Introduction

Globally, the most common colours of bird-dispersed fleshy fruit are red and black1,2,3,4. Other fruit colours occur at lower frequencies and are dispersed chiefly by non-avian frugivores1,5. Fruit colours are commonly considered a key feature in attracting avian dispersers because birds have excellent colour vision systems and probably use colour to locate fruits. As such, the frequency of fruit colours in many habitats may be the evolutionary result of the differential foraging preference of birds for these colours. In other words, it has been hypothesized that fruit colours are adaptive in attracting bird frugivores6 (and references therein [6]). This traditional hypothesis has recently been strengthened by some studies indicating that fruit colour and trait diversity can be well explained by their animal consumers7,8,9, more so than by phylogenetic relationships among plants10. In addition to color, conspicuousness is also an important visual signal affecting bird food preferences and red and black fruit are generally more conspicuous than other colours4,11,12,13,14.

Although red and black are the fruits dispersed chiefly by birds, it is still contentious as to whether the frequency of fruit colours, especially black and red, in the wild match the colour preferences of bird consumers. For example, some studies have shown that birds select fruit of a certain colour, but that colour was not necessarily red or black15,16. Other studies found that birds prefer red fruit17,18,19, but there is wide individual variation in fruit-colour preference12,16,19. These inconsistencies in colour preferences suggest that birds may not provide strong and constant positive selection for red and black fruit. It should be remembered that other factors, such as phylogenetic inertia of plants20 and other functions of pigment in fruit may also influence fruit colours. For example, the important pigment class of anthocyanins may play multiple roles in protecting vegetative tissue and fruit and reducing fungal growth21. Further, the selection of fruit by birds may be driven by factors other than colour: traits like nutritional contents and secondary compounds also influence bird food preference6,21. For example, birds may actually prefer anthocyanins because of their antioxidative properties22. As nutritional contents and colour often positively covary in bird-consumed fruits and birds are visually guided foragers, frugivorous birds can use colour as indicators of nutritional rewards9.

Experience is another important aspect of birds' fruit colour preference23,24,25. In the wild, early experiences of food items may be influenced by social learning26,27,28. These early experiences may help young birds to identify favourable food sources and avoid unfavourable food sources later in life. In some species, the effect of early social learning on aspects of food lasts for life28. Thus hand-raised naïve birds are advantageous for determining the influence of experience on food selection by fruit-eating birds because, unlike wild-caught conspecific birds, their colour preferences have not been influenced by previous experience. If colour preference is innate, hand-raised birds are expected to prefer fruits of the most common colour in their environments. This is because common fruit colours (e.g., red) can facilitate the detection and recognition of food sources by inexperienced birds25,29. Therefore, one important aspect for understanding whether the frequency of fruit colour is influenced by the preferences of frugivorous birds is to determine whether innate preferences match common fruit colours. Although this question has been addressed in some studies by comparing the colour preference of wild-caught animals to hand-raised naïve animals (e.g.24,25,30), the extent to which both learnt and innate colour preferences reflect the frequency of fruit colour in nature has not yet been clearly established. Furthermore, in the few studies that tested naïve colour preference, there was strong inter-individual variation (e.g.30). In the absence of a stable pattern, naïve preferences do not seem likely to be related to the frequency of fruit colour. Therefore, more studies are needed to improve our knowledge of the colour preferences of bird dispersers, especially using hand-reared birds.

Tropical Asia is an ideal location for expansion of our knowledge of the colour preferences of birds because tropical Asia has the largest diversity of frugivorous bird species in the world31. These birds also play important roles in the seed dispersal of tropical Asia32,33. However, very few studies have been conducted in this region (but see Duan and Quan34). In this study, we experimentally addressed several questions concerning the preferences of frugivorous birds. Four bird frugivores from tropical Xishuangbanna, southwest China, were used in this experiment, selected because of their abundance at the study site and because they are representative of two important families of Asian frugivores (bulbuls and barbets)31. The following questions were addressed: (1) Do naïve hand-raised birds prefer fruits of a certain colour, in a way similar to wild-caught birds? Does this preferred colour correspond to the frequency of ripe fleshy fruit colours in this area? It was hypothesized that black and red fleshy fruits would be preferred by such hand-raised birds, as well as wild-caught ones. (2) Does experience influence the colour preferences of birds for fruit? Do the colour preferences of experienced or naïve birds better reflect local frequencies of fruit colour? We hypothesized that the colour preferences of naïve birds would be closest to the frequency of fruit colour in the wild. (3) Are the colour preferences of both experienced and naïve birds constant over time (specifically over several days of repeated trials)? We used the following colours for testing these hypotheses on artificial and natural fruit: black, blue, green, red and yellow; however, note that the artificial black colour was found subsequently to be close to the artificial blue colour in its spectral reflectances34 (see discussion).

Results

Artificial fruit colour selection

Juvenile birds of all species differed in colour preference of artificial fruit from what would be expected randomly (ANOVA, F4, 555 = 639.1, 117.5 for Pycnonotus jocosus, P. aurigaster respectively and F4, 380 = 79.3 for Megalaima asiatica; P < 0.001 for all). Hand-raised P. jocosus, P. aurigaster and M. asiatica (14, 14 and 9 individuals, respectively) all preferred red artificial fruits over artificial fruits of other colours (Figure 1A).

Figure 1
figure 1

Mean (±SE) number of artificial fruits consumed by hand-raised (A) and wild-caught bird species (B).

Different letters within each panel indicate statistically significant differences (p < 0.05) between these five colours (ANOVA followed by multiple Bonferroni-corrected two sample T-tests).

Wild-caught P. jocosus, P. aurigaster and P. melanicterus (10, 5 and 7 individuals, respectively) also showed a strong difference in colour preference for the artificial fruits (ANOVA, F4, 295 = 74.4, F4, 160 = 22.2 and F4, 235 = 47.8 respectively; P < 0.001 for all) with red fruits consistently preferred over fruits of any other colour (Fig. 1B). Wild-caught barbets did not eat any artificially coloured fruit during the experiment.

There was weak inter-individual variation in the artificial fruit colour preference among hand-raised juveniles (Appendix S1-1). All 14 juvenile P. jocosus showed a consistent preference for red fruits over fruits of other colours. However, only 11 of the 14 juvenile P. aurigaster and 7 of the 9 juvenile M. asiatica preferred red fruits over fruits of other colours. Approximately 86% (32/37) of all hand-raised birds showed a significant preference for red fruits.

There was also weak inter-individual variation in the artificial fruit colour preference of wild-caught adult birds. All 10 P. jocosus, four of the five P. aurigaster and six of the seven P. melanicterus individuals showed a significant preference for red fruits (Appendix S2-1). In total, about 90% (19/21) of the wild-caught individuals had a significant preference for red artificial fruits.

Natural ripe fruit colour selection

Hand-raised juveniles also showed a strong difference in colour preference for the fleshy fruits from what would be expected randomly (ANOVA, F4, 535 = 139.6, F4, 455 = 298.4, F4, 370 = 78.7 for hand-raised P. jocosus, P. aurigaster and M. asiatica, respectively, P < 0.001 for all). Juveniles of all species preferred black fruits to red fruits and ripe fruits of other colours (ANOVA followed by multiple Bonferroni-corrected two sample T-tests, P < 0.01 for black vs. red, black vs. yellow, black vs. green and black vs. blue). In general, black fruits were always the most preferred and ripe green fruits were the least preferred for all species of hand-raised juveniles (Fig. 2A). It is important to recognize that the juveniles' preferences may have been influenced by nutritional differences among the fruits, once they had consumed them. However, if one only considers the first fruit an individual consumed, the pattern remain similar: 16 of the 31 hand-raised juveniles first ate black fruits, with red fruits being the second most consumed category (8 individuals) (Appendix S3).

Figure 2
figure 2

Mean (±SE) number of fleshy fruits consumed by hand-raised (A) and wild-caught birds (B).

Different letters within each panel indicate statistically significant differences (p < 0.05) between these five colours (ANOVA followed by multiple Bonferroni-corrected two sample T-tests).

Wild-caught adults also showed a strong colour preferences for fleshy fruits, different than what would be expected by chance (F4, 390 = 55.6, F4, 135 = 14.5, F4, 250 = 44.7 and F4, 130 = 12.6 for wild-caught P. jocosus, P. aurigaster, P. melanicterus and M. asiatica, respectively; P < 0.001 for all). However, with the exception of P. jocosus, which preferred red over other colours, wild-caught adults showed a significant preference for both red and black fruits over fleshy fruits of other colours. For example, M. asiatica and P. melanicterus adults showed no difference in preference for flesh black and red (P > 0.5), but they preferred black/red fruits to yellow, green and blue flesh fruits (P < 0.05 for all). Green fruits were also the least preferred by all species of wild-caught adults (Fig. 2B).

All juvenile individuals displayed a consistent preference for one or both of the common fruit colours (black and red) over ripe fruit of other colours. Overall, approximately 71% (22/31) of the total number of hand-raised individuals showed a preference for black fruit, 13% (4/31) preferred red fleshy fruit and the remaining 16% (5/31) preferred both black and red fleshy fruit over fruit of other colours (Appendix S1-2).

There seemed to be a larger inter-individual variation in natural fruit colour preference among wild-caught adults than was observed for hand-raised juveniles. However, if red and black are lumped together as “common colours”, then 25 of the 26 wild-caught birds (96%) showed a significant preference for one of the common fruit colours (including 4 individuals that preferred both a common colour and one other) and only one individual (4%) totally avoided the common colours of natural fruits, preferring to consume yellow fruit (Appendix S2-2).

Temporal variations in preference

Artificial fruit colour preference was stable over time for hand-raised juveniles and red was always the most preferred colour (Table 1). For natural fruits, black was the most stable and preferred colour by juvenile P. aurigaster during the entire experimental period. Hand-raised P. jocosus and M. asiatica showed a similar pattern in that they preferred black fruit on the first few days of the trial period, then expanded their preference to include both black and red fruit on subsequent days of the trial period (Table 1). In general, only the common colours were constantly preferred by all hand-raised birds for the entire trial period.

Table 1 Stability of fruit preference by hand-raised juvenile birds during the continuous trial period (day)

Artificial red fruit was also the most stable and preferred colour for wild-caught adults over time (Tables 2). The fleshy fruit colour preferences of wild-caught adults varied more substantially than those of juveniles of the same species. Wild-caught P. jocosus preferred red natural fruit on the first day of trials, both black and red fruit on days 2–4, black fruit on day 5 and again preferred black and red fruit on days 6–8. P. melanicterus showed a preference for red and yellow fruit on the first day of trials and then mainly focused on black and red fruit during the remainder of the trial period. During the first few days of the trial, wild-caught P. aurigaster and M. asiatica showed a preference not only for black and/or red fruit, but also for blue fruit. These birds then preferred only black and/or red fruits for the remainder of the trial period (Table 2). In general, a preference for common-coloured ripe fruit was the most stable and a consistent preference shown by all wild-caught birds over time (Table 2).

Table 2 Stability of fruit preference by wild-caught adult birds during the continuous trial period (day)

Discussion

Colour preferences of wild-caught experienced birds

The significance of fruit colour diversity in animal foraging decisions remains contentious, but if plants have adapted fruit colours that attract bird frugivores6, frugivorous birds should exhibit a preference for the common colours, black and/or red fruits, over fruits of uncommon colours. We examined this assumption and found that all experienced wild P. jocosus, P. melanicterus and P. aurigaster caught in Xishuangbanna, tropical Asia exhibited a strong and consistent preference for red artificially coloured fruits over others. This result is similar to that of some studies in which birds displayed a preference for red fruit6,17,19 and consistent with our previous work34, but contrasts with other studies in which birds displayed a preference for colours other than red or for no specific colour15,24,25,35. In the present study, the consistent preference of all wild-caught birds for red artificial fruit seems to support our hypothesis that birds prefer the most commonly encountered fruit colours, as blue, green and yellow are comparatively rare in the forest20. Surprisingly, red was preferred to black, even though black is actually more frequent in the study area (40% of fruit) than red (19%). However, the fact that no bird species in the experiment preferred artificially coloured black fruit is likely because the artificially coloured ‘black’ used in our experiment had a reflectance peak in the blue range that may have led birds to perceive it as blue34. All other artificial colours had reflectances which would lead birds to perceive them as the same colour humans perceive them as34. In these experiments more birds consumed red and yellow fruits than would be predicted by the frequency of natural fruits of those colours, but this was possibly caused by the lack of truly black artificial fruits in the experiment.

Although black is one of the most common colours of fleshy fruit dispersed by birds, only a few studies of bird fruit colour preference have included black fruits in their experiments (e.g.12,15,19,24,36) and in very few of these studies have birds showed a preference for black fleshy fruits36. Furthermore, these previous studies on colour preference primarily used cultivated fruits or migrant bird species. In comparison to resident birds, migrants may have experienced more variable environments and learned to eat a broader range of differently coloured food items during migration37. Therefore, it is probably not reasonable to expect migrant birds to choose fruits based primarily on the frequency of fruit colours in the environment where they were caught. This might explain why the authors of many previous reports failed to find a positive relationship between the frequency of a fruit colour in the environment and the colour preference of frugivorous birds. In contrast, the natural experience of resident birds could be more reasonably expected to reflect the frequency of fruit colour in the local environment.

Using five common fleshy fruits of native plant species and four wild-caught resident bird species, our study showed a consistent and strong preference for both black and red fleshy fruits over ripe fruits of other colours. Although the different foraging experiences of the wild-caught experimental individuals would no doubt influence food choices24,25,38, the wild caught resident birds overwhelmingly preferred black and red fleshy fruit. This is consistent with our previous studies39. However it should be noted that this study used different species of natural fruits than Duan and Quan39, which used Litsea glutinosa for black, Rauvolfia vomitoria for red, Duranta erecta for yellow and Microcos nervosa for green. The fact that the two studies showed similar results despite using different fruiting species suggests that that black and red are the preferred colours of fruit for birds in our study site. As indicated by previous studies, the reason birds prefer these colours might be because black and red are especially conspicuous against natural backgrounds compared to other fruit colours12, as well as the frequency of these colours in nature in this region.

To posit birds as a selective force on fruit colour, we must also demonstrate a preference against ripe fruit of less common colours. All wild-caught birds in our study consistently avoided artificial and natural ripe green fruits. These results are consistent with most previous studies, in which green fruits were avoided by frugivorous birds25,30 and our previous work34,39. Some studies attributed the avoidance of green to its relative inconspicuousness6,12, while others showed green was conspicuous against background12,19. However, most studies agree that green might signal unripeness and birds might use green as a cue to recognize unripe fruits6,19. Although the reason for the avoidance of green fruits by birds is still an open question, if bird preferences affect fruit colour, the consistent and constant pattern of avoiding ripe green fruits shown by all wild-caught individuals might explain why ripe green is the least frequent fruit colour in our study site. As in some previous reports1,3, we also observed that ripe blue and yellow fruits, which occur relatively infrequently in our study site, were only occasionally consumed by wild-caught individuals (Fig. 2, Table 2, Appendices S2-2).

Besides contrasts, some other selective pressures have also been reported to influence fruit colour. For example, anthocyanins may be important drivers of fruit colour evolution because of their protective properties for plants (e.g. they reduce the growth of fungi found in fruit tissue21). Further, birds also may seek out anthocyanins for their antoxidative properties and birds could use colour as a reliable indicator of the anthocyanin contents of fleshy fruits22. In addition, phylogenetic effects are also reported to be the factors that influence fruit colour evolution in our study site, accounting for approximately 52% of the variance in fruit colour20. However, a recent article that investigated phylogenetic effects on fruit colour at both global and local levels found “little indication of phylogenetic conservatism10”.

Colour preferences of hand-raised naïve birds

The naïve birds used in this choice trial had no prior experience with fruit colours. As hand-raised juveniles of all species showed a strong and consistent preference for red artificial fruits, we suggest that juveniles have an unlearned preference for red fruit. Hand-raised blackcaps (Sylvia atricapilla; Schmidt and Schaefer25), blackbirds (Turdus merula; Larrinaga19) and northwestern crows (Corvus caurinus; Willson and Comet30) also showed a similar tendency to prefer red foods. An innate preference for red might suggest that the commonness of red fruits helps juveniles to recognize palatable food resources25, or at least function as an important signal in food discovery29. However, in interpreting these results it should be remembered that the artificial black colour in this study may have been perceived as blue, as explained above.

Despite a consistent preference for artificial red food, all hand-raised juveniles from all three species showed a strong and consistent preference for black fleshy fruits to red and to fruits of other colours. The same pattern was seen when only analysing the first fruit an individual consumed (Appendix S3). This differs from the preferences shown by conspecific wild-caught birds, which preferred both black and red fleshy fruits over the fruits of other colours. In Xishuangbanna, the majority of ripe fruit are black (40%) and red are the next most common (19%). Therefore, the juveniles' unlearnt consistent preference for the black fleshy fruits of native species over the less common red fruit suggests that the frequency of natural fruit colour is better reflected by the innate preferences of these birds than by the preferences of wild-caught adults. Indeed, red fruits may be either ripe or unripe in nature and therefore red flesh is not always a reliable signal of palatability. In contrast, the colour of natural black fruits always signals palatability2,30 and thus it would be adaptive for juveniles to prefer black fruit over red. In addition, wild-caught birds have experienced a variety of differently coloured fruits in the wild. Because previous experience affects future choices24,28,38, it is probably not adaptive for these wild-caught birds to choose food based only on colour frequency. Consequently, red (the second-most common colour) may operate more strongly as a cue for food detection in experienced birds than in inexperienced birds.

Two other factors, the contrast of fruit colour against a background and the nutrition of fruits, could have influenced the decisions of birds. Conspicuousness against a background has been shown to be an important factor in birds' fruit preferences11,12,13,22. In our experiments it is not clear whether conspicuousness was a factor because all fruits were displayed against an artificial white background. In previous studies34, we have looked at the effect of contrast on fruit colour preferences of artificial fruit and have found that it doesn't change birds' basic preferences for certain colours like red; nevertheless, the effect of contrast should be explored in further work on the colour preferences of frugivores for wild fruits in this region. As to nutrition, it is possible that the natural fruits had different nutritional values; however if this was the case, we would expect changes over time in the birds preferences, which in general were not found (please see section on temporal variation, below). Further, as mentioned above, the hand-raised birds' first fruits chosen were predominantly black and then red.

Inexperienced birds of all three species showed a strong and consistent innate avoidance of both artificial and natural ripe green fruits. Although the avoidance of green may be easily overcome by experience35, the significant and consistent pattern of avoidance of ripe green fruits by both experienced and inexperienced frugivores in this study implies the existence of constant selective pressures exerted by bird dispersers against some less-common fruit colours (e.g., green).

Inter-individual variations in colour preferences

To assess whether food preferences by bird dispersers exert a stable selection pressure on the frequency of fruit colour, the preference pattern should be consistent and not easily changed6,16,30. Although some studies have found a preference for certain colours, most of them have shown strong intra-specific variation12,17,30. These strong inter-individual variations imply that there is a lack of consistent and constant selective pressure exerted by bird dispersers on the frequency of fruit colour.

In our study, the inter-individual variations in colour preference were very weak (Appendices S1-1 and S2-1). Nearly all of the hand-raised and wild-caught birds preferred red artificial fruits (86% and 90%, respectively). The inter-individual variation in natural fruit colour preferences might appear to be relatively large among wild-caught individuals, as some individuals preferred black, some preferred red and some preferred both. However, approximately 96% (25/26) of wild-caught birds significantly preferred one or both of the common-coloured (red or black) fruit (Appendix S2-2). It seems the differences in the experiences of wild-caught birds did not ultimately lead to strong inter-individual variation in colour preference for uncommon colours. Perhaps more significantly, all hand-raised birds (3 species, 31 individuals) showed a consistent preference for common-coloured fleshy fruits (Appendix S1-2). The consistent pattern exhibited by almost all individuals, whether wild-caught or hand-raised, implies the potential for directional selective pressures on natural fruit colour towards the common black and red fruit colouration observed in this area.

Temporal variation in colour preferences

Learning and experience can alter colour preference pattern of birds25,38,40, but if birds change their preferences during a short period in a response to external stimuli, they could not exert a constant selective force to drive fruit colour evolution. Therefore, temporal variation in colour preference might have important evolutionary consequences for plant-animal interactions. Both wild-caught and hand-raised birds in our study showed a stable preference for red artificial fruits over a several-day period (Tables 1 and 2). The colour of natural ripe fruits favoured by wild-caught birds shifted somewhat between trials; during the first few trials, wild-caught birds preferred both common (red and/or black) and rare (blue) fruit, but as the trials continued, wild-caught birds preferred only the common black and red fleshy fruits. All hand-raised juveniles showed a stable preference for black natural fruits, although their preference for red fruits increased over subsequent trials until they preferred both black and red fruit (Table 2). It seems that the preference for natural black by hand-raised birds reflected an inherited preference at the beginning of experiment which was modified with ongoing positive experience with red fleshy fruits. The highly constant preference towards the most common black and red fruit over time shown by wild-caught and hand-raised birds also implies a stable selective pressure that is reflected in the commonness of black and red fruit.

A limitation of our study is that for each fruit colour we only presented one species of fruit (as mentioned above, another study of our group exclusively looking at experienced adults found similar results with a different set of species). This limitation could be a problem for the interpretation of our results if the fruit species had different nutritional contents, because nutritional contents could influence food selection9,21. Hence further studies testing nutritional contents of these plant species are needed to see whether these traits also influence fruit choice. For now, we think it is reasonable to suggest that fruit colour is the dominant force behind birds' preferences, because if fruit nutrition is very poor we would expect wild birds to avoid such fruits. In contrast, we observed wild birds eating these fruits frequently (Duan, personal observation). Further, if there had been large nutritional differences between the fruit species, we would expect to see changes over time in the birds' preferences in our study. For example, we might expect birds to focus on only just red or just black fruits by the end of the experiment if one of these fruits had the highest nutritional content, not a mixture of both. It is possible that red and black fruits both had higher nutritional content than other fruits; yet we point out that for juvenile birds the first fruit consumed tended was most often black and next often red. Fruit size and shape are other important characteristics that may influence bird food selection41,42. However, the fruits used in our experiment were similar in size and in (roughly oval) shape, so we believe fruit colour explains most of the differential response to fruits in this study.

Conclusions

Our study, using resident hand-raised and wild-caught birds and both artificial and naturally ripe fruit of native plants, allows us to conclude the following: (1) All experimental birds have a strong preference for the most common fruit colours, black and red. We must acknowledge, however, that these same colours are also the most contrasting colours against a constant background and hence contrast, as well as colour per se, may play a role in our results. (2) Relative to adult wild-caught birds, the innate colour preference of hand-raised birds, with a strong preference for black natural fruit, better reflects the frequency of natural fruit colour in our study site. Although experience influenced the selection of natural fruit colours by wild-caught birds to a certain extent, the colours (black and red) preferred by them were still the most common colours of fleshy fruit. (3) The preferences for common-coloured fruit and avoidance of ripe fruit of an uncommon colour are relatively invariant at inter- and intra-specific levels and are also stable over time. Although we used only four species, which is a small number considering the diversity of Asian frugivores, these species are representative of some avian families of frugivores (bulbuls and barbets) that are dominant in human-modified landscapes in which larger frugivores are often extirpated43 and such landscapes are, unfortunately, increasingly common.

Methods

Ethic statements

This study was carried out in strict accordance with the Guideline for the Care and Use of Laboratory Animals of China. The protocol was approved by the Administrative Panel on the Ethics of Animal Experiments of Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences (Permit Number: XTBG2011-3).

Study area and species

The study was carried out at Xishuangbanna Tropical Botanical Garden (XTBG, centred at 21°55′N, 101°16′E) in central Xishuangbanna Dai Autonomous Prefecture, southwest China. This region has an especially high biodiversity, which includes 16% of the higher plant species and 36% of the wild bird species found in China44. The forests in this region can be grouped into four types: tropical rain forest, tropical seasonal moist forest, tropical montane evergreen broad-leaved forest and tropical monsoon forest45. In this tropical area, 40% of the 412 ripe fleshy-fruited plant species have black fruit, while 19% are red, 13% are brown, 13% are yellow, 8% are bicolour and a smaller number have green, white, or blue fruit20.

We tested the fruit-colour preferences of four bird species: Pycnonotus jocosus, P. melanicterus, P. aurigaster and Megalaima asiatica. Both wild-caught and hand-raised individuals were tested for each species except for P. melanicterus, for which only wild-caught birds were used. P. jocosus, P. melanicterus and P. aurigaster are members of the bulbul family (Pycnonotidae) and have mean body lengths of 19 cm, 19 cm and 21 cm, respectively46. M. asiatica is a member of the barbet family (Megalaimidae) and has a mean body length of 20 cm47. We selected these species because they are the most abundant bird species at our study site and the use of common birds facilitates finding nests and the acquisition of juveniles. Further, these birds are representative of important families of Asian avian frugivores. For example, of the bulbuls, R. Corlett (1998) writes: “Altogether, the available evidence suggests that the bulbuls may be most important seed dispersal agents in anthropogenic open habitats in the Oriental Region”. The barbets are “among the most highly frugivorous birds in the region, west of Wallace's Line”31. In Xishuangbanna, these birds consume large amounts of fruit from a wide variety of plant species, as shown by their stomach contents, which are mainly made up of the remains of different fruits46,47.

Ten Pycnonotus jocosus, seven P. melanicterus and four Megalaima asiatica were captured in January and February 2011, before the breeding season. Five P. aurigaster were captured in December 2011 in the non-breeding season. We planned to catch 7–10 individuals of each experimental bird species, but the number of P. aurigaster and M. asiatica was constrained as they are canopy dwellers and are therefore difficult to capture. Birds were kept individually in 30 × 30 × 40 cm cages. The maintenance diet of all birds consisted of apple, pear, banana and mealworms (which was given in a yellowish mixture); water was available ad libitum.

Juvenile P. jocosus (12), P. aurigaster (10) and M. asiatica (9) were taken from nests in XTBG between April and June 2011, 4–6 days after hatching. Juveniles were hand-raised and fed with apple, pear, banana and mealworms. All fruit was peeled prior to feeding in order to prevent juveniles from acquiring colour preferences based on fruit peels. Field video observation showed that prior to collection, juveniles were fed mainly caterpillars and insects (Li H., unpublished data). Therefore, the juveniles in our experiment had no experience with any natural fruit items before the start of the experiment. Juveniles and adults were held in different rooms to prevent feeding imitation of adults by juveniles. Juveniles were placed in individual cages (30 × 30 × 40 cm) at 5–7 weeks of age, when they were able to feed themselves.

Fruit type and colour

Both artificial and natural fruits were used for colour selection experiments. Artificial fruit used in our experiments was a mixture of apple, pear, banana, wheat and corn flour (1:1:1:1:1). We dyed the artificial fruit black, red, yellow, green, or blue using 0.01% tasteless food colouring34. The artificial fruits were shaped spherically with a diameter of 5–7 mm, similar to the natural fruit size consumed by bulbuls and barbets at our study site. Because the artificially coloured fruits were totally new to both wild-caught and hand-raised birds, the experiences of wild-caught birds with fleshy fruits might influence artificial food selection to a lesser extent than natural food selection. We therefore expected that both experienced and naïve birds would have similar patterns of preference for artificial fruit.

We also used ripe fruits of native species to test colour preference. Because we assumed that wild-caught birds had experience with these common fleshy fruits and previous experience influences food preference later in life, we expected the colour preferences of experienced (wild-caught) and inexperienced (hand-raised) birds to differ and specifically, for the innate preferences of naïve birds to better reflect the natural frequency of differently coloured fruit. We used naturally ripe fruits from five common species: Phaeanthus saccopetaloides (black), Psychotria fluviatilis (red), Debregeasia orientalis (yellow), Ficus langkokensis (green) and Ophiopogon japonicus (blue). While we unfortunately do not have reflectance spectra data for these species, spectra from other species of trees in this forest have generally been consistent with the qualitative categories of Chen et al20 (see Quan and Duan39) and have little reflectances in the ultraviolet, making it likely that human categories may match the perceptions of birds. These fruits have no secondary structure, matured simultaneously during the trial period, were similar in size and in (roughly oval) shape and were normally consumed by the experimental bird species in the field (Duan observ.). All species have dark green leaves.

Fruit traits, such as size and shape, nutritional value and conspicuousness14,35,42,48,49 affect fruit-bird interactions. However, as birds are visually-guided foragers, they commonly use colour/colour contrast as important signals to find and assess fruits12,22,50. Moreover, Valido et al.9 found that traits (e.g. morphology, nutrients and colour) of bird-consumed fruits showed a higher integration with colour than fruits consumed by other frugivorous animals. As these recent studies also indicated fruit traits and colour covary in birds consumed fruits, we ignored the potential effects of other fleshy fruit traits on bird food preferences in this study (as for the possible influence of variation among fruits in their nutritional value and morphology, please see discussion).

Experimental design

Four experiments were conducted to test bird colour preference: adult × artificial fruits, adult × natural fruits, juvenile × artificial fruits and juvenile × natural fruits.

Artificial fruits

Adult × artificial fruit food-choice experiments were conducted in March–April 2011 and February 2012 after the wild-caught adults had habituated to the experimental cages. Juvenile × artificial fruit experiments were conducted in late August 2011, after juveniles were able to eat independently. Each adult bird participated in six trials and each juvenile bird participated in eight trials. Trials were conducted on consecutive days from 09:00–11:00 or 13:00–15:00. Before each trial, maintenance food was removed from the experimental cage for 1–1.5 h, water remained available ad libitum. At the start of each trial, birds were exposed simultaneously to 40 artificial fruits of five colours (eight fruits/colour × five colours) and all fruits were placed randomly in a transparent Petri dish with white background. We allowed the bird to feed until approximately 8–10 fruits were eaten (roughly equal to the number corresponding to any individual colour of fruit) and then counted the number of unconsumed artificial fruits. Each individual was tested separately to avoid influence on colour selection from other birds.

Natural fruits

Following the same method as artificial fruits, we displayed 40 naturally ripe fruits (eight fruits/colour × five colours) simultaneously to each of the birds. Each adult bird participated in eight trials and each juvenile bird participated in nine trials. Trials were conducted on consecutive days. Adult × natural fruit trials were conducted in December 2011. Juvenile × natural fruit trials were conducted in January 2012, three months after the end of juvenile × artificial fruit trials to minimize any affect that the experience of juveniles in the artificial food choice experiment might have on their decisions regarding natural fruits.

Data analysis

As a response variable, we used the number of fruits consumed by an individual bird of each colour per trial. We used ANOVA to determine whether food preferences of bird species of different ages (hand-raised juveniles versus adults) differed between these five colours. We then conducted multiple Bonferroni-corrected two sample T-tests, to further examine the differences between each pair of colours.

We first examined colour preferences pooling all individuals of the same age group of the same species. Because during the continuous experimental period, a bird might learn that certain characteristics of fruits (e.g. nutrition, texture) were identical (for artificial fruits) or different (for natural fleshy fruits) and decrease or increase their degree of selection among the five colours, we examined whether colour preferences were stable over the 6–8 consecutive days of trials. Preferences of each species for each day were determined by ANOVA followed by multiple Bonferroni-corrected two sample T-tests, as above. Finally, we measured inter-individual variation among individuals of a species in an age class in colour preference.