FLIES AND FLOWERS II: FLORAL ATTRACTANTS AND REWARDS

This paper comprises Part II of a review of flower visitation and pollination by Diptera (myiophily or myophily). While Part I examined taxonomic diversity of anthophilous flies, here we consider the rewards and attractants used by flowers to procure visits by flies, and their importance in the lives of flies. Food rewards such as pollen and nectar are the primary reasons for flower visits, but there is also a diversity of non-nutritive rewards such as brood sites, shelter, and places of congregation. Floral attractants are the visual and chemical cues used by Diptera to locate flowers and the rewards that they offer, and we show how they act to increase the probability of floral visitation. Lastly, we discuss the various ways in which flowers manipulate the behaviour of flies, deceiving them to visit flowers that do not provide the advertised reward, and how some flies illegitimately remove floral rewards without causing pollination. Our review demonstrates that myiophily is a syndrome corresponding to elements of anatomical, behavioural and physiological adaptations of flower-visiting Diptera. The bewildering diversity of anthophilous Diptera and of the floral attractants and rewards to which they respond allows for only broad generalizations on myiophily and points to the need for more investigation. Ecological relationships between flies and flowers are critical to the survival of each group in many habitats. We require greater understanding of the significance of flies in pollination, especially in the face of recent pollinator declines.


INTRODUCTION
Pollination by insects, including flies, is commonly a mutualistic interaction, in which both the plant and the insect benefit. The plant receives self or outcross pollen on receptive stigmas and/or exports pollen to a conspecific plant's stigmas, and the insect gains a nutritional or other reward. The most often sought visitor rewards are nectar, which fuels flight and may contain a variety of compounds in addition to sugars, and to a lesser extent pollen, an important protein resource, but there are also various non-nutritive rewards. Flowers are able to signal their presence at both long and short distances, and to guide the movements of visitors, using a variety of attractants. Over evolutionary time, animal and flower interactions have produced plants with visual, chemical, and other traits that entice pollinators to visit conspecific flowers. Conversely, animal visitors, such as flies, have behavioural, physiological, and morphological traits that facilitate their interactions with flowers. And, as with any mutualistic interaction, there are plenty of opportunities for deception and cheating on both sides.
Despite the frequency with which Diptera are recorded as flower visitors, relatively little is known about fly pollination (myophily or myiophily) compared to other groups of pollinators such as bees, birds, and bats. We do know that numerous families of Diptera include flower visitors, with examples of the most common including Mycetophilidae, Bibionidae, and Culicidae (Nematocera); Syrphidae, Bombyliidae, Conopidae, Stratiomyidae, and Nemestrinidae (lower Brachycera); and among the higher Brachycera (Cyclorrhapha), the Muscidae, Anthomyiidae, Tachinidae, Calliphoridae, and others (Kastinger & Weber 2001;Larson et al. 2001;Rotheray & Gilbert 2011). Floral visitation by small Diptera, including acalyptrates, has often been discounted as unimportant in pollination, and in many cases empirical confirmation of successful pollination is lacking (Larson et al. 2001;Willmer 2011). However, Diptera can be highly important pollinators in some systems (e.g. Ssymank et al. 2008;Barraclough & Slotow 2010), and may also be important in the reproduction of certain threatened and endangered plants (e.g., Wiesenborn 2003;Murugan et al. 2006;Humeau et al. 2011). We have previously reviewed the importance of Diptera as pollinators (Larson et al. 2001), and in this paper we turn to reviewing what is known about the attractants and rewards that entice Diptera to visit flowers. We have separated floral rewards ("primary attractants" of Faegri & van der Pijl (1979)) from advertisements ("secondary attractants" of Faegri & van der Pijl (1979) or simply "attractants" of many anthecologists), and discuss them in that order. Rewards are the main attractants for flower visiting insects and include pollen and nectar, whereas advertisements are visual, chemical, or structural cues that provide information to potential pollinators about location of and access to floral rewards. The final section of the paper will discuss the numerous types and examples of cheating by both sides in the flyflower relationship, in which flies obtain rewards without J Poll Ecol 12 (8) effecting pollination and flowers may attract visitors by advertising rewards that do not exist.

REWARDS FOR FORAGING FLIES
The Diptera are among the most ancient pollinators of flowering plants (Gottsberger 1974;Crepet 1979;Bernhardt & Thien 1987;Kato & Inoue 1994;Labandeira 1998), primarily visiting flowers to procure highly nutritious nectar or pollen (Faegri & van der Pijl 1979;Simpson & Neff 1981;Larson et al. 2001). As the onset of diversification in Diptera predates angiosperm diversification (Grimaldi 1999), it is likely that adaptation to flower visitation evolved independently in different lines of Diptera. Pollen grains (or microspores) are essential to terrestrial plant reproduction and probably were an original reward and attractant for spore-vectoring and pollinating insects Kevan & Baker 1983;Willemstein 1987). Nevertheless, pollination droplets of sugary liquids, such as those from the sexual structures of various ancient plants, may have been consumed by insects prior to the evolution of the Angiospermae (Taylor 1981;Wetschnig & Depisch 1999;Ren et al. 2009). Diptera and Hymenoptera, with mouthparts appropriate to imbibing liquids, are known from the fossil record during this period, and may have sought those sugary liquids (Labandeira 1997;Krenn et al. 2005). Regardless of the reproductive importance of nectar in ancient plants, production of floral nectar throughout the angiosperms is indicative of its value and it is now the foremost reward for many pollinators (Kevan & Baker 1983, 1999, including a diverse array of Diptera (Larson et al. 2001). In addition to pollen and nectar rewards, flowers attract Diptera of both sexes and so may be mating sites, or they may function as sites for predatory flies to await prey. Flowers may also give protection from predators and inclement weather, or provide brood places for larvae. The warmth generated or concentrated by some flowers may act as a secondary attractant, but may be more important for its physiological effects on anthophiles, so we consider it a reward.

Primary floral rewards -Nectar
Nectar is the most commonly sought reward by flowervisiting flies (Tab. 1). It is predominately a source of carbohydrates, most commonly containing the disaccharide sucrose and the hexose sugars glucose and fructose. Nectar can also contain other sugars, various amino acids, proteins, lipids and vitamins (Baker & Baker 1973a, b, 1983a. Nectar constituents and concentrations are often adapted to the desired group of pollinators. Nectars with a preponderance of sucrose, for example, are taken mostly by long-tongued Diptera (e.g., Goldblatt et al. 1997). The nectar used by most flies (short-tongued or lapping) is characteristically hexose-rich and of relatively high (but variable) sugar concentration (Percival 1965;Baker & Baker 1983b;Kevan & Baker 1999). The sugars in nectar may crystallize, but many generalist flies are able to re-liquify the nectar with saliva and then imbibe it (Elton 1966;Baker & Baker 1983a;Willmer 2011), as is also reported for Diptera feeding on honeydew (Downes & Dahlem 1987).
Both sexes of most flies use the carbohydrates in nectar for short-term energy needs (Hocking 1953;Downes 1955;Downes & Smith 1969), especially during periods of peak activity such as swarming, mating and oviposition, dispersal, and migration (Kevan 1973;Haslett 1989a, Service 1997Branquart & Hemptinne 2000;Willmer 2011). In female mosquitoes, at least, nectar feeding and blood feeding are antagonistic and mutually exclusive, but little is known about how such insects choose which resource to pursue (Foster 1995). Vrzal et al. (2010) found that female mosquitoes can acquire appreciable quantities of amino acids from nectar, increasing longevity and possibly decreasing the need for a blood meal. Rather than consuming nectar, David et al. (2011) reported Drosophila suma scarifying the floral tissues of Ipomoea and Crinum with spiny fore tarsi adapted to the purpose and consuming the forthcoming liquid.
It is surprising that there is little information on the importance of micro-nutrients in the physiology of Diptera; even economically important Diptera have not been rigorously studied. Numerous authors have indicated that nectars can contain amino acids (Baker & Baker 1973a, b, 1983a, b, 1986Kevan & Baker 1983, 1999Potter & Bertin 1988;Rathman et al. 1990;Gardener & Gillman 2002;Vrzal et al. 2010), but experimental evidence for an adaptive role of amino acids in the attraction and nutrition of pollinators is sparse. Nectar availability can increase insects' life spans (Nayar & Sauerman 1971;Grimstad & DeFoliart 1974;Kevan & Baker 1983;Vrzal et al. 2010), presumably due to the presence of minor nutrients (Haslett 1989a). Stomoxys calcitrans (L.) (Muscidae) fed dilute honey lived longer than those fed sugar syrup (Jones et al. 1992). Rathman et al. (1990) found that Sarcophaga bullata Parker (Sarcophagidae) chose an amino acid-containing nectar over a sucrose-only nectar only when protein deprived and previously without access to protein or amino acid sources. This suggests a nutritional role for the amino acids in nectar, especially for reproductive females and when other protein rich foods are scarce (Vrzal et al. 2010). Foraging flies that have nectar as their primary source of protein-building amino acids feed from flowers with higher concentrations of amino acids than do those that have alternative protein sources (e.g., pollen) (Baker & Baker 1983a. The possible benefits to Diptera of other nectar constituents, such as lipids, antioxidants and proteins, are also not well understood. Lipids are common in the nectar of flowers visited by flies, which presumably have digestive lipases or esterases to break them down (Dadd 1973;Chapman 1998;Kevan & Baker 1999). Dethier's classic book (1976) makes little mention of lipids in the diets of flies. Lipids likely provide energy as polyunsaturated fatty acids, or have a role in hormonal biosynthesis as sterols (Downer 1978). Reports on Diptera feeding on lipid-rich floral tissues invoke special floral structures (Dafni & Werker 1982;Aronne et al. 1993). Antioxidants such as organic "reducing" acids (ascorbic acid) are often found in lipidic nectars, where they may help to delay rancidity (Baker & Baker 1983a). Nectar may also contain small quantities of vitamins and minerals (Baker & Baker 1983a) and these too may play roles in dipteran nutrition. Various salts have phagostimulative effects and may be available as a reward in Gladiolus spp. Goldblatt et al. 1997Goldblatt et al. , 2001 Melasphaerula ramosa (Burm. f (8) the nectar of some flowers (Waller et al. 1972;Dethier 1976). However, high concentrations of salt in solution are rejected by Eristalis tenax (L.), possibly because they inhibit the water detection cells of the labial taste sensilla (Wacht et al. 2000).
Pollen contains proteins, carbohydrates and lipids, in addition to various amino acids, minerals and vitamins (Stanley & Linskens 1974;Alba et al. 1995;Bonvehi & Jorda 1997;Roulston & Cane 2000;Villanueva et al. 2001). To obtain the nutrients within the indigestible exine, some Diptera consume entire grains, whereas others puncture them and suck out the protoplasm. In the former case, the nutrients presumably diffuse out through germination pores in the exine during digestion or the exine is destroyed by osmotic shock (Haslett 1983;Kevan & Baker 1983;Roulston & Cane 2000). Some syrphids soak pollen in saliva exuded by their proboscis, then imbibe the slurry (Proctor & Yeo 1973). They may ingest so much pollen that their abdomens appear bloated and yellow, and digested pollen is observed easily in their feces. The mouthparts of many Bombyliidae are poorly suited to pollen collection directly from anthers, but a few species can harvest pollen using specialized structures on the tarsi that rake the anthers and convey material to the mouth (Neff et al. 2003).
Pollen contains numerous amino acids that may have other nutritional benefits (Kevan & Baker 1983, 1999Roulston & Cane 2000), though proline is the most significant of them; plants usually need it in large amounts (e.g., 2% of dry weight) to support pollen tube growth (Linskens & Schrauwen 1969). The proline in the pollen of anemophilous flowers (Stanley & Linskens 1974) can be used by insects to power their flight (Gilbert 1985a;Candy et al. 1997). The importance of proline in fly nutrition was confirmed by Wacht et al. (2000) who found that it is the only amino acid of 20 they tested that can be perceived in the sub-millimolar range by the salt receptor of the labellar chemosensory hairs of Eristalis tenax. In the Syrphidae, adults obtain protein solely from pollen and nectar (Oldroyd 1964), with the relative intake determined by their immediate needs (Haslett 1989a) and the constraints of proboscis structure (Gilbert 1981(Gilbert , 1985b. In general, as the length of the proboscis increases, the apparent importance of pollen in the diet decreases as the flies concentrate on flowers with longer corolla tubes and greater nectar rewards (Gilbert 1981).
Some syrphids, such as members of the genera Melanostoma and Platycheirus in Europe, feed almost exclusively on the pollen of various anemophilous plants that offer no nectar reward (Goot & Grabandt 1970;Leereveld et al. 1976Leereveld et al. , 1991Gilbert 1981;Leereveld 1982Leereveld , 1984Sharma et al. 1993;Ssymank & Gilbert 1993). The pollen of anemophilous plants tends to be smoother and has less of a pollenkitt (sticky or oily material on the outside of pollen grains) than that of entomophilous plants (Ackerman 2000, Pacini & Hesse 2005, so syrphids may not contribute significantly to pollination. However, intraspecific variation in pollen characteristics of some plants corresponds to the environments in which the plants grow, suggesting that syrphids may sometimes be significant pollinators of anemophilous plants. For example, pollen from Plantago lanceolata L. (Plantaginaceae) growing in forested areas in Luxembourg, where wind is minimal and biotic pollen vectors are most adaptive, has a higher adhesive capacity than pollen from conspecifics in exposed coastal Netherlands (Stelleman 1980(Stelleman , 1984.

Secondary floral rewards -Location of mates and prey
Because the females of many species of flies must visit flowers to obtain nectar and pollen, flowers could be excellent places for males to locate mates (Tab. 2). Maier (1978Maier ( , 1982, Waldbauer (1979) andWaldbauer (1984) have shown that males of Temnostoma and Mallota species, Somula decora Macquart, and Spilomyia hamifera Loew (Syrphidae) patrol flowers in the morning, and oviposition sites in the afternoon, and so locate mates (Waldbauer & Ghent 1984). In some cases, they establish a patrol route territory or "trap-line", containing ten to twenty prime flowers, which they over-fly repeatedly (Maier & Waldbauer 1979). They feed periodically at various flowers along the route and thereby may act as pollinators. According to Speight (1978), various syrphids commonly hold floral territories where they await females. Certain bombyliid flies also use flowers as mating rendezvous sites, possibly defending territories for this purpose (Evenhuis 1983, Johnson & Dafni 1998). As a final example, Nagasaki (2007) reported numerous individuals of Notiphila maritima Krivosheina (Ephydridae) mating and ovipositing on flowers of Nuphar subintegerrima (Casp.) Makino and carrying pollen. The larvae were not found to feed in the flowers, but presumably in submerged sediments near the water lily roots, which they may have tapped for oxygen as described by Larson & Foote (1997) for Notiphila associated with the yellow water lily, Nuphar luteum.
Pollen movement may occur as individual flies visit multiple flowers during multiple mating or oviposition events, and may occur in concert with the breeding system of the plant. For instance, in the pollination of protogynous Peltandra virginica Kunth (Araceae) by Elachiptera formosa Loew (Chloropidae), the flies oviposit mainly in femalephase inflorescences (differentiated by floral odour), and their larvae feed on the copious pollen released during the male phase. Pollen transfer is frequent and occurs as the ovipositing flies move between female-phase inflorescences, where they mate and oviposit, and male-phase inflorescences, where they consume nutritional pollen (Patt et al. 1995).
Similar to the mutualisms between Yucca (Agavaceae), Ficus (Moraceae), Elaeis (Arecaceae) and their non-Diptera pollinators, larval flies may be provisioned with resources, such as pollen or seeds, in exchange for pollination service. Mostly, insects that use flowers as brood places are destructive, but the relationship will persist if there is a net benefit to the plant in terms of propagules produced (Pettersson 1992;Brody 1992;Zimmerman & Brody 1998;Brody & Morita 2000;Despres et al. 2002;Gao 2011). However, larvae may consume floral resources to the extent that the reproductive success of the plant may be lower due to reduced rewards to pollinators (du Toit 1987;Tribe 1991;Nicolson 1994a, b;Weiss 1996).

Secondary floral rewards -Warmth, protection and shelter
Although some bees are known to sleep in flowers (Dafni et al. 1981;Willis & Kevan 1995), similar behaviours have been reported rarely among other flower-visiting insects (Tab. 3). In the Arctic, Hocking & Sharplin (1965) and Hocking (1968) noted Aedes spp. (Culicidae) basking at the foci of the parabolic corollas of heliotropic Dryas integrifolia and Papaver radicatum Rottb., where the temperature was often 6̊ C above that of the ambient air. Basking flies were observed disproportionately often on flowers aligned with the sun, so heliotropism and the sun-focussing shape of the flowers provide heat rewards that increase the probability of pollinator visitation (Kevan 1989;Stanton & Galen 1989;Kudo 1995;Krannitz 1996;Totland 1996;Luzar & Gottsberger 2001;Yuan et al. 2008). Heliotropism is recorded in numerous plant families, most notably in the Asteraceae, Papaveraceae, Ranunculaceae and Rosaceae (Kevan 1972a;Stanton & Galen 1989;Totland 1996;Luzar & Gottsberger 2001;Orueta 2002), and seems relatively common in arctic and alpine regions where heat rewards could be sought by foraging Diptera. In two species of tropical Convolvulaceae (Ipomoea pes-capri and Merremia borneensis), Patiño et al. (2002) noted that seasonal heliotropism and floral orientation combine to regulate internal temperatures for both seed fertilization and development, and that insects preferentially visit the sunlit flowers. If the insects are reluctant to leave the heat, basking behaviour may not be conducive to cross-pollination. Extra warmth may benefit the plant by optimizing temperature for pollen germination and fertilization (Orueta 2002). Kevan (1970) has suggested that Diptera that bask in the diaheliotropic and parabolic corollas of arctic flowers such as Dryas integrifolia M. Vahl. (Rosaceae) are able, from such vantage points, to detect the shadows of potential predators. Some Diptera bask in flowers and the increased warmth may speed metabolism and ovarian maturation, and preheat flight muscles (Knutson 1974(Knutson , 1979Kevan 1975Kevan , 1989Yafuso 1993). Tatler et al. (2000) have shown that extra warmth results in accelerated optical neuronal processing in insects (e.g., Calliphora vicina, Calliphoridae). Certain parabolic, campanulate, and funnel-shaped flowers may be used by flies for physical protection from adverse weather conditions (Drabble & Drabble 1927;Parmenter 1958;Kevan 1973Kevan , 2007. The capacity for heat production, particularly in basal angiosperm lineages and in Araceae (Lamarck 1777;Thien et al. 2009), may have several functions, including acceleration of floral growth and development, improved volatilization of pollinator attracting molecules, and shelter and warmth for the entrapped pollinators (Meeuse 1966;Knutson 1974Knutson , 1979Moodie 1976;Thien et al. 2000;Patiño et al. 2002;Quilichini et al. 2010;Willmer 2011). Luo et al. (2010) demonstrated that heat generated by floral tissues was not necessary for seed or fruit development in Illicium spp. (Schisandraceae), but that larvae of cecidomyiid pollinators developing in the flowers would die without the added warmth.

FLORAL ADVERTISEMENTS TO FORAGING FLIES
The visual and olfactory advertisements used by flies to locate flowers operate over different distances. Distinctive floral advertisements and guides benefit the plant because they increase the likelihood that the pollinator can learn their "search image" and thus more easily revisit other plants of the same species. This constancy greatly improves the effectiveness of visiting insects as pollen vectors (Waser 1983(Waser , 1986Fenster et al. 2004), but has not been well studied in Diptera. Goulson & Wright (1998), for example, noted that the hover flies Episyrphus balteatus (De Geer) and Syrphus ribesii (L.) (Syrphidae) can be remarkably constant to the flowers they visit, although reasons for this constancy and the floral features involved are undocumented. Other visual and chemical cues operate at closer range, guiding the flies' behaviour on the flower. The whole suite of floral advertisements probably operates to attract anthophiles through both innate and learned responses.

Background on fly vision
As in other diurnal pollinators, most flower-visiting flies use floral colour as their primary cue to recognize preferred flowers (Tab. 4). Due to interactions between different visual cues in the orientation process, it is difficult to separate the elements of the suite of cues presented by flowers, and much of what is known about insect vision and the implications of floral colours for anthecology comes from studies of bees (e.g. Menzel & Shmida 1993;Kevan & Backhaus 1998;Chittka & Menzel 1992;Vorobyev & Brandt 1997;Kelber et al. 2003). Weiss (2001) has pointed out that vision and learning in various lesser understood groups of pollinators, including Diptera, is a fertile ground for research. It has been shown in behavioural experiments that Lucilia cuprina (Calliphoridae), Bombylius fuliginosus (Bombyliidae) and Eristalis tenax (Syrphidae) possess colour vision (Ilse 1949;Kugler 1950;Lunau & Wacht 1994;Kelber et al. 2003 for review), and thus discriminate colour shades due to their chromaticity and independent of their brightness.
The insect eye is less able to resolve shape and details of pattern than is the human eye (Land 1997, Dafni et al. 1997), but insects have remarkable abilities to perceive and resolve rapid motion. Because insects are fast flyers and lack the ability to move the eyes, their orientation is supported by effective motion resolution of more than 200 pictures per second (Ruck 1961). Thus, flies can perceive the forms of flowers even when in rapid flight (Kevan & Baker 1983;Dafni et al. 1997;O'Carroll et al. 1996). It is noteworthy that the motion-sensitive neurons in the eyes of hovering insects (e.g., Bombyliidae) are tuned to detection of lower rates of motion than are non-hovering insects (O'Carroll et al. 1996). The Diptera preference toward characteristically "open" flowers is probably not related to visual advertisements, but caused by their limited abilities to manipulate flowers. J Poll Ecol 12(8) Eristalis tenax (L.) Ilse 1949;Kugler 1950;Haslett 1989b;Lunau & Wacht 1994;Lunau & Maier 1995;Lunau 1996 Rhingia campestris Meigen Haslett 1989b Volucella pellucens (L.) In contrast to other insects, flies have two morphologically and physiologically separated visual subsystems. The apposition subsystem consists of two receptor cell tandems, only one of which is present in each ommatidium. Providing the eye with a set of four retinula cells that differ in their spectral sensitivities enables tetrachromatic colour vision, but with less sensitivity and less acuity than the neural superposition subsystem (Hardie 1979;Pichaud et al. 1999). The neuronal superposition subsystem is monovariant and thus colourblind but highly sensitive, collecting visual input from six retinula cells of six neighbouring ommatidia (Anderson & Laughlin 2000). This system enables flies to orient visually in dim light conditions without loss of spatial information, unlike optical superposition eyes of nocturnal Lepidoptera. Together, the two subsystems allow for a parallel processing of motion perception and colour vision (Kelber et al. 2003).

Visual advertisements -Colour
Most flies have tetrachromatic colour vision with sensitivity from ultraviolet (UV) over blue and green to yellow wavelengths (Menzel & Backhaus 1991;Troje 1993;Kelber et al. 2003). According to the colour vision model of Troje (1993), flies exhibit categorical colour vision such that they discriminate colours in only four categories. These categories are determined by the superior excitation of one of the two photoreceptor cells in each tandem and can be regarded as fly-blue, UV, fly-green and fly purple (Arnold et al. 2009).
Many Diptera are observed either on yellow or white flowers, although some species prefer blue or red (Weems 1953;Sandholm & Price 1962;Kevan & Baker 1983;de Buck 1990;Campbell et al. 2010;Willmer 2011). These responses are plastic to some degree. It is also unknown whether the flower visits were determined by an underlying innate colour preference, by a limited variety of alternatively coloured flowers, or by experience of individual flies. Interestingly, the categorical colour vision model of Troje (1993) predicts that flies have a limited ability to discriminate between yellow and white flowers, if both absorb ultraviolet light. In wild radish (Raphanus raphanistrum L. (Brassicaceae)), the yellow flower morphs are preferentially visited by flies and have a UV-reflective target pattern that the white morph lacks (Kay 1976(Kay , 1978. Innate colour preferences of naive flies have not yet been studied systematically; the hoverflies Eristalis tenax and Episyrphus balteatus exhibit preferences for yellow colours (Lunau & Maier 1995).
Some studies provide information about colour preferences in distinct species (Ssymank 2001, Laubertie et al. 2006. The notion that in regions where Diptera predominate as floral visitors, flowers are more often white and yellow than elsewhere (Inouye & Pyke 1988 for Australian alpine; Kevan 1972bKevan , 1973 for Canadian Arctic; Kevan (unpubl. data) for Rocky Mountain alpine; Primack (1978Primack ( , 1983 for New Zealand) has been questioned by Arnold et al. (2009), who found no significant change of flower colour with altitude in an alpine region of Norway.
Evidence from behavioural experiments shows that flies can be trained to colours that initially were less attractive to them. Kelber (2001) trained syrphid flies, which were initially attracted to yellow flowers, to visit flowers of other colours. After a period of training on blue flowers, the flies had no preference for yellow flowers when given a choice of blue or yellow. Haslett (1989b) studied the pollen foraging of six species of Syrphidae in the wild and at experimental arrays of artificial flowers of different colours, and showed that these species had colour preferences ranging from white and yellow to blue and violet. The preference for yellow colours in Eristalis tenax is innate, wavelength-specific and varies with proximity to the flower (Lunau & Wacht 1994;Lunau & Maier 1995;Lunau 1996). Once the flies land, however, only deep yellow colours with strong absorption in the ultraviolet waveband release proboscis extension behaviour in inexperienced flies. This innate response is finetuned to the spectral reflection properties of pollen that reflects green and yellow wavelengths > 510 nm. Ultraviolet and blue wavelengths < 510 nm, which are typically absorbed by yellow pollen, strongly inhibit the proboscidial reaction (Lunau 2000(Lunau , 2007. Wacht et al. (1996) noted that both optical and chemical stimuli control pollen feeding in Eristalis tenax and other syrphid flies. The fixed innate search images for pollen signals constrains the evolution of flowers' signalling devices and leads to the standardisation of floral signalling components, for example the colour signals of pollen, anthers, and pollen-and anther-mimicking floral guides (Lunau 2004(Lunau , 2007. Red flowers are normally associated with specialist pollinators, such as birds and those insects (butterflies, some beetles) that have red-sensitive vision (Kevan & Backhaus 1998;Bernhardt 2000). In this context it has been shown that red bird-pollinated flowers are less attractive to potentionally nectar thieving bees and thus provide a private niche for their pollinators (Lunau et al. 2011). Most flies are unable to perceive red, but flowers with reddish hues are also found within the mimetic complex of sapromyophilous blooms and this may relate to the sensitivity of the eyes of some Diptera to red (Autrum & Stumpf 1953). The colours of sapromyophilous blossoms tend to be dark and mimic the substrate associated with their odours (carrion, dung, etc.), but little is known of the linkage between the visual systems of the pollinators and orientation and attraction cues from the plants involved. There are assemblages of flies that are frequent visitors to blue-violet, pink or red flowers, but these often have tubular corollas and hidden nectar, thus only flies with elongate, specialized mouthparts can forage from them (Knuth 1906(Knuth -1909Robertson 1928;de Buck 1990;Proctor et al. 1996;Goldblatt et al. 2001;Kastinger & Weber 2001;Goldblatt & Manning 2007a, b;Willmer 2011). In numerous South African flowers visited predominately by long-tongued flies, a high incidence of scentless red, pink, or blue-violet flowers, which would usually suggest pollination by birds, has been recorded (Rebelo & Siegfried 1985;Goldblatt et al. 2001;Goldblatt & Manning 2007a, b). In addition to Bombyliidae, Nemestrinidae and Tabanidae, long-tongued Syrphidae, Conopidae, Empididae and Tachinidae are able to feed from these deep-tubed flowers, as well as from shallow, unspecialized ones. Observations by numerous investigators have shown the predilection of bombyliids for bluish flowers, which are often pollinated by bees (Knuth 1906(Knuth -1909Knoll 1921;Scott 1953;Kevan & Baker 1999;Kastinger & Weber 2001).
Despite the generalizations above, floral colour preferences vary both spatially and temporally, depending on the availability of flowers, floral rewards, competitors, and those factors in combination. Fruit flies react differently to the colours and scents of natural and artificial (trap) fruits depending on age and sex (Owens & Prokopy 1986;Duan & Prokopy 1994). Physiologically-induced changes (Browne 1993) and sexual differences in flies' reactions to flowers may be especially important in some of the mutualistic relations. In some flowers, corolla colour changes over time, and the colour phase of the rewarding flowers is often more attractive to naive visitors than the colour of spent flowers (Kugler 1950;Cameron & Troilo 1982;Gori 1983;Casper & La Pine 1984;Cruzan et al. 1988;Zietsman 1990;Robertson & Lloyd 1993;Weiss 1991;Lunau 1996).

Visual advertisements -Size and form
There is a positive correlation between the size of flowers, inflorescences or patches of bloom, and attractiveness to insects, and small, inconspicuous flowers are often aggregated into large inflorescences to attract pollinators more effectively (Faegri & van der Pijl 1979;Kevan & Baker 1983;Dafni et al. 1997). However, this relationship has been investigated in detail for only a few flyvisited flowers. Andersson (1991) showed that ray and inflorescence size of Achillea ptarmica L. (Asteraceae) have additive effects on the attraction of Syrphidae (Eristalis, Syrphus, Volucella). Abbott & Irwin (1988) found that ligulate inflorescences were visited much more frequently by Syrphidae than were the smaller non-ligulate ones in polymorphic populations of Senecio vulgaris L. (Asteraceae). The visitation rate by the tachinid Protohystricia huttoni (Malloch) to Myosotis colensoi (Kirk) Macbride (Boraginaceae) in New Zealand increased linearly with display size, but because the number of flowers visited on a plant concomitantly decreased, visitation rate to individual flowers remained more or less constant, regardless of display size (Robertson 1992;Robertson & Macnair 1995). The number of insect visits per unit time to clones of Edelweiss (Leontopodium alpinum Cass. (Astercaeae)) remained constant across all clone sizes, decreasing the chance of a given flower being visited in larger clones, but enhancing out-crossing (Erhardt 1993). Golding et al. (1999) noted that visits by Episyrphus balteatus (Degeer) (Syrphidae) to variously manipulated flowers of Brassica rapae oleifera (Brassicaceae) were more strongly related to the number of anthers present than with the size of corolla display. Moller (2000) indicated that floral symmetry is perceived by pollinators, and that symmetrical (radial or bilateral) flowers are preferentially visited and pollinated over flowers that deviate from symmetry. He suggests that deviations represent developmental instability and reflect reduced Darwinian fitness, and thus may be correlated with minor floral rewards. Nevertheless, asymmetries in flowers are more common than are generally recognized (Dafni & Kevan 1996) and can result from physical corolla damage or natural variation even in populations of plants bearing flowers that are described as symmetrical. However, the nature of the optical boundary between a flower and its background (i.e., edge effect) may also be important in attracting foraging flies. Consideration of edge effects also apply to flowers with broken outlines and appendages that may further stimulate the eyes of foraging flies (Faegri & van J Poll Ecol 12(8) der Pijl 1979), but the data are equivocal. Kugler (1950) detected no innate preference of Eristalis for star-shaped over circular models of flowers of the same size, although they were able to discriminate between them. Others have found preference for flowers with dissected outlines over those with simple outlines (e.g., Kugler 1956;Johnson & Dafni 1998). Despite the findings noted above, the generally held view that flowers with broken outline lengths are more visible to foraging anthophiles requires further critical examination and empirical testing.

Visual advertisements -Guides, staminodes and motion
Many flowers have visual guides that direct the foragers to hidden rewards, and ensure contact with the stigma, stamens or both (Sprengel 1793;Dafni & Giurfa 1999;Lunau 2006) (Tab. 5). These guides also affect the foraging efficiency of the visitor and have a positive effect on pollinator attraction and successful interaction with the flowers (Kevan & Baker 1983;Dafni et al. 1997;Hansen et al. 2012). Johnson & Dafni (1998), for example, showed that landing and orientation of Usia bicolor Macquart (Bombyliidae) to the centres of models were encouraged by converging lines. Dinkel & Lunau (2001) used circular dummy flowers to show that inexperienced Eristalis tenax were quicker to find a potential food source in the centre if black line markings were present than if not. In contrast to the yellow, UVabsorbing central dot used as a potential food source, the black line markings were never touched with the proboscis.
A row of yellow dots instead of black lines increased the handling time of Eristalis tenax flies, because the yellow dots are probed with the proboscis. On natural flowers the probing of a number of floral guides may intensify the movements of the flies and thus increase the chance of touching anthers and stigma. Vertically-oriented staminodia are often important structures that act as landing platforms, and may act as a physical guide to nectaries at the base of the flower (Brew 1987;Pellmyr 1992) or themselves exude nectar (Anderson & Hill 2002), and their reflectance patterns may play a role in attraction of Diptera (Kugler 1951;Percival 1965;Young et al. 1987a). Sparkling or glistening surfaces may attract some flies (Kugler 1951;Percival 1965;Patt et al. 1989;Bänziger 1996). Nectar guides often offer contrast in the UV (Kugler 1963;Lunau 2000;, to which insects are sensitive (Bishop 1974;Kevan & Backhaus 1998). Some flowers take advantage of the aggregation instinct of insects (Elvers 1980;Morse 1981) and trick foragers into believing that there are already insects feeding at the flower. Dark spots or florets may give the illusion of small, crawling insects being present and attract certain floral visitors (Dodson 1962;Weismann 1962;Faegri & van der Pijl 1979;Eisikowitch 1980;McDonald & van der Walt 1992;Westmoreland & Muntan 1996;Johnson & Midgley 1997;Johnson & Dafni 1998;Lamborn & Ollerton 2000).
However, the dark central florets of the wild carrot Daucus carota L. have been shown to repel egg-laying females of the  (Polte & Reinhold 2013). Physical vibration of flowers or flower parts as they move in the wind may also serve as an attractant, and some flowers accentuate their movement with motile appendages (Tab. 5) and may attract flower-visiting Diptera (Percival 1965;van der Pijl & Dodson 1966;Kevan 1970;Kevan & Baker 1983;Young 1984;Young et al. 1987a).

Chemical advertisements
Flies perceive chemical stimuli in a way that is unique among insects. In contrast to bees and other insects, many flies have no flagelliform antennae with which to touch floral surfaces. Flies have taste sensilla on the labellum, the tarsi of the forelegs and the ovipositor. Fly taste sensilla have four receptor cells: a water receptor, a salt receptor, a sugar receptor and an anion receptor. Taste sensilla have an additional mechanoreceptor providing those flies with the combined gustatory and tactile information of potential food touched with the tarsi or with the extended proboscis (Hansen 1978;Hanson 1987). Using an electrophysiological tip-recording technique for single labellar chemosensory hairs of Eristalis tenax, Wacht et al. (2000) showed that pollen extract and the amino acid proline stimulate the salt receptor, but in contrast to salt stimuli, do not inhibit the water receptor at the same time. Thus, the across-fibres pattern of excitation of taste receptors is responsible for the perception of phagostimulants in pollen.
Odours are often liberated from flowers, either to attract pollinators from a distance or to direct them to certain portions of the flower (Dobson 1994;Dobson & Bergström 2000;Knudsen et al. 2006;Willmer 2011). However, Diptera are usually attracted to flowers by a combination of visual and olfactory stimuli (e.g., Beaman et al. 1988). Generally it is thought that long-range attraction is visual (but see Giurfa et al. 1996;Kevan & Backhaus 1998) and that odour acts nearer the flower (Kugler 1951(Kugler , 1970, at least in diurnally active flowers. However, Liebermann (1925) showed that numerous species of muscoid flies approached covered plants of Cicuta (Apiaceae) only when they passed downwind of them. Others have observed the attraction of nocturnal mosquitoes to flowers or floral extracts, and their downwind approach in the absence of visual stimuli suggests airborne attractants (Brantjes & Leemans 1976;. Although difficult to see in the field, zig-zag flights from flower to flower are indicative of olfactory orientation, whereas direct flights, regardless of wind direction, signify visual orientation. At close range, odour can act similarly to visual nectar guides, by directing the visitor to the resources, and position them appropriately to contact sexual structures and effect pollination (Brantjes 1981;Young 1984Young , 1985Young et al. 1984Young et al. , 1987aPatt et al. 1989). Strength of attractiveness of the odour is apparently proportional to hunger (i.e. time since previous feeding; Jhumur et al. 2006).
Flower-visiting Diptera can be attracted by odours that are either pleasant or unpleasant to human beings (Tab. 6), but the latter are often accented in the literature in connection with deceitful attraction and the syndrome of sapromyiophily. There is a preference among many Diptera for "unpleasant" odours (Galen & Kevan 1980;Galen 1983Galen , 1985Galen & Newport 1988;Bänziger 1996;Pombal & Morellato 2000;Goldblatt et al. 2009). Nonetheless, numerous fly-visited flowers have a sweet, musty, or yeasty scent that is devoid of amine components (e.g., indole and scatole) associated with unpleasant smells (Erhardt 1993;Meve & Liede 1994;Johnson & Steiner 1995;Patt et al. 1995;Pombal & Morellato 2000;Goodrich et al. 2006;Hansson et al. 2010). Roy & Raguso (1997) did not find a strong attraction of some flies to indole, further indicating that flies respond differently to floral odours.
Certain Diptera do not find mates at flowers, but obtain rewards from flowers in the form of chemical precursors to sexual attractants. The gathering of floral scents is known among the Euglossinae (Apoidea), in which it is used by the males to attract mates (Eltz et al. 1999). In the Diptera, possibly the best studied is the collection of the scent components methyl eugenol, zingerone, and related phenylpropanoids, for use in pheromone synthesis by some male Bactrocera (Tephritidae) (Nishida et al. 1997(Nishida et al. , 2004Shelly 2001;Tan et al. 2002Tan et al. , 2006Tan et al. , 2011Tan & Nishida 2007. Some Bactrocera also collect raspberry ketone, another phenylpropanoid, as a defensive compound against predators (Tan & Nishida 2005).

Tactile advertisements and guides
Although texture has been invoked as a sensory cue by which pollinators can orient on flowers, and distinguish between flowers of different species (Kevan & Lane 1985), the sense of touch in Diptera is hardly studied. The fly Exorista japonica Townsend (Diptera: Tachinidae) may use, through tarsal examination, the texture and curvature of substrates in oviposition behaviour (Tanaka et al. 1999). Vogel & Martens (2000) noted that tactile cues characteristic of fungi may be important in attracting Mycetophilidae and Sciaridae to spathes of Arisaema spp.
(Araceae), and Kevan (1970) observed mosquitoes using grooves in the corollae of Pedicularis (Scrophulariaceae) flowers to guide their mouthparts to the nectar. Pollen texture and exine ornamentation may play a role in feeding by anthophiles, as in Eristalis tenax (Wacht et al. 2000).

FLORAL ATTRACTION OF FLIES BY DECEIT
The classification of deceptive attraction for purposes of pollination used below generally follows that presented in the review by Dafni (1984). Many mutualisms are open to "cheating" by one partner or the other, and plant-pollinator interactions are no exception (Dafni 1984;Addicott 1998;Schiestl et al. 1999;Paxton and Tengö 2001;Willmer 2011). In many instances flowers mimic the cues used by flies to locate other flowers or substrates that they normally visit for sustenance or oviposition, activating the innate or learned searching behaviour of prospective visitors to search for a nonexistent reward. Pollen or nectar rewards may be available to the visitor, but the advertised reward is not. J Poll Ecol 12(8)  Tan et al. 2002Tan et al. , 2006Tan et al. , 2011Nishida et al. 2004;Tan & Nishida 2005 Epipactis palustris (L.) Crantz Syritta pipiens L. (Syrphidae) Brantjes 1981 Listera cordata (L.) R. Br.

F Mycetophilidae
Ackerman & Mesler 1979 Paphiopedilum villosum  (8) is not. Because it is often difficult to determine whether or not a reward is present in a deceptive flower, we consider flowers that initially mislead a visitor as deceitful even if a reward is later offered (Ackerman & Mesler 1979;Mesler et al. 1980). For example, Rafflesia spp. (Rafflesiaceae) are usually considered deceptive (Beaman et al. 1988), but Rafflesia kerrii Meijer apparently delivers substantial rewards to its calliphorid pollinators (Bänziger 1991; see also Meve & Liede 1994). The flowers attract pollinators by their form and "cadaveric stench of rotting snakes", so acting as a sensory trap upon flies searching for food or a brood site when neither of these are present. In addition, a fruity scent arising from the center of the flower (not previously reported because it was hidden by the stench) may deter the flies from laying doomed eggs. The anthers of male flowers exude a nutritious pollen mush that is likely consumed by the flies, and flowers of both sexes secrete a potentially rewarding slime. Some mimetic trap flowers do provide an apparent reward for visitors, but it may not be used, such as nectar in Arum maculatum L. (Lack & Diaz 1991). In trap flowers such as aroids, stigmatic exudates and nectaries may function to increase the lifespan of the visitor by maintaining the relative humidity of the chamber (Daumann 1971;Wolda & Sabrosky 1986).
In such blooms, the unpleasant odours are often considered the long-distance lure, and in dense forests where some species are found, flowers may be visually inconspicuous (Percival 1965;Bänziger 1991Bänziger , 1996. On the other hand, Faegri & van der Pijl (1979) suggested that the carrion flies patrol large areas on the wing and are likely to see the flowers. Certainly, some such blooms are huge and visually conspicuous (e.g., Amorphophallus (Araceae), Rafflesia, and some Aristolochia spp.), and also emit large quantities of volatiles. Many aroids (Araceae) volatilize the amines by rapid respiration and the concomitant release of heat and carbon dioxide also simulate a substrate for oviposition (Dormer 1960;Meeuse 1966Meeuse , 1978Moodie 1976;Albre et al. 2003;Seymour et al. 2003;Quilichini et al. 2010;van der Niet et al. 2011but see Uemura et al. 1993. Many orchids in the huge genus Bulbophyllum (c.2000 species) smell of carrion, sometimes of rather specific varieties, and attract appropriate fly species as pollinators (Pemberton 2010).
The nectar of sapromyophilous flowers offering it tends to be rich in amino acids, by an order of magnitude higher than in other flowers. These high levels of amino acid may reward female flies lured from their normal oviposition substrates (Baker & Baker 1983a, b) with fuel for flight and nutrients for ovarian maturation. Nectar may also be a means of correctly orienting the visitors to reproductive structures .
Visual attraction of carrion flies to sapromyophilous flowers is presumably enhanced by their brown, red and purplish, and often blotchy, flowers, which further the mimicry of breeding and feeding venues for the flies (Proctor & Yeo 1973;Pemberton 2010). Wrinkled surfaces, filamentous appendages, and blotches (perhaps resembling hordes of indulgent flies) are also common and may serve as attractants, and windows (transparent areas) direct the movements at close range of positively phototactic visitors (Faegri & van der Pijl 1979;Seymour et al. 2003). The blotches also produce patterns that are more attractive than monotone surfaces to landing Diptera (Steiner 1948). Ultraviolet reflectance may be important for the attraction of sapromyophilic Muscidae (e.g., Moring 1978;Agee & Patterson 1983), but has not been investigated. Although most flies are thought unable to perceive red, certain carrion flies can detect red reflections (Autrum & Stumpf 1953), but it is not known if those are involved in colour perception (see . In a series of experiments, Kugler (1951Kugler ( , 1956 showed that greenbottle flies (Lucilia sp.), blow flies (Calliphora sp.) and flesh flies (Sarcophaga sp.) favour yellow and white coloured models over brown and purple ones in the presence of a sweet scent, but the opposite in the presence of a carrion scent. Unscented models were ignored.
It follows that the suite of characters, common to deceitful flowers such as Ceropegia, Aristolochia, Rafflesia and Stapelia, act synergistically to mislead anthophiles.
Overall, it appears that olfaction and vision (including colour vision) are important, but that each sensory modality may take precedence at different distances depending on plant species and floral stage, pollinator species and physiological state, and environmental conditions.

Other host mimicry
Some flowers mimic odours and substrates that are attractive to guilds of Diptera other than the carrion flies mentioned above (Tab. 7). Best known in this regard are the flowers that mimic Basidiomycetes and entice fungus gnats (Mycetophilidae, Sciaridae) to oviposit. Plants of this type occur in various genera, such as Arisarum and Arisaema spp. (Araceae), Aristolochia, Asarum and Heterotropa spp. (Aristolochiaceae), Corybas, Dracula, Masdevallia spp. and Cypripedium spp. (Orchidaceae) (Vogel 1973(Vogel , 1978Dressler 1981;Sugawara 1988;Mesler & Lu 1993; (Kato et al. 1995). Typically, fungusmimetic flowers pollinated by fungus gnats are dark purplish brown and borne close to the ground, and they often exude a strong fungus-like odour, which is the main attractant (Vogel 1973). Although other mimetic characters such as visual and tactile cues are presumably important attractants, odours have the most important role, at least in Arisaema (Vogel & Martens 2000;Barriault et al. 2010) and in Dracula (Dentinger & Roy 2010;Endara et al. 2010). High humidity caused by intense local transpiration further intensifies the mushroom guise. In Artocarpus integer (Thunb.) Merr., female gall midges pick up pollen while ovipositing and feeding on fungus that infect the male flowers; female flowers do not contain fungus, but mimic the scent of infected male flowers to attract the pollen-bearing flies (Sakai et al. 2000).
Tactile cues may play a role in short-range attraction and guidance of the exploring flies once they alight on the flower. Fungoid lamellae and pores are found on parts of the blossom in bolete-mimics, and the deception is so complete that the flies oviposit, and in the process pick up pollen grains in their hairs (Sugawara 1988). On Asarum hartwegii in northern California, Mesler & Lu (1993) found eggs of several fly species (see Tab. 7). The fungivorous larvae usually die upon eating the toxic tissue of these plants. Monoecious Araceae are protogynous, luring pollen-bearing flies and trapping them in contact with the female flowers, then releasing pollen and allowing the flies to escape (Albre et al. 2003). Sciaridae and Mycetophilidae attracted to the kettle trap blossoms of dioecious Arisaema escape through exit holes of male spathes, but die in female spathes after potential pollination (Vogel & Martens 2000;Barriault et al. 2010). In fungus-mimetic flowers pollinated by mycophagous Diptera such as Muscidae and Phoridae, there may be severe reduction of reproductive success of females ovipositing into the flowers. These factors may result in a coevolutionary arms race between fungus gnats (the search image for brood substrate) and plants with fungusmimicking flowers (deceptive stimuli).
Other flowers imitate the odours and, in some cases, morphologies of prey or host animals (Knoll 1926;Vogel 1973Vogel , 1978Crosswhite & Crosswhite 1984). Certain female simuliid flies are strongly visually attracted to distinct body parts of their hosts, and Wilhelmia equina L.
Epipactis veratrifolia Don. (Orchidaceae) accomplishes the same effect with scent by mimicking aphid alarm pheromone (Stoekl et al. 2011). While nectar may be offered to the ovipositing syrphids, if the larvae are unable to find real aphids to eat they will die (Pemberton 2010;Stoekl et al. 2011). Many milichiid flies are kleptoparasites, taking nourishment from seeping wounds of insects captured by spiders, and Ceropegia dolichophylla Schltr. (Apocynaceae) mimics the odour of insect hemolymph to lure milichiid pollinators (Heiduk et al. 2010).

Pseudonectar and pseudopollen
Pseudonectaries (organs resembling nectaries but not producing nectar) take the form of glistening, shiny, or refractile structures or areas on the blossom, and may distract pollinators from true nectaries, if they exist (Faegri & van der Pijl 1979;Chase & Peacor 1987;McDonald & van der Walt 1992;Vogel 1993;Bänziger 1996;Pemberton 2010). The exact role of the pseudonectaries in reproduction is unclear in many cases (Tab. 7). Glistening structures, such as the "mesmerizing wart" of Paphiopedilum villosum (Lindl.) Stein (Orchidaceae) may lure visitors into a trap where they contact the pollen (Bänziger 1996), or possibly distract less desirable visitors from rewards intended for pollinators, as in Parnassia palustris L. (Saxifragaceae) (Kugler 1956). Many flowering plants have a corolla bearing yellow, UVabsorbing spots that imitate the colour of pollen and anthers (Lunau 2000;Pansarin 2008). Visitors attempting to collect pollen from these spots may be manoevered into contact with the real pollen, and thus serve as a form of floral guide although no reward is given (Dafni & Calder 1987;Vogt 1990). The petals of some Saxifraga flowers (Saxifragaceae) display a particular type of floral guide consisting of an array of small spots on the petals that are apically red and basally yellow and combined with red or white pollen and anthers. Using artificial flowers that simulated this colour pattern, naive Eristalis tenax L. (Syrphidae) were selectively directed by the transition of colour from red over orange to yellow and moved towards the yellow coloured dots more often than towards the red coloured dots (Lunau et al. 2005). In actual flowers, attraction to the yellow spots acts to bring the pollen-foraging fly into contact with the actual pollen, which has less visibility or contrast, without it being eaten. Some plants have floral staminodes that produce pseudopollen (Tab. 7). The material may be used to provision brood by bees (Apoidea), but actual feeding by Diptera on pseudopollen has not been reported

Mimicry of rewarding flowers by non-rewarding flowers
In "mistake pollination" male and female unisexual flowers differ in the reward that they offer, and potential pollinators would be less likely to visit the poorer-rewarding flowers if they were distinguishable, adversely affecting crosspollination (Baker 1976). The most common example is the lack of pollen in female flowers, which then mimick the appearance of male flowers (e.g., have sexually functionless anthers), and attract visitors seeking the missing reward (Agren et al. 1986;Bierzychudek 1987;Aronne et al. 1993;Charlesworth 1993;Yuan et al. 2007Yuan et al. , 2008. Eristalis tenax and other Syrphidae visit the rewarding flowers of male plants of "cryptically dioecious" Rosa setigera Michx. (Rosaceae) in southern Ontario, Canada, early in the morning, but seem to be displaced to the female plants by bees later in the day. The malformed and inviable "pollen" J Poll Ecol 12 (8) from the female plants probably provides some nutrition. This strategy provides systematic pollen transfer from males to females over the course of the day, but the relative importance of the native anthophiles to the flowers of this plant for pollination could not be assessed given the preponderance of European honey bees (Kevan et al. 1990).
There are also examples of heterospecific non-rewarding flowers that are pollinated via their mimicry of rewarding flowers (Tab. 7). At a community level, nectarless plants may ensure visitation if their flowers are similar to those of rewarding species, in a form of Batesian mimicry. There are numerous examples, particularly among the tubular flowers visited by long-tongued Nemestrinidae, Bombyliidae, and Tabanidae in southern Africa, of plant species that offer no nectar reward (e.g., Schiestl 2005; Pemberton 2010; see Tab. 7). These plants often co-occur with a suite of plants that have similar, but nectariferous flowers, and it has been proposed that floral mimicry alone ensures visitation to the nectarless species. This system encourages out-crossing in the mimic and acts as an evolutionary selective force in maintaining populations of both model and mimic.

Sexual deception
Some flowers mimic the appearance of female insects or emit scents that mimic sex pheromones to attract males, sometimes at concentrations much greater than the female insects themselves (Schiestl 2005) (Tab. 7). The best known example is pseudocopulation by male bees and wasps with flowers that mimic females (e.g., Ophrys orchids; Faegri & van der Pijl 1979). Pseudocopulation is less widely documented in the Diptera, but is a possible mechanism in Microdon hoverfly pollination of some Ophrys orchids (Paulus 2005), and confirmed in sciarid pollination of several Lepanthes species (Blanco & Barbosa 2005;Pemberton 2010). There is also evidence that spots in certain Asteraceae, Geraniaceae, and Linaceae are reflective mimics of female bombyliids that may be attractive to males seeking mates (Johnson & Midgley 1997;Ellis & Johnson 2010).

Floral larcenists
Some insects are ill-adapted for the pollination of flowers they visit (Faegri & van der Pijl 1979;Inouye 1980), and may become nectar (or pollen) robbers or thieves (see Inouye 1980 for the terminology of floral larceny). Robbers bite holes at the bases of corollas in order to reach hidden rewards. Because they do not interact with the reproductive structures of the flower in obtaining the reward, they do not cross-pollinate the flower. Although it is unlikely that many Diptera are nectar robbers due to lack of suitable mouthparts, many small Diptera are probably secondary robbers that utilize the holes created by primary robbers. For example, Kevan (unpubl.) often noted mosquitoes feeding through holes made by bumble bees (Bombus spp.) in the bases of the tubular flower of Mertensia ciliata (Torr.) G.
Don (Boraginaceae) in the subalpine zone of Colorado. Other nectar (or pollen) thieves may enter flowers to obtain nectar without damaging the flower, but also without contacting anthers or stigmas.
Any fly species that commonly visits a plant for its rewards, but seldom pollinates it, is a thief. For example, flies were the most common visitors of Listera ovata (L.) R. Br.
(Orchidaceae) in Sweden, but they were considered unimportant as pollinators (Nilsson 1981). Pont (1993) lists some Muscidae that commonly visit flowers for nectar, but avoid contact with the anthers. Delia flavifrons Zetterstedt abuses the system even further: These flies feed, mate and lay their eggs on flowers of Silene vulgaris (Moench) Garcke (Caryophyllaceae), but do not pollinate them (Pettersson 1992). Given the small size of many Nematocera and Acalyptratae, it is likely that they are often nectar and pollen thieves, but the importance of this behaviour to either the thieves or the flowers has been littlestudied. Zhang et al. (2013) found that flies of the genus Scatopse (Scatopsidae) were nectar thieves of Corydalis ambigua (Papaveraceae) flowers, and that their presence deterred the pollinating bumble bees (Apidae: Bombus sp.).

CONCLUSIONS
The taxonomic, ecological and behavioural diversity of Diptera as anthophiles is probably greater than for any other insect order, including Hymenoptera. Myiophily is a major component of pollination, and the floral features reflect fly biology. Nectar is the most sought reward by flies, many of whom depend on the energy it provides for flight, and often has other nutrition benefits such as amino acid content. Pollenophagy is widespread in Diptera, but not as universal as in Apoidea, which primarily collect pollen for larval provisioning. Pollen is highly variable in its chemical and nutritional characteristics, and even pollen form sometimes reflects the importance of pollination by Diptera. Flies also seek mates, prey, protection, and brood sites in flowers. They find and forage at flowers by visual and olfactory cues that present a complex array of attractants of differing relative importance depending on the types of flies, their sexes, their immediate nutritional needs, and innate and learned behaviours. Sensory capabilities of flies are outstanding among insects and are highly varied from colour vision to shape, size, and motion perception and contact chemoreception to olfactory discrimination.
Our synthesis illustrates the complexity of myiophily in terms of rewards and attractants, but elucidates some general and unifying principles. Overall, it is difficult to separate attractants and rewards, and even more difficult to tease apart the relative importance of the different sensory modalities of floral attraction to Diptera, or even for particular taxa within the order. Moreover, the complexity of myiophily, from neurophysiology to community ecology, begs for imaginative and synthetic research. There is need for metabolic and physiological studies on pollen and nectar consumption and neurophysiological studies on the perception of floral colour and odour stimuli for a better understanding of fly visitation and activity in pollination. It becomes evident that myiophily is not a clear-cut phenomenon but rather multifaceted, integrating over groups of Diptera with different physiological capabilities and different motivations to visit flowers. Especially, competition with bees has not been adequately studied, though many flowering plants achieve reproductive success through both bee and fly pollinators (e.g., Kearns & Inouye 1994). The trade-offs for flowering plants that either specialise in fly pollination or compromise by attracting and nourishing bees as well as flies and use both as pollen vectors, are not well understood, even though "generalist" pollination by diverse arrays of pollinators has excited recent scientific interest (Waser et al. 1996;Herrera 1996;Ollerton 1996;Johnson & Steiner 2000).