed consumption by herbivorous invertebrates, mainly insects, dates back to the Devonian (c. 416 million yr ago). However, those insects were probably granivorous and contributed little to the evolution of fleshy fruits (Mack, 2000; Fleming & Kress, 2011). Frugivorous insects comprise mainly taxa from the orders Lepidoptera, Hemiptera, Coleoptera, Hymenoptera and Diptera (Sallabanks & Courtney, 1992). Fruit location is a key issue for feeding, mating and reproduction of specialist insects, and involves the perception of a sequence of olfactory and visual cues (Schoonhoven et al., 2005). Generally, specialized insects are able to distinguish the VOCs emitted by vegetative tissues and unripe and ripe fruits; they are mainly attracted by particular VOC blends of ripe fleshy fruits and, in some cases, they are repelled by green tissues (Vallat & Dorn, 2005; Piñero & Dorn, 2009). For example, the codling moth Cydia pomonella (Lepidoptera: Tortricidae) is attracted by mature apple fruits, but repelled by green fruits, probably through the emission of benzaldehyde and butyl acetate (Vallat & Dorn, 2005). The preference for mature fruits has also been shown for females of the oriental fruit moth (Cydia molesta; Lepidoptera: Tortricidae) in apple and peach fruits, whereas VOCs released by vegetative tissue are behaviorally ineffective (Piñero & Dorn, 2009). For Ceratitis capitata (Diptera: Tephritidae) females, the odor of ripe or almost ripe coffee drupes is more attractive than that of unripe drupes, leaves or stems (Prokopy & Vargas, 1996).

Host fruit recognition usually depends on specific blends of VOCs and not just on the detection of a single compound; however, some blend components are biologically more important than others for the interaction (Light et al., 2001; Reddy & Guerrero, 2004). Moreover, the recognition of a host plant by insects could occur using either specific ratios of ubiquitous compounds or species-specific compounds (Bruce et al., 2005). For example, polyphagous insects, such as Anastrepha obliqua and C. capitata fruit flies (Diptera: Tephritidae), are attracted by different blends of monoterpene compounds emitted by mango and citrus fruits (Papadopoulos et al., 2006; Malo et al., 2012). In contrast, monophagous insects, such as the olive fly Bactrocera oleae (Diptera: Tephritidae), are attracted by a specific VOC blend present in ripening fruits and in leaves. Therefore, these specific VOC cues may have evolutionary significance for monophagous insects. In this context, the case of the monophagous apple maggot fly Rhagoletis pomonella (Diptera: Tephritidae) is particularly interesting, because this insect shifted from its ancestral host hawthorn (Crataegus spp.) to cultivated apple c. 150 yr ago, and it has been shown that apple and hawthorn native flies use fruit VOCs to distinguish between the two hosts (Linn et al., 2003). Genetic analysis of F2 and backcross hybrid insects indicates that differences in host choices based on VOC discrimination pertain to a few loci, imply cytonuclear gene interactions and have resulted in reproductive isolation, which has facilitated sympatric insect speciation in the absence of geographic isolation (Dambroski et al., 2005).

In addition, insects are sensitive to volatiles for social communication, and some acquire host plant compounds to use as sex pheromones or sex pheromone precursors (Bruce et al., 2005). Insects such as Tephritidae and Drosophilae Diptera release sex pheromones in response to host fruit chemical emissions that additionally enhance the response of insects to sex pheromones. For example, the combination of male pheromone and host fruit odor is more attractive to female papaya fruit flies, Toxotrypana curvicauda (Diptera: Tephritidae), than is either male pheromone or host fruit aroma alone (Landolt et al., 1992). Oriental fruit fly Bactrocera dorsalis (Diptera: Tephritidae) males are attracted to and feed on methyl eugenol, a VOC emitted by Terminalia catappa ripe fruits (Siderhurst & Jang, 2006). Males that have eaten methyl eugenol are more successful in courting and mating with females than males that have not (Shelly & Dewire, 1994).

V. Fruit VOCs and interactions with microbes

Microbes are the most abundant frugivores of fleshy fruits, although they have been rarely studied as such in ecological contexts (Levey, 2004). Fungi that naturally infect a wide range of wild fruits include, predominantly, Colletotrichum, Phomopsis, Cladosporium, Penicillium and Fusarium species (Tang et al., 2003, 2005). Coincidently, these species cause the most conspicuous opportunistic diseases in commercial fleshy fruits. Fruit softening during ripening facilitates the establishment of opportunistic microbial infections. Ripening is a developmental process usually associated with increased susceptibility to microbial infections in crops (Prusky, 1996) and in wild plants (Tang et al., 2003, 2005).

The effect of VOCs emitted by different plant organs and tissues on microorganisms, either as volatiles or through direct contact, has been widely investigated in crops and forest trees. Many studies indicate that VOCs are toxic to diverse fungi, yeasts and bacteria; however, these studies were performed with individual compounds in vitro and, sometimes, the assays used levels far in excess of those actually present in a fruit (Dorman & Deans, 2000; Daferera et al., 2003). For fleshy fruits of agricultural importance, it has been proposed that VOCs could be used as inhibitors of postharvest fruit spoilage (Archbold et al., 1997). However, when compounds able to control fungal or bacterial growth in vitro were tested in fruits, they were inefficient or even stimulated microbial growth. For example, (E)-2-hexenal vapor at different doses inhibited Botrytis cinerea spore germination and mycelial growth in Petri dishes, but the same doses of this compound applied on strawberry fruits enhanced fungus incidence during storage (Fallik et al., 1998), indicating that VOC toxicity experiments should be performed in planta. In orange fruits, the d-limonene content is usually low in the exocarp during the 2–3 months postanthesis, increases dramatically when the green fruit develops seeds, and remains at a high level until the fruit becomes fully mature (Dugo & Di Giacomo, 2002). When oranges were engineered to accumulate very reduced levels of this monoterpene, they became resistant to the bacterium Xanthomonas citri subsp. citri, and to Penicillium digitatum and other specialized fungi (Rodríguez et al., 2011). Therefore, d-limonene is required for pathogens to establish infections in mature oranges.

VI. VOCs as mediators of indirect interactions

Recent works have shown that the consideration of the third (and fourth) partners within the community context converts a previously considered parasitism into a multispecies mutualism (Dunn et al., 2008; Palmer et al., 2010). In the fleshy fruit–frugivore context, the ecological importance of such interactions is difficult to predict because, in many cases, it is not unequivocally known whether consumers are actually seed dispersal or predator agents under different environmental circumstances. Pulp feeder ‘antagonists’ may have a positive effect on seed and seedling fate (Fedriani et al., 2012), seed predators may facilitate seed dispersal (Norconk et al., 1998), legitimate seed dispersal animals may have negative effects on plant population dynamics (Loayza & Knight, 2010) and the interference between mammal seed dispersers and insect seed predators may ultimately benefit seed dispersal (Visser et al., 2011).

Fruit VOCs and interactions among insects

In spite of the high cost for plants to produce VOCs in ripe fruits and the importance of ripe fruit aromas for the life cycle of specialized insect frugivores, as mentioned previously, insects are considered to be harmful to plant fitness and are much more involved in fruit and seed predation than in seed dispersal (Janzen, 1977). Fruit VOCs might be involved in indirect defense against insect consumers by attracting pest parasites. For example, Leptopilina boulardi (Hymenoptera: Figitidae), a parasite of Drosophila melanogaster (Diptera: Drosophiladae), is attracted to VOCs emitted by fly-infested banana or pear fruits, but not to noninfested ones (Couty et al., 1999). VOCs emitted by coconut fruits infested by Aceria guerreronis (Acari: Eriophyidae) are more attractive for two mite predators (Neoseiulus baraki and Proctolaelaps bickleyi) than uninfected ones (Melo et al., 2011). Although indirect defenses are widely documented in plant vegetative tissues (Heil, 2008), there are no reported cases of indirect defenses against pathogens in ripe fruits.

Fruit VOCs and interactions with insects and vertebrates

VOCs may be involved in how vertebrates distinguish between infested and uninfested fruits. In general, avian consumers prefer intact mature fruits and reject fleshy fruits infested by insects (Traveset et al., 1995; García et al., 1999). Deterrent effects of infested fruits on avian frugivores are considered to be an evolutionary necessity for insect frugivores to escape predation (Sallabanks & Courtney, 1992). In another scenario, legitimate seed dispersers may be attracted by VOCs from infested fruits and consume them without major problems (Drew, 1987; Valburg, 1992). In a recent work, the attraction of birds to heavily insect-infested trees was directly correlated with the emission of several specific terpene VOCs by the trees (Mäntylä et al., 2008). In these cases, insects would directly benefit birds by enhancing the nutrient content of the fruits and indirectly benefit host plants by facilitating vertebrate seed dispersal. In addition, a recent review has shown several examples of insects inhabiting seeds from wild ripe fruits that can survive passage through the entire digestive tract of seed-dispersing vertebrates, including many bird species and also primates, which suggests that this process may also favor insect dispersal (Hernández, 2011). As described above, most mammals may primarily use the sense of smell instead of sight to locate fruits. There are references to ungulates, primates and rodents being attracted by fruits infested by insects (Redford et al., 1984; Rader & Krockenberger, 2007; Bravo, 2008), which suggests that they may also be able to distinguish VOCs of infested fruits. However, the results for any of the intervening elements of these tritrophic interactions are often unpredictable and could conversely lead to killed larvae, destroyed seeds or toxicity for the vertebrate (Or & Ward, 2003).

Fruit VOCs and interactions with microorganisms and insects

Ripe fruit VOCs are also important in trophic interactions involving microbes and insects. Insects that feed on overripe, wounded or decomposing fruits commonly exploit VOCs induced by microbial action on damaged tissues for host finding (Hammons et al., 2009). The microbial detoxification of pulp secondary metabolites and the breakdown of carbohydrates refractory to insect digestive enzymes, on the one hand, and microbial dissemination on the other, may explain such mutualisms (Berenbaum, 1988). The Japanese beetle Popillia japonica (Scarabaecidae: Rutelinae) facilitates feeding of the green June beetle Cotinis nitida (Scarabaecidae: Cetoniinae) on grapes by biting through the skin and introducing yeasts in such wounds. Yeasts eliciting fermentation VOCs are exploited by both sexes of C. nitida for host finding (Hammons et al., 2009). Nitidulid sap beetles (Carpophylus humeralis; Coleoptera: Nitidulidae) are attracted to VOCs from fermenting fruits and vegetables (Nout & Bartelt, 1998). There are other insects that prefer damaged fruits, such as the Asian lady beetle Harmonia axyridis (Coleoptera: Coccinellidae) (Koch et al., 2004) and the medfly C. capitata (Papadopoulos et al., 2006). Fruit flies, such as Bactrocera tryoni and B. oleae, have symbiotic bacterial associations, which can improve the nutritive quality of their fruit diet and may play a role in the detoxification of plant secondary chemicals (Fletcher, 1987). Recently, it has been demonstrated that specific odors from rotten fruits sexually attract male fruit flies (D. melanogaster) (Grosjean et al., 2011). D. melanogaster larvae consume yeasts growing on rotting fruits and have evolved resistance to fermentation products. Ethanol is produced in overripe and rotten fruits through sugar fermentation by infecting microorganisms. Interestingly, it has been shown that alcohol protects D. melanogaster from endoparasitoid wasps; therefore, flies consuming alcohol do not need to activate the stereotypical antiwasp immune response. Therefore, fly larvae seek for ethanol-containing food and probably use it as an antiwasp medicine (Milan et al., 2012).

Many butterflies in tropical forests feed on fruits that have fallen to the ground. This substrate differs in many ways from floral nectar, and it has been established that fruit-feeding butterflies use specific VOC cues from the fruits and fermentation products to locate their food (Molleman et al., 2005; Sourakov et al., 2012). From the plant's perspective, the presence of microbes and insects in damaged fruits for predation may favor the possibility of undamaged fruits to attract legitimate seed dispersers. Alternatively, VOC compounds emitted by wounded fruits may play an indirect role in plant defense by facilitating attraction of natural enemies of the damaging fungus and/or insect.

Fruit VOCs and interactions with microorganisms and vertebrates

Little is known about whether VOCs emitted by ripe fleshy fruits infected by microbes are distinguished by vertebrates. Primates, rodents and bats have demonstrated sensitivity to ripe fruit-associated odors, such as those of esters, aldehydes and alcohols (Laska et al., 2006; Sánchez et al., 2008). Thus, it is possible that these frugivores are able to recognize rotten fruits through the VOCs emitted by the fruit, the microbe or both. The only volatiles from rotten fruits that have been studied in some detail for their interaction with vertebrates are alcohols, specifically ethanol. Dudley (2000) proposed that ethanol could represent an important sensory cue to primates because of its association with caloric and physiological rewards. Moreover, Dominy (2004) suggested that the ethanol content (together with soft texture) could have been a cue with strong adaptive advantages for primates, and the selection of fruits on this basis may be a long-standing trend in primate evolution. However, Levey (2004) concluded that frugivores usually prefer ripe, nonrotting fruits over damaged or rotting fruits (in which the concentration of ethanol is supposed to be higher). In Egyptian fruit bats, ethanol did not stimulate visits to or the ingestion of ripe fruits (Sánchez et al., 2008). Studies performed with wild individuals of several frugivorous and nectarivorous bat species have shown that these animals tolerate relatively high levels of ethanol without negative effects on their flight and echolocation performance (Orbach et al., 2010). These authors believe that frugivorous bats may be used to eating fruits rich in ethanol when other healthy fruits are unavailable.

Most birds and small mammals prefer ripe, uninfected fruits to rotten fruits (Borowicz, 1988; Cipollini & Stiles, 1993), except for some specialized rodents (Borowicz, 1988). The omnivorous diet of such rodents may be an adaptation for enhanced tolerance to microbes in rotten fruits and for efficient competition with most vertebrates for these resources. Nevertheless, rotten fruits are generally nontoxic to vertebrates. When just rotten fruits are offered or ripe fruits are scarce, these are readily consumed (Borowicz, 1988; Cipollini & Stiles, 1993; Levey, 2004; Sánchez et al., 2008). Therefore, microbes and vertebrates may not be strong competitors, especially when ripe fruit resources are limited. Microbes may benefit by being ingested by frugivores and dispersed in their feces (Abranches et al., 1998). From the plant's perspective, there could be two different scenarios. Deterrence to microbes may be important for the fruit if infected seeds may compromise their viability (Janzen, 1977), particularly in the case of small fruits from shrubs and small trees or in the case of seeds without coats. Neutral or attraction responses may be favored by ripe fruits when seed dispersal is not compromised, mainly in the case of large fruits with large pericarps and/or coated seeds. Microbial infection and/or insect infestation would favor fruit crushing and/or abscission, and therefore access of terrestrial animals to the fruit.

VII. VOCs in fruit crops and agriculture

Most VOC research has been conducted in agricultural species; it has contributed greatly to our understanding of the role of VOCs in plant–insect and plant–microbe interactions, and it is providing many applied tools in agriculture. For example, the pear ester ethyl (2E, 4Z)-2,4-decadienoate is highly attractive and has been used to monitor both males and females of the codling moth C. pomonella (Light et al., 2001). Field-trapping tests show that Argyresthia conjugella (Lepidoptera: Yponomeutidae) females are attracted to VOCs identified from rowanberries, and that a blend of 2-phenylethanol and anethole is sufficient to show a strong attraction (Bengtsson et al., 2006). In addition, the identification of VOCs specifically emitted from infested fruits and attractive to natural enemies would allow the development of lures to be used in integrated pest control programs. Alternatively, based on the synergism between insect pheromones and VOCs, it has been suggested that mating disruption dispensers could be developed for certain pests using small amounts of expensive pheromonal ingredients and small amounts of inexpensive plant VOCs (Reddy & Guerrero, 2004). In addition, pheromone-based mass annihilation strategies are nowadays successfully employed to control Diptera and Coleoptera insects in agriculture (Witzgall et al., 2010). These strategies, unlike detection and monitoring (where only a small proportion of a population needs to be sampled), require the use of the most attractive lure and may become far more efficacious if lures include fruit VOCs involved in ovipositional and/or feeding cues. Blends of VOCs emitted by nonhosts are usually neutral, but they could also be repellent, although this aspect has been largely overlooked (Reddy & Guerrero, 2004). For example, the psyllid Diaphorina citri (Hemiptera: Psyllidae), transmission vector of the bacterium that causes the Huanglongbing (HLB) disease of citrus, is attracted to VOCs emitted by citrus host plants (Patt & Sétamou, 2010), whereas VOCs from the nonhost guava have been shown to be repellent and also to inhibit the psyllid response to the normally attractive citrus odor (Rouseff et al., 2008; Onagbola et al., 2011). The identification of repellent VOC blends from guava leaves and fruits would allow the development of strategies to control the psyllid population and thus HLB spreading. Finally, the growing number of reports on the involvement of specific VOCs in plant defense, together with the current progress on the knowledge of their biosynthesis and regulation, is allowing the use of plant genetic engineering to improve plant resistance to pests and diseases. For example, d-limonene production, which represents up to 97% of total VOCs in orange fruit peel, has been down-regulated by the overexpression of an antisense construct of a d-limonene synthase gene (Rodríguez et al., 2011). Transgenic orange fruit peels with up to 85 times reduced d-limonene accumulation were less attractive to males of the citrus pest medfly (C. capitata) and strongly resistant to fungal and bacterial pathogens (Rodríguez et al., 2011). This work illustrates how fruit VOC emissions can be manipulated to provide novel strategies for pest and disease management, without altering important agronomic traits. Our most recent results indicate that d-limonene up-regulation is highly associated with a general depletion of defenses in mature fruit peels (A. Rodríguez et al., unpublished), suggesting that there is a tradeoff between the costly production of monoterpenes for the attraction of frugivores and decreased general defense.

The foundations for the aromas associated with most fruits existed long before crop domestication (Goff & Klee, 2006). However, in recent times, breeding with the aim of improving yield, size and postharvest fruit shelf life has affected some original sensory qualities, including aromas. For example, in a commercial tomato cultivar, modifications in some VOC concentrations have been detected in comparison with a wild relative (Goff & Klee, 2006). In strawberry, marked differences in the production of specific sesqui- and monoterpenes between cultivated and wild strawberry species have been related to the activity of just one enzyme (Aharoni et al., 2004). The sesquiterpene profile also varies greatly in the rind of melons resulting from breeding programs (Portnoy et al., 2008).

Some fruit tree crops, such as Malus (apples), Pyrus (pears) and Prunus (peaches, nectarines, plums, etc.) species, have been subjected to extensive breeding programs, but, in general, fruit trees have very long juvenile periods that have delayed the possibilities of producing new varieties through breeding, at least in comparison with annual crops. Consequently, varieties from Citrus (including oranges, lemons, mandarins, limes, grapefruits, etc.) and most (although not all) varieties from other tropical and subtropical fruit trees, including some with highly odorous fruits (mango, guava, avocado, durian, passionfruit, breadfruit, pitanga, mangosteen, loquat, quince, etc.), are species, natural hybrids or budsport mutants selected in nature by humans in more or less recent times. Within a given genus, VOC profiles may be similar, at least qualitatively, in ancestral types (maintained in germplasm banks) and in relatively recent cultivated hybrids (e.g. citrus types; Table 2), or may be variable among close species with drastic changes in some specific major compounds (e.g. Psidium species; Table 1). Such similarities or differences in VOC profiles may have obvious consequences in fleshy fruit interactions with their biotic environment.

Table 2. Changes in the total amount of volatile organic compounds in different ancestral, wild and natural hybrid Citrus types

		Monoterpenes		Sesquiterpenes
		Total	Limonene	Total
Obtained from Dugo & Di Giacomo (2002). Tr, traces.
Ancestral species	Citrus grandis	50.3–100.0	48.9–95.6	Tr-4.5
	Citrus medica	53.4–100.0	51.2–93.6	Tr-8.3
	Citrus reticulata	90.6–100.0	87.4–91.7	Tr
Wild species from subgenus Papeda	Citrus hystrix	55.3–100.0	2.8–14.2	Tr-3.1
Hybrids	Citrus aurantium	82.9–100.0	80.1–95.8	Tr
	Citrus paradisi	86.31–100	83.4–93.8	Tr-5.1
	Citrus aurantifolia	69.7–94.9	38.4–50.0	5.36–12.87
	Citrus clementina	85.1–100.0	83.0–95.1	Tr-2.2
	Citrus bergamia	31.5–100	24.1–54.9	Tr-2.2
	Citrus limon	71.6–100	59.6–76.2	Tr-3.0
	Citrus junos	70.7–100.0	60.4–82.4	Tr-4.7
	Citrus unshiu	42.8–100.0	41.2–90.7	Tr-2.7
	Citrus sinensis	93.4–100.0	91.0–97.0	Tr-1.1

VIII. Concluding remarks and future prospects

As illustrated in this review, fruit VOC profiles are diverse, change during ripening and have important effects on both mutualists and antagonists. Most research on this topic has been conducted so far in agricultural species, and information regarding wild fruits is scarce or totally absent when pertaining to the role of VOCs in interactions with vertebrates, insects or microbes in nature. Therefore, it would be indispensable to obtain additional information on wild fruit VOCs and on their interaction with frugivores in order to assess their ecological relevance and to draw strong evolutionary inferences on how aromas may have evolved by selective pressure from surrounding living organisms. Moreover, little information exists on how VOC changes might affect species interaction. Wild and domesticated species and cultivars with different VOC profiles are excellent tools to investigate the importance of VOCs for fruit interactions with their frugivores. Moreover, the possibility of generating mutants and transgenic plants affected in VOC biosynthetic or signal transduction pathways could allow the determination of the key compounds involved in fruit–frugivore interactions.

Information on fruit VOC evolution and the influence of frugivores in this process is also scarce. Selective pressures on fruit VOC production and emission may be exerted not only by legitimate seed dispersal animals, but also modulated and/or abolished by less apparent, but often more common, frugivore agents. Whether the attraction of seed dispersal and fruit predator agents through fleshy ripe fruit-emitted VOCs is positive or detrimental for plant fitness, and therefore the net effect of these trophic interactions with multiple partners, should be carefully considered and investigated. Seed predation could be a selective force on fruit VOC emission in some cases, as has been suggested for fruit color polymorphisms in Acacia ligulata (Whitney & Stanton, 2004). Therefore, to understand whether and to what extent the diversification of VOCs in fleshy fruits has been shaped by frugivores, we must overcome the traditionally considered dichotomy of seed dispersers vs seed predators and investigate the interactions among the multiple partners of the network as a whole. In addition, a broad view of VOCs is necessary, together with other traits in each specific fruit species as integrated cues for frugivory, because it is unlikely that such different cues have evolved independently. Furthermore, it is necessary to deepen our understanding of how VOCs are perceived by frugivorous animals and to what extent the responses to odors are learned or innate, which may have important consequences when considering the presumed coevolution of fleshy fruits and frugivores. As envisaged from this review, the role of VOCs in species interaction may be quite complex because of the multitrophic, direct and indirect interactions, synergistic effects of compounds, etc.; thus, integrative studies are necessary to take into account the full fitness costs and benefits of particular traits.

To unravel how differences and singularities between fruit VOCs of different species, cultivars and mutants or transgenic plants can be explained from a molecular and biochemical perspective, and how they are linked to different direct and indirect trophic interactions, will require multidisciplinary collaborative work from chemists, geneticists, ecologists and biologists in the coming years. Comparative transcriptomic, proteomic and metabolomic datasets in both fruits and vertebrates/insects/microbes would provide novel valuable data for the clarification of these highly complex and interactive processes. As not all studies accomplished to date have been reproducible beyond the laboratory setting, it is of major importance to establish the ecological and/or agricultural realism of new experiments. In agricultural contexts, studies on fruit VOCs may help to develop potential alternatives to toxic synthetic agrochemicals for the control of devastating pests and diseases. In conclusion, future work will improve our basic understanding of plant ecology and evolution and may have important applications in agriculture.

Acknowledgements

We wish to acknowledge Dr Pedro Moreno for his comments on this review. This work was supported by grant AGL2009-08052, co-financed by Fondo Europeo de Desarrollo Regional (FEDER) and the Ministry of Economy and Competitivity of Spain. A.R. was supported by a PhD fellowship from the Instituto Valenciano de Investigaciones Agrarias (IVIA, Spain). We apologize to the authors whose work is not cited because of space constraints.