On the reliability of visual communication in vertebrate-dispersed fruits

Authors

  • Eliana Cazetta,

    Corresponding author
    1. Departamento de Ecologia, Universidade Estadual Paulista – UNESP, C.P. 199, 13506-900 Rio Claro, SP, Brazil
      Correspondence author. Departamento de Ciências Biológicas, Universidade Estadual de Santa Cruz, Rodovia Ilhéus-Itabuna km 16, Ilhéus, Bahia, CEP 45662-900 Brazil. E-mail: eliana.cazetta@gmail.com
    Search for more papers by this author
  • Mauro Galetti,

    1. Departamento de Ecologia, Universidade Estadual Paulista – UNESP, C.P. 199, 13506-900 Rio Claro, SP, Brazil
    Search for more papers by this author
  • Enrico L. Rezende,

    1. Departament de Genètica i de Microbiologia, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain
    Search for more papers by this author
  • Hinrich Martin Schaefer

    1. Faculty of Biology, Evolutionary Ecology, University of Freiburg, Hauptstr. 1, 79104 Freiburg, Germany
    Search for more papers by this author

Correspondence author. Departamento de Ciências Biológicas, Universidade Estadual de Santa Cruz, Rodovia Ilhéus-Itabuna km 16, Ilhéus, Bahia, CEP 45662-900 Brazil. E-mail: eliana.cazetta@gmail.com

Summary

1. Several fundamental aspects of communication between plants and their mutualists are poorly understood. It remains unclear whether plant signals are accurate and reliable, that is, whether they recruit mutualists by informing them about the nutritional rewards provided by flowers and fleshy fruits.

2. We evaluated fruit characteristics of 105 vertebrate-dispersed species from an Atlantic rainforest community to examine the relationship between visual fruit stimuli and fruit quality. We combined morphological, biochemical and spectrophotometric methods and employed comparative analyses to assess the evolutionary association among morphological, biochemical and visual characters.

3. We detected a significant phylogenetic signal in most morphological and nutritional traits but not in colour traits, which suggests that fruit colours are evolutionarily labile. We also found that two properties of fruit coloration (brightness and chroma) explain 24–29% of variation in protein and sugar contents, respectively. High sugar content in fruit is associated with dark colour and low chroma (colour saturation). However, fruit brightness is not strongly correlated to protein content and its signalling role remains unknown.

4. We suggest that the negative relationship between chroma and sugar content in fruit can be explained by the fact that sugars up-regulate the synthesis of anthocyanins, the pigments primarily imparting achromatic coloration in fruits. Biochemical pathways could consequently explain the covariance between visual fruit traits and nutritional fruit traits. In this scenario, information could arise as by-product rather than as an adaptation to signal to seed dispersers.

5.Synthesis. We show that fruit visual traits are evolutionary labile and that fruit chroma and brightness can reliably indicate sugar and protein content, respectively. We propose that shared biochemistry is a likely mechanism inducing the covariance between colour (the stimulus) and nutritional quality (as the unseen quality). We hypothesize that by-product information represents a widespread and so far neglected mechanism leading to reliable visual traits in mutualistic and possibly also antagonistic plant–animal interactions.

Introduction

To overcome the constraints imposed by immobility, many plants produce conspicuous flowers and fruits to recruit animal vectors for reproduction and dispersal. In accordance with the widespread evidence that animals use colour and odour to find plant rewards, the colours and bouquets of flowers and fruits are thought to have evolved primarily or secondarily as communicative traits facilitating the recruitment of mutualistic animals. However, the fundamental mechanisms by which plants communicate with their mutualists remain contentious, particularly because defining communication is generally a controversial subject. The process of information transfer is key to many definitions of communication (e.g. Bradbury & Vehrencamp 1998), while others suggested that communication is not necessarily informative (i.e. transferring accurate information) but should instead be considered as a process of influencing or manipulating others (Dawkins & Krebs 1978; Maynard-Smith & Harper 2003; Rendall, Owren & Ryan 2009; Scott-Phillips 2010).

The reliability of plant–animal communication – apart from when indicating the location of a plant – is not well known. This uncertainty is surprising because plants are an ideal system to evaluate the reliability of this information because plant displays are relatively constant in space and time; they are often less complex and dynamic compared with animal displays and lack a movement component (Schaefer, Schaefer & Levey 2004). In contrast to animal communication where the interests of senders and receivers can quickly change (e.g. depending on the dynamics within social groups), it is also easy to define the potential information content of plant traits because animals’ interest in plant nutritional quality is relatively invariant and predictable. Clearly, animals can select for reliability, that is an accuracy between signal and reward (Armbruster, Antonsen & Pelabon 2005; Benitez Vieyra et al. 2010). However, fruit colours can also be strongly determined by selection by antagonists or abiotic factors (Strauss & Whittall 2006; Schaefer, Rentzsch & Breuer 2008; Schaefer & Braun 2009; Steyn et al. 2009), and it is unknown whether these selective regimes produce reliable communication between plants and animals.

There are currently two major limitations in the field of plant–animal communication that limit progress in the field. First, the correlations between colours (and likewise odours) as sensory stimuli that animals perceive and the nutritional qualities of plant rewards are mostly unknown. Only very few studies analysed whether colour traits are reliable on a community level. Schaefer & Schmidt (2004) found that fruit colours in the Venezuelan rainforest indicated fruits’ nutritional contents. Similarly, Raine & Chittka (2007) reported that violet was the most rewarding floral colour in a plant community in Germany and that bumblebees have an innate sensory bias towards violet. The second limitation is that these studies did not control for phylogenetic effects on fruit or flower colours, even though such effects strongly influence morphological and nutritional traits of plant reproductive organs (Herrera 1992; Jordano 1995). The lack of a phylogenetic perspective on the relationship between the colours of fruits and flowers and the nutritional rewards associated with them precludes an understanding of the evolutionary history of the covariance between those traits. Moreover, many comparative studies have shown that species are often not the appropriate sampling unit because traits do not evolve de novo in response to the current selective pressures in each species (Fischer & Chapman 1993).

To study the relationship between fruit stimuli and fruit quality, we evaluated morphological, biochemical and visual fruit traits in a large data set (105 species) from a phylogenetic perspective in an Atlantic rainforest community in Brazil. If fruit traits are reliable, we expect colour and morphology to indicate fruit nutritional quality. Fruits can contain both palatable and unpalatable components. In general, secondary compounds (responsible for fruit unpalatability) have complex functions in ripe fruits, and according to the most influential hypothesis, there may be a trade-off between defence against damaging agents and palatability for dispersers (Herrera 1982; Cipollini & Levey 1997). In the context of plant–animal communication, we thus expect information to similarly represent a trade-off between attracting visually oriented dispersers and fending off visually oriented fruit predators. Specifically, we predict that if seed dispersers have primarily influenced fruit communication, fruit colours should indicate the contents of nutrients such as carbohydrates, lipids or proteins. In the discussion, we consider the proximate mechanisms underlying the observed covariance among visual and nutritional traits and contrast the scenario of adaptive, reliable fruit communication to seed dispersers against a scenario of nonadaptive, reliable fruit communication.

To assess the evolutionary association among fruit characters, we tested whether the covariance among traits exhibits a phylogenetic signal by employing comparative analyses. We specifically evaluated whether any correlations among colour and morphology and fruit nutritional contents are associated with phylogeny or independent of it. We further tested whether fruit colour is an evolutionary conserved trait differing among higher plant taxa or whether it is an evolutionary labile trait that may adapt to different selection regimes. In our large data set, fruits were mainly dispersed by birds and primates. In our phylogenetically explicit analyses, fruit colour traits need to be assigned unambiguously. Hue, chroma and brightness are three colour properties that provide such unambiguous notation (Endler 1990). They are therefore the most commonly used colour variables in comparative studies (Montgomerie 2006). These three colour variables account for primate vision. To account for bird vision, we also included in our analyses the chromatic contrasts between fruits and their background, as seen by birds. Thus, this approach would allow us to assess the detectability and reliability of information transmitted by fruit visual stimuli.

Materials and methods

Study system

Fruits were collected at Ilha do Cardoso (25°03′ S and 48°53′ W), a pristine land-bridge subtropical island, located in the south of São Paulo state, Brazil. The study site is a 15 100 ha protected island composed exclusively of Atlantic rainforest, including lowland and highland tropical rainforest, mangroves, dune vegetation and restinga forest (Barros et al. 1991). The climate at Ilha do Cardoso is generally warm and wet throughout the year, with a mean annual rainfall of roughly 3000 mm and average ambient temperature of 23.8 °C (Funari, Struffaldi-De-Vuono & Salum 1987; Castro, Galetti & Morellato 2007).

We collected quantitative information on fleshy fruit traits for 105 vertebrate-dispersed species belonging to 44 families (see Appendix S1 in Supporting Information). The sample was dominated by trees (59 spp.), followed by shrubs (29 spp.), vines (eight spp.), herbs (five spp.), hemiparasites (two spp.) and bromeliads (two spp.), and represents c. 50% of the known vertebrate-dispersed plants in the area (E. Cazetta unpublished data). Fruits from these species are mainly dispersed by birds (74 spp.), followed by mixed agents (birds and primates as well as other mammals; 15 spp.) and primates and other mammals (16 spp.) (Appendix S1), which is a common pattern found in vertebrate-dispersed plants in the Atlantic forest (Almeida-Neto et al. 2008).

Fruit phenotypic traits

For each species, we collected c. 30 ripe fruits (range 8–50) from at least three individuals per species (for some species, we obtained fruits from up to five individuals). From these samples, we obtained data on fruit morphology, chemical composition and colour. Due to logistic constraints, sample sizes varied depending on the type of measurement: morphological variables were measured in all species (= 105 spp.), whereas fruit colour and chemical composition were obtained from smaller subsets of the total (= 58 spp. and 78 spp., respectively). Therefore, we report the total number of species or the degrees of freedom of each analysis.

Fruit morphology

Morphological traits were measured on 30 fruits per species. For each fruit, we quantified the following variables: length and diameter of the fruit and its seeds, fruit fresh mass, pulp fresh and dry mass, seed mass and number of seeds per fruit. From these measurements, we also calculated the relative yield of fresh fruits (dry mass of the pulp/fresh mass of whole fruit) and the pulp to seed ratio (dry mass of pulp/mass of the seeds).

Colour measurements and contrast calculation

To estimate fruit colour, we measured the reflectance spectra of 15 fruits per species, as well as 10 leaves and 10 accessory structures (nongreen pedicels, capsules) if present. We performed all measurements with an Ocean Optics USB2000 spectrometer and a Top Sensor System Deuterium-Halogen DH-2000 (both Ocean Optics, Duiven, the Netherlands) as a standardized light source. Fruit reflectance was measured as the proportion of a standard white reference tile (Top Sensor Systems WS-2). We calculated three variables that characterize reflectance spectra: total brightness, chroma and hue (Endler 1990). Brightness is the intensity of reflected light from a given surface and is measured as the cumulative sum of the light intensity reflected (R) between 300 and 700 nm. Variation in brightness is perceived on the scale of white to black. Humans (and presumably other primates as well) perceive chroma and hue as chromatic aspects of visual information. Chroma (or saturation) describes the colours’ similarity to a neutral grey; colours that contain no or very little grey are deeply saturated, whereas a grey with a little tint of colour has a low saturation (see Kelber, Vorobyev & Osorio 2003 for a thorough review). Chroma is calculated as reflectance variation between 300 and 700 nm ((RmaxRmin)/Raverage). In contrast, hue defines colour differences that are related to human colour categories, for example red, yellow and green. While there is no evidence that animals other than primates perceive hue and chroma as humans do, some experiments on chickens suggest that they can categorize colours in a way similar to humans (Jones, Osorio & Baddeley 2001). Hue (or spectral shape) is measured as the wavelength of maximal reflection (λ(Rmax)).

The visual system of our two most important seed disperser groups, birds and primates, differs with birds being superior in discriminating colour differences (Håstad & Odeen 2008). We selected the three variables that characterize colour properties that primates can extract from visual stimuli and that are also likely to be perceived by birds because they have superior visual abilities. To account specifically for avian vision, we calculated additionally chromatic contrasts between the mean reflectance of fruits and leaves or secondary structure according to eye models of avian vision (Vorobyev & Osorio 1998). The model first calculated the quantum catch of each class of single cones presented in a tetrachromatic visual system (LWS, MWS, SWS and UVS), based on analytical approximation of cone visual pigments and oil droplet spectra. The quantum catches are used to find relative contrasts against fruits and background. Contrasts were characterized in units of ‘just noticeable differences’ (jnd), following the model proposed by Vorobyev & Osorio (1998). One jnd is the threshold of discrimination; higher values indicate discrimination of two colours. For example, we found on Ilha do Cardoso that increasing fruits’ contrasts also increases fruit detection by birds up to a threshold of 40 jnds (Cazetta, Schaefer & Galetti 2009). We opted to employ this model because most dispersers in Ilha do Cardoso are birds, and our previous study showed that birds attend primarily to chromatic contrasts to locate fruits in this community (Cazetta, Schaefer & Galetti 2009).

Chemical analyses

To estimate fruit chemical composition, the pulp or aril from fruits was frozen at −18 °C following collection, until the moment of analysis. For each fruit sample, we ran three replicates and used the mean of the three values for statistical analyses. The following variables were estimated from chemical analyses: protein composition, lipid content, total soluble sugar content and secondary compounds (phenols).

Proteins were determined with the Kjeldahl method, which consists in estimating total nitrogen in the fruit as a surrogate for total protein (protein composition was calculated by multiplying total nitrogen by 6.25) (Jeffery et al. 1989). This conversion factor may overestimate the total amount of protein found in fruits, but it does not alter conclusions on the covariance between protein contents and fruit visual traits. Lipid content was determined employing the standard macro-gravimetric method (Bligh & Dyer 1959). Total sugar content (glucose + fructose + sucrose) was estimated with gas chromatography–mass spectrometry (GC–MS), because the interference of coloured material in fruit pulps or arils prevented the use of more standard colorimetric methods (e.g. the Anthrone method; see the study by Pooter & Villar 1997).

To determine the content of secondary compounds (phenols), we employed the extraction method proposed by Price & Butler (1977), employing butanol and methanol extracts (see details in the study by Schaefer, Schmidt & Winkler 2003). Although not all phenolic compounds have similar extraction probabilities, this index of phenolic compounds correlated negatively in our community with the consumption of fruits by birds and fruit pathogens alike (Cazetta, Schaefer & Galetti 2008), demonstrating that this index is biologically meaningful. We focused on phenols because they commonly occur in fleshy fruits, and they are the most widespread secondary compounds in ripe fruits (Buren 1970; Herrera 1982; Cipollini & Stiles 1992; Foley, McLean & Cork 1995). The contents of these compounds were analysed with photometric measurements (for details, see the study by Cazetta, Schaefer & Galetti 2008).

Statistical analyses

We reconstructed the phylogeny of plant species from Ilha do Cardoso with the Phylomatic software, employing the Phylomatic conservative supertree (Webb & Donoghue 2005). We analysed the phylogeny of fruit traits to evaluate to what extent our results on trait covariances are influenced by phylogeny. It was not our principal aim to study the evolution of fruit traits given that our sample did not comprise an adequate and balanced sample within the phylogeny. Because estimates of divergence times were not available, we tested four different arbitrary branch lengths: constant branch lengths (all branch lengths = 1), Grafen (1989), the variant Nee’s (Purvis 1995) and Pagel (1992). Statistical adequacy of the different branch lengths were tested by plotting the absolute values of standardized phylogenetic independent contrasts against their standard deviations, and branch lengths were considered statistically adequate if there was no correlation between these values. In essence, this means that there is no consistent bias as a function of divergence times during the calculation of contrasts (i.e. contrasts are uniformly distributed across long and short branches); hence, the standardization procedure can be considered adequate for statistical purposes (Garland, Harvey & Ives 1992). These analyses were performed with PDTREE (available in the PDAP package; Garland et al. 1993), employing log10-transformed phenotypic values to improve normality, except for sugar content which was square-root-transformed. Because we were interested in performing analyses employing a single phylogeny, we selected Nee’s arbitrary branch lengths transformed with Grafen’s rho of 0.5 (see Results), which was statistically adequate for all studied phenotypic variables.

To investigate whether fruit morphology, chemistry and colour of the plant species in Ilha do Cardoso show significant phylogenetic signal, and whether this signal differs between these suites of characters, we employed the randomization procedure of the matlab program Physig.M developed by Blomberg, Garland & Ives (2003). This method consists in shuffling tip values across the real phylogeny and obtaining the distribution of variance in contrasts under this null model. If the variance of the real data set is lower than 5% of the values obtained from the randomizations (1000 in our study), we rejected the null hypothesis of no signal with < 0.05. We also compared the k-statistic obtained in each analysis, which describes the amount of phylogenetic signal in the continuous-valued traits and allows for comparisons across phylogenetic trees and phenotypic traits (Blomberg, Garland & Ives 2003). A value of 1 indicates exactly the amount expected under Brownian motion evolution along the specified phylogeny (topology and branch lengths). Values smaller than 1 indicate less phenotypic resemblance between relatives than expected and larger than 1 indicate stronger than expected phylogenetic clumping of trait values.

We performed Pearson’s correlations to test for pairwise associations between visual cues and estimates of fruit quality and defence to document the covariance among phenotypic fruit traits. Importantly, animals may combine a set of indicators rather than base their decisions during foraging on pairwise correlations only. Multivariate analyses may thus be more adequate. Therefore, we analysed the relation between macronutrients and phenols employing a model comparison approach. In this context, we were primarily interested in determining the association between fruit morphology, colour and nutrients, and we evaluated whether visual cues such as colour and morphology are good indicators of fruit quality, as judged by its chemical composition. While regular correlation and regression analyses are adequate to test whether dispersers can rely on visual cues to select better fruits, phylogenetically informed tests are generally more adequate to address if traits at different levels of organization have coevolved in response to selection.

Subsequently, we used the matlab program RegressionV2.M (Lavin et al. 2008) to perform linear statistical models via both ordinary least squares (OLS) (i.e. nonphylogenetic) and phylogenetic generalized least square regressions (PGLS) (Garland & Ives 2000; Garland, Bennett & Rezende 2005). Ordinary least square regressions assume that the unexplained residual variation is independent among species, while PGLS assumes that residual variation among species is correlated, with the correlation given by the degree of phylogenetic relatedness between species. We used both approaches to understand whether relationships among fruit traits are evolutionary conserved (and thus better explained by PGLS models) or evolutionary labile (and thus better explained by OLS models).

Regression models included chemical compounds (lipids, sugar, proteins and phenols) as the dependent variables and fruit morphology and colour as independent ones. Because morphological traits were highly correlated, we first performed a diagnostic principal component analysis and selected the variables that explained most of the variance in our data based on the score values (data not shown). Therefore, the morphological variables we selected were fruit diameter, seed number and relative yield. We included these variables even though the relative yield of fruits or the number of seeds is not involved in signalling. However, fruit consumers may learn that some fruits are relatively unrewarding and thereby use colour or other external stimuli (fruit size and shape) to avoid these fruits. The colour variables included in the model were brightness, hue, chroma and fruits’ chromatic contrasts as seen by birds.

We compared models including morphology or colour as independent variables employing the Akaike Information Criterion (AICc, which corrects for small sample sizes; Turkheimer, Hinz & Cunningham 2003), using the smaller-is-better formulation. When comparing a series of models, the one with the lowest AICc is considered to be the best. We also calculated the Akaike weights (wi) that are particularly intuitive because they may be expressed in terms of percentage and can be interpreted as ‘weight of evidence’ in favour of the model with the lowest AICc when compared with all alternative models (Turkheimer, Hinz & Cunningham 2003). Comparisons between models allowed us to determine (i) which visual cues can be employed as accurate indicators of fruit quality and (ii) whether the inclusion of phylogenetic information significantly improves the fitness of the tested regression models. We did not employ Bonferroni corrections in this analysis because this approach can be very conservative in the context of model comparison (Burnham & Anderson 2002). Instead, we opted to report relations that were significant with a < 0.05 weighed by the evidence provided by different models. That is, if a given relationship is statistically significant in more than one model, the total weight of the evidence favouring a significant relationship corresponds to the Akaike weights wi summed across models.

Results

Phylogenetic signal

Our working phylogeny, based on the diagnostic test suggested by Garland, Harvey & Ives (1992), presented little hierarchical structure. The branch lengths that showed better adequacy for all traits were Nee’s arbitrary branch lengths, subsequently transformed with Grafen’s rho of 0.5 (Fig. 1). Randomization tests suggest that most morphological and chemical traits exhibit significant phylogenetic signal, except for pulp fresh mass and sugar content (Table 1). Conversely, we did not detect phylogenetic signal in colour traits and phenols. The k-statistics revealed that the amount of signal is lower than expected by Brownian motion for all studied traits (Table 1). This result, concomitantly with the low degree of hierarchy of our working phylogeny (i.e. where most of the phenotypic variation is explained by recent history, and not by shared evolutionary history deeper in the phylogeny), suggests that phylogenetic signal in fruit characteristics in the plant community of Ilha do Cardoso is generally weak, in spite of being significant in many instances (Table 1).

Figure 1.

 Phylogenetic hypothesis employed to analyse the vertebrate-dispersed species at Cardoso Island, Atlantic rainforest (Brazil), including some representative plant orders belonging to this community. Randomization analyses were employed to test whether fruit morphology, chemical composition and colour showed significant phylogenetic signal, which is illustrated by the clustering of values below (black bars) and above (white bars) the median, shown by the triangle at the bottom of the panels. Generalized linear models were subsequently employed to test whether visuals cues were accurate indicators of fruit quality. The species and the data sets analysed are provided in the Supporting Information.

Table 1.   Range of phenotypic values observed for fruits from the community of Ilha do Cardoso and phylogenetic effects
 Phenotypic traitObserved rangenPearson rkP
  1. The number of species (n), results from diagnostic tests of branch length adequacy for Nee’s branch lengths with 0.5 rho transformation (Pearson’s r), the amount of phylogenetic signal (k) and its significance according to the randomization tests are listed for each phenotypic trait. Values in boldface are significant at P < 0.05.

MorphologyFruit diameter (mm)3.8–35.41050.11340.75110.044
Fruit length (mm)3.7–43.91050.06530.77980.017
Fruit fresh mass (g)0.02–23.411050.23610.77020.027
Pulp fresh mass (g)0.01–19.991050.28770.73580.076
Pulp dry mass (g)0.001–8.091051.84120.8643<0.001
Seed diameter (mm)0.4–27.91050.36500.8963<0.001
Seed length (mm)0.6–40.11050.42510.87320.001
Seed number (#)1–c. 5401050.50630.9214<0.001
Seed mass (g)0.002–13.841050.33910.82220.004
Relative yield0.003–0.4931050.00170.80650.003
Pulp/seed ratio0.04–4401050.06330.82750.005
Chemical compositionLipids (%)0.29–83.3781.55390.90960.001
Proteins (%)2.08–18.30780.87040.83910.023
Sugar (%)<0.001–3.76780.05890.78310.077
Phenols (%)0.50–8.22780.27440.76680.074
ColourBrightness (QT)0.77–3118.6580.50570.79980.082
Chroma0.22–16.42580.00480.75330.270
Hue (nm)335.0–695.0580.16480.71870.419
Chromatic vision (jnd)5.40–155.03580.02190.76030.181

Visual cues and chemical composition

Results from correlation tests and regression models designed to determine whether fruit morphology and colour could be employed as reliable indicators of fruit composition and quality varied depending on the nutrients analysed. Pairwise correlations between visual cues and chemical composition suggested that chroma may be employed to assess fruit sugar content (two-tailed < 0.05). Pairwise correlations also indicated that the relative yield of fruits was positively correlated with lipids and negatively correlated with proteins (Table 2), which means that fruits with high lipid values contain a large amount of pulp, while the opposite is true for protein-rich fruits. Multivariate analyses were used because animals may combine a set of indicators during foraging. Also, if seed dispersers tend to choose fruits based on their phenotypic attributes, their choice may also show a degree of phylogenetic structure (if they choose closely related species based on phylogenetically conserved cues).

Table 2.   Pairwise regular Pearson’s correlations between fruit characteristics and chemical composition
 Fruit sizeSeed numberRelative yieldBrightnessChromaHueChromatic contrast
  1. Correlations performed between log10-transformed data, except for sugar that was square-root-transformed to meet normality.

  2. *Two-tailed < 0.1, **two-tailed < 0.05.

Lipidsr76 = −0.138r76 = −0.155r76 = 0.224r45 = 0.167r45 = 0.220r45 = 0.145r45 = 0.145
P = 0.228P = 0.175P = 0.048**P = 0.262P = 0.137P = 0.330P = 0.330
Proteinsr76 = 0.166r76 = −0.006r76 = −0.223r45 = 0.247r45 = −0.041r45 = −0.080r45 = 0.125
P = 0.146P = 0.959P = 0.049**P = 0.093*P = 0.783P = 0.593P = 0.402
Sugarsr76 = 0.0025r76 = −0.0084r76 = −0.074r45 = −0.086r45 = −0.296r45 = 0.039r45 = −0.110
P = 0.982P = 0.942P = 0.516P = 0.566P = 0.043**P = 0.794P = 0.460
Phenolsr76 = −0.035r76 = −0.130r76 = 0.196r45 = −0.172r45 = 0.035r45 = 0.246r45 = −0.029
P = 0.764P = 0.255P = 0.085*P = 0.248P = 0.813P = 0.095*P = 0.846

Overall, regression models including fruit colour, but not fruit morphology, were the best predictors of sugar concentration, protein content and concentration of phenols, as judged by lower AICc values and Akaike weights wi that were nearly three times larger than models including morphological traits (Table 3). Thus, two variables employed to characterize fruit colour (brightness and chroma) were indicative of different aspects underlying fruit composition in these models, while fruit hue and the chromatic contrast perceived by birds did not indicate fruit composition. Brightness was associated with protein content with a < 0.05 in the two best models that, taken together, support a significant association between these variables with a total Akaike weight of 71.7 (Table 3). The positive relation between fruit brightness and protein content indicates that bright fruits (e.g. yellow and white ones) in this community apparently contain more protein than darker fruits (Fig. 2). Similarly, the Akaike weight of 64.47 indicated that the best model of sugar content is negatively associated with the chroma of fruits, meaning that fruits with low saturation, such as black fruits, have higher sugar contents than strongly saturated fruits (Fig. 2). The discrepancy between the Akaike weights of the OLS and PGLS models on sugar content is consistent with a lack of phylogenetic structure on sugar contents (see Table 1).

Table 3.   Results of regression models testing for the relation between fruit morphology, colour and chemical composition
 ModelRegressionAICcF statistic
wiFruit sizeSeed numberRelative yieldBrightnessChromaHueChromatic contrast
  1. Results from ordinary and phylogenetic generalized least square regressions (OLS and PGLS, respectively). Significant results (< 0.05) are highlighted in boldface. AICc is the Akaike Information Criterion corrected for small sample sizes and wi is the Akaike weight. All models with a low likelihood of being the true model (wi values <0.2) were nonsignificant, and F values were omitted.

LipidsMorphologyOLS77.4112.99       
PGLS74.7349.60F1,43 = 1.654F1,43 = 0.078F1,43 = 1.540    
ColourOLS77.1215.01       
PGLS76.3222.40   F1,42 = 2.691F1,42 = 3.674F1,42 = 0.030F1,42 = 0.547
ProteinsMorphologyOLS−12.5813.34       
PGLS−12.8114.96       
ColourOLS−14.1328.95   F1,42 = 4.425F1,42 = 0.004F1,42 = 0.300F1,42 = 2.703
PGLS−14.9142.75   F1,42 = 4.223F1,42 = 0.123F1,42 = 0.103F1,42 = 2.777
SugarsMorphologyOLS93.6919.32       
PGLS94.9210.45       
ColourOLS91.2864.47   F1,42 = 2.468F1,42 = 4.651F1,42 = 0.027F1,42 = 0.900
PGLS96.115.76       
PhenolsMorphologyOLS7.5224.72F1,43 = 0.020F1,43 = 0.635F1,43 = 1.250    
PGLS9.967.30       
ColourOLS6.2347.12   F1,42 = 2.514F1,42 = 0.009F1,42 = 3.592F1,42 = 0.061
PGLS7.8620.86   F1,42 = 3.720F1,42 = 0.016F1,42 = 2.423F1,42 = 0.700
Figure 2.

 Association between fruit colour parameters and chemical composition. The grey symbols represent the arithmetic average of fruits falling below the 10th quantile and above the 90th quantile, to illustrate the mean difference in chemical composition between fruits in the extremes of the colour distribution. Raw values are plotted, even though statistical analyses were performed with transformed data to meet the criterion of normality (results are listed in Tables 2 and 3).

Conversely, the best model explaining the variation in lipid fruit content across species included morphological traits but not colour. However, none of these variables was statistically significant in this model (Table 3). These results suggest that interspecific variation in lipid composition cannot be directly assessed from morphological traits or differences in fruit colour.

Comparisons between OLS and PGLS results suggest that inclusion of phylogenetic information improved the models (i.e. resulted in lower AICc values) for regressions with lipid and protein content as dependent variables but not for sugar and phenol concentration (Table 3). These results are qualitatively similar to those obtained from randomization analyses, which suggested that phylogenetic signal is significant for lipid and protein, but not for sugar and phenol, concentrations (Table 1).

Discussion

Our statistical approach incorporating phylogenetic information shows that fruit chroma and fruit brightness can reveal reliable information on the sugar and protein contents of fleshy fruits, respectively. Thus, seed dispersers may be able to evaluate sugar content and possibly also protein content based upon the total brightness and chroma of fruit colours. This covariation among visual and chemical plant traits may reflect the outcome of natural selection favouring particular combinations of traits over others. This explanation is particularly likely because phylogenetic relationships appear as an important constraint for morphological fruit traits but not for colour traits. In contrast to the covariation between colour and nutritional traits, interspecific variation in any chemical content cannot be directly assessed from morphological traits. To a large extent, morphological traits show independent patterns of covariation relative to nutritional traits. This result supports the finding by Jordano (1995) and by Valido, Schaefer & Jordano (2011) who also found a marked decoupling of fruit morphology and fruit nutrients.

Phylogenetic signal in fruit characteristics

We observed significant phylogenetic signal in most morphological and nutritional traits, except for pulp fresh mass, carbohydrates and phenols. Significant signal was expected because morphological traits tend to be more conserved across the phylogeny, and error is generally low in morphological measurements. Even though the probability of detecting a signal depends on the number of species and on the K values, Blomberg, Garland & Ives (2003) found that 90% of the evaluated traits with sample sizes >20 showed statistically significant phylogenetic signal. Conversely, none of the parameters describing fruit colour showed significant signal. Although this may reflect a decrease in statistical power due to a lower sample size or intra-individual variation (e.g. in ripeness and thus colour), it is also possible that fruit colour is a strong target of selection and consequently more evolutionarily labile.

As a corollary, the lack of significant phylogenetic signal is per se not evident that fruit coloration is an adaptation to frugivores. It rather shows that phylogenetic constraints on fruit colours are weak compared to most other fruit traits as another broad interspecific comparison had shown previously (Voigt et al. 2004). Supporting fruit coloration as adaptation requires demonstrating that it evolved in response to established selection regimes by frugivores. Two recent studies suggested the importance of frugivores to fruit evolution finding that fruit colours differ among plant species dispersed by primates or bats and those dispersed by birds independently of phylogeny (Lomáscolo & Schaefer 2010; Lomáscolo et al. 2010).

The presence of significant phylogenetic signal in most traits illustrates the importance of incorporating phylogenetic information in statistical analyses. Models on protein and lipid that incorporated phylogenetic relationships (PGLS) yielded consequently higher Akaike weights than OLS models, while the reverse was true for sugars and phenols. This result indicates that the relationship between protein and brightness is evolutionary stable, while the relationship between chroma and sugar is evolutionary labile. These findings stress the importance of applying a phylogenetically based statistical method to document the distinct evolutionary history of trait associations.

Visual cues and fruit composition

Our study shows that chroma and to some extent fruit brightness can be indicative of sugar and protein contents, respectively. It needs to be noted, however, that the relationship between fruit brightness and proteins in pairwise correlation analyses was marginally significant at best, and that the possibility that fruit brightness indicates protein content should be therefore be interpreted cautiously and investigated in the future. This is particularly important because the relative importance of fruit brightness for detection under natural conditions is poorly known. In particular, a background of vegetation usually varies 3000-fold in brightness (Sumner & Mollon 2000), and this natural variation may make it more difficult for fruit consumers to detect fruit targets using brightness only.

High sugar contents in fruits are indicated by low chroma. We suggest that this relationship is best understood by considering the underlying biochemical mechanisms. Fruits with low chroma values are mainly achromatic, being predominantly black or purple black in our sample. These colours are imparted by high concentrations of anthocyanins, the predominant fruit pigments that often override the coloration imparted by other pigments such as carotenoids (Schaefer, McGraw & Catoni 2008). Importantly, sugars strongly up-regulate the anthocyanin biosynthetic pathway (Solfanelli et al. 2006) because the gene expression of key enzymes of that pathway, such as chalcone synthase, is directly induced by sugars (Takeuchi, Matsumoto & Hayatsu 1994). As such, the correlation between high sugar content and high anthocyanin content in fruits (resulting in unsaturated colours) is likely to be explicable because carbohydrates directly induce anthocyanin biosynthesis. If this conjecture is correct, we expect the negative correlation between chroma and sugar content in fruits to hold more widely across distinct vegetation communities. In this scenario, the covariance between chroma and sugar content is explicable by biochemistry and does not necessarily represent a signal that evolved this covariance to communicate.

In general, traits that indicate content will be especially effective in stimulating fruit removal if they signal rare compounds. Although our study showed that fruit brightness is not strongly correlated with protein content, it might be advantageous for plants to indicate high protein content as this is generally limited in fruits (Herrera 1987). A large proportion of fruits with high brightness values were orange or yellow in our study, corroborating the relationship between protein content and these fruit colours that Schaefer & Schmidt (2004) found in a Venezuelan rainforest. This is particularly noteworthy because Endler (1993) suggested that in the prevailing ambient light within forests yellow and yellow–green fruits maximize brightness.

Is the reliability of fruit colours in indicating fruit nutrients adaptive? In general, because multiple perceivers can react to fruit coloration, the design of a signal should present a trade-off between conspicuousness to intended perceivers such as seed dispersers and relative inconspicuousness to unintended perceivers such as visually oriented fruit and seed predators (see Endler 2000). According to the hypothesis of adaptive plant communication, we could thus predict that visually oriented fruit and seed predators should respond more weakly to variation in sugar content than to variation in contents that are not signalled (e.g. lipids). If this prediction is not supported, by-product information may seem a more plausible scenario. The hypothesis on by-product information predicts that information arises if it is not selected for by animals that perceive plant traits. For example, information could evolve as by-product of pleiotropy, correlational selection or selection imposed by biotic agents such fungi (Schaefer & Ruxton 2011).

The current evidence in favour of the hypothesis of adaptive communication or by-product information is mixed. In contrast to the expectations of adaptive fruit communication, phenolic content is not indicated by fruit coloration even though phenols increase fruit longevity in our community by deterring fruit predators (Cazetta, Schaefer & Galetti 2008). This could be explicable because seed dispersers are likewise deterred by phenols in fruit pulp (Schaefer, Schmidt & Winkler 2003) or because many of the fruit predators such as fungi or microbes are not visually oriented. Similarly, both seed dispersers and seed predators preferentially consume lipid-rich fruits (Schaefer, Schmidt & Winkler 2003; Cazetta, Schaefer & Galetti 2008), but lipid content is not indicated by fruit coloration. We thus emphasise that the determination of the relative selective pressures of seed dispersers and fruit and seed predators on specific fruit compounds is necessary to distinguish between adaptive and nonadaptive scenarios of fruit communication.

In sum, we suggest that shared biochemistry is a likely mechanism inducing the covariance between colour (the stimulus) and nutritional quality (as the unseen quality). By-product information represents a simpler mechanism for the reliability of fruit colours than adaptation to signalling and should therefore serve as a null hypothesis in plant–animal communication. Even though the information represented by fruit colours may not have evolved to inform seed dispersers on nutritional contents, fruit colours might still evolve as signals that enhance the detectability of fruit crops and might thereby ultimately increase plant fitness. Likewise, it remains to be tested whether plants can increase their reproductive success by transmitting informative signals. Finally, the concept of adaptive fruit communication should be tested rigorously against by-product information.

Acknowledgments

We thank Cláudio Bernardo and Vanessa G. Staggemeier for help with field work, Dr. José Roberto Trigo for laboratory facilities and help with chemical analyses. This work received financial support from FAPESP (Proc. 05/52726-9). E.C. thanks FAPESP (Proc. 03/08447-2) and M.G. receives a fellowship from CNPq. E.C. and H.M.S. received a DAAD fellowship during this project. E.L.R. is a Ramón y Cajal fellow supported by the MICINN (Spain).

Ancillary