Introduction
The coexistence of species and the compositions of communities can be interpreted under different ecological aspects based on different theoretical points of view, such as niche conservatism (Wiens and Graham Reference Wiens and Graham2005) and the neutral theory of biodiversity (Hubbell Reference Hubbell2001), in addition to classical niche concepts such as that of Hutchinson (Reference Hutchinson1957). In turn, competition theory predicts that community structure can be shaped by partitioning resources between coexisting species (Geange et al. Reference Geange, Pledger, Burns and Shima2011). These theoretical developments demonstrate that aspects of resource partitioning and the way in which species exploit them still remain central and relevant themes for understanding the structures of animal communities (e.g., Maynard et al. Reference Maynard, Ananda, Sides, Burk and Whitehead2019, Mohd-Azlan et al. Reference Mohd-Azlan, Noske and Lawes2014, Naoki Reference Naoki2007).
Research evaluating resource partitioning in vertebrates often investigates habitat use, diet and activity period as parameters to explain why certain species occur in syntopy (e.g., Munin et al. Reference Munin, Fischer and Gonçalves2012, Riegert et al. Reference Riegert, Fainová, Antczak, Sedláček, Hořák, Reif and Pešata2011, Winck et al. Reference Winck, Hatano, Vrcibradic, Van Sluys and Rocha2016). Some studies also consider foraging behaviour to analyse niche overlap (e.g., Kent and Sherry Reference Kent and Sherry2020, Oyugi et al. Reference Oyugi, Brown and Whelan2012) because differences in the way of consuming fruits, capturing mobile prey or using the foraging substrate may be related to morphological and behavioural differences that influence spatial occupation by consumers and establish limits on the exploitation of these resources for each species (Moermond and Denslow Reference Moermond and Denslow1983, Pearson Reference Pearson1977, Whelan Reference Whelan2001). Thus, quantifying species overlap in relation to these parameters is a key component in revealing how they interact and coexist and how communities are structured (Geange et al. Reference Geange, Pledger, Burns and Shima2011).
Many bird species consume a wide spectrum of food resources, especially fruits and arthropods, which may require different behaviours to obtain. When feeding on fruits, for example, the behaviour to access them depends on the way the fruits are arranged on branches (Denslow et al. Reference Denslow, Moermond, Levey, Estrada and Fleming1986, Stanley and Lill Reference Stanley and Lill2001). The nature of bird–fruit relationships ranges from generalisation to high specialisation (Dalsgaard et al. Reference Dalsgaard, Schleuning, Maruyama, Dehling, Sonne, Vizentin-Bugoni, Zanata, Fjeldså, Böhning-Gaese and Rahbek2017, Malanotte et al. Reference Malanotte, Machado-de-Souza, Campos, Petkowicz and Varassin2019), but mutualistic bird × plant specialisation in frugivory seems to be lower in tropical areas than in temperate environments (Schleuning et al. Reference Schleuning, Fründ, Klein, Abrahamczyk, Alarcón, Albrecht, Andersson, Bazarian, Böhning-Gaese, Bommarco, Dalsgaard, Dehling, Gotlieb, Hagen, Hickler, Holzschuh, Kaiser-Bunbury, Kreft, Morris, Sandel, Sutherland, Svenning, Tscharntke, Watts, Weiner, Werner, Williams, Winqvist, Dormann and Blüthgen2012). On the other hand, when hunting arthropods, birds face the challenge of subduing prey that lives on a variety of substrates provided by the environment, using them for camouflage, shelter or foraging (Robinson and Holmes Reference Robinson and Holmes1982, Whelan Reference Whelan2001). Finally, seasonal variations may lead different bird species to utilise food resources in different ways over time, probably facilitating their coexistence (Manhães Reference Manhães2003a).
Foraging behaviours may be more similar between closely related species than between more distant species (Brumfield et al. Reference Brumfield, Tello, Cheviron, Carling, Crochet and Rosenberg2007), although some taxonomically more distant species show behavioural convergence (Mansor and Mohd Sah Reference Mansor and Mohd Sah2012, Robinson and Holmes Reference Robinson and Holmes1982). In taxonomic terms, congeneric species presumably have more behavioural similarities than those more distantly related (Naoki Reference Naoki2007), while species within a family sometimes show large differences between their foraging behaviours (Chapman and Rosenberg Reference Chapman and Rosenberg1991). In any case, the diversity and variability in the behaviours of ecologically or phylogenetically restricted groups of birds must be smaller than the diversity of food items included in the diet of these birds, especially when considering ornithochoric plants from hyper-diverse neotropical botanical families, such as Melastomataceae and Rubiaceae. On the other hand, the same may not occur regarding the capture of arthropods on different types of substrates, as these are not as diverse as different ornithochoric fruits, but such comparisons are unexplored topics in bird behaviour.
This study analysed the trophic ecology of Passeriformes from the Thraupidae family, a taxon with several representatives from forests and open areas such as cerrado and rocky outcrops (Ridgely and Tudor Reference Ridgely and Tudor1994), where they consume a wide spectrum of food items, although predominately fruits and arthropods (Rodrigues Reference Rodrigues1995, Snow and Snow Reference Snow and Snow1971). The objective was to investigate the foraging behaviour of Thraupidae and analyse the niche relationships between the species studied. I proposed two lines of analysis: 1) comparing the similarities of bird foraging behaviours with those of the frugivorous diet and use of substrates for foraging on arthropods; and 2) testing the fruit–frugivore network specialisation, the similarity of the frugivorous diet, and the similarity of substrate use against null models. In the first case, I expected to find that a) birds show more similarities in foraging behaviours than in frugivorous diet composition, and b) the similarities of foraging behaviours for capturing arthropods do not differ from the similarity among the categories of substrates used to capture those prey. In the second case, I expected a) fruit–frugivore network specialisation similar to that expected by chance because the birds are generalist species and despite the possible minor similarities in frugivore diets in relation to behaviours; and b) due to the expected absence of specialisation in frugivory and probable limitation of substrates as a resource in the environment, the similarities of frugivorous diet and substrate use among tanagers are equal to that expected by chance.
Materials and methods
Study area
The study was carried out at Ibitipoca State Park. The park covers an area of 1488 ha, between coordinates 21° 40’ to 21° 44’ S and 43° 52’ to 43° 55’ W, in the state of Minas Gerais, south-eastern Brazil (Menini Neto and Salimena Reference Menini Neto, Salimena, Forzza, Menini Neto, Salimena and Zappi2013). Altitudes vary from 1050 m to 1784 m (Rodela Reference Rodela1998), and the climate is classified as Koppen’s Cwb with dry winters and rainy summers (Carvalho et al. Reference Carvalho, Fontes and Oliveira-Filho2000). The landscape is heterogeneous but dominated by rocky outcrops, grasslands and riparian forests (Carvalho et al. Reference Carvalho, Fontes and Oliveira-Filho2000, Dias et al. Reference Dias, Fernandes Filho, Schaefer, Fontes and Ventorim2002).
Foraging behaviour sampling
Data were collected over periods of 5–8 days per month, from March 1999 to February 2000, totalling 84 days and 840 hours of observations. A main transect and some secondary trails, totalling around 4,000 m, were walked randomly every day from 6:00 h to 12:00 h and 14:00 h to 18:00 h, walking slowly until finding one or more individuals whose foraging behaviour could be recorded. The covered area was divided into two sectors, the first covering the initial 840 m of the main transect and the second the remaining 1440 m, both with secondary trails. On one sampling day, one of the sectors was travelled in the morning and the other in the afternoon, alternating on the following day, seeking to adequately cover different sections of the area with equal frequency, as proposed by Eckhardt (Reference Eckhardt1979). The transects ran through open shrubby savannas and grassland areas or denser vegetation of cloud or dwarf forest areas (Oliveira-Filho et al. Reference Oliveira-Filho, Fontes, Viana, Valente, Salimena, Ferreira, Forzza, Menini Neto, Salimena and Zappi2013).
I recorded the first movement of a solitary bird observed foraging or the first bird observed in a monospecific flock or pair (Hejl and Verner Reference Hejl and Verner1990) feeding on the same item (e.g., fruits, insect swarms). Occasionally, some individuals were seen feeding repeatedly on the same item, and in these circumstances, a new record was only considered after a 5-minute interval. If a bird moved to forage on another plant or consumed another type of food item, a new foraging movement was recorded regardless of the time elapsed (Snow and Snow Reference Snow and Snow1971) so as not to overestimate one item in relation to another. Similar approaches were also adopted by Blendinger et al. (Reference Blendinger, Ruggera, Montellano, Macchi, Zelaya, Álvarez, Martín, Acosta, Sánchez and Haedo2012) and Silva and Pedroni (Reference Silva and Pedroni2014). Different individuals feeding on different items were also considered distinct foraging bouts. It was very unusual to obtain more than one foraging bout in an observation from a solitary individual or flock, so possible data dependency due to sequential observations was minimised (see Craig Reference Craig1990). Observations were carried out with binoculars or the naked eye and were recorded on a portable recorder for later transcription. Foraging behaviours were described according to Remsen and Robinson (Reference Remsen and Robinson1990). Eleven tanager species were found during the study. However, the analyses were carried out for the six most common bird species (Schistochlamys ruficapillus (Vieillot, 1817), Tangara desmaresti (Vieillot, 1817), Stilpnia cayana (Linnaeus, 1766), Stephanophorus diadematus (Temminck, 1823), Thraupis sayaca (Linnaeus, 1766) and Dacnis cayana (Linnaeus, 1766)) because they comprised more than 80% of the observations.
Metrics and data analyses
Similarity was calculated with the Horn-Morisita index (Horn Reference Horn1966):
$$simijk = {{2\mathop \sum \nolimits_i {x_{ij}}{x_{ik}}}\over{{((\mathop \sum \nolimits_i x_{ij}^2\;/{{(\mathop \sum \nolimits_i {x_{ij}}\;)}^2} + \mathop \sum \nolimits_i x_{ik}^2\;/\left( {\mathop \sum \nolimits_i {x_{ik}}\;{)^2}\;} \right)\mathop \sum \nolimits_i {x_{ij\;}}\mathop \sum \nolimits_i {x_{ik}}}}}$$
where simijk is the similarity between species j and k in resource use or behaviour, and xij and xik refer to the proportion of item i used by species j and k, respectively. This formula was used to evaluate the similarities of foraging behaviours, species of fruit consumed and substrate for capturing arthropods. The Horn-Morisita index ranges from 0 to 1, where 0 indicates no similarity in resource use and 1 means total similarity (Fernandes et al. Reference Fernandes, Dáttilo, Silva, Luna, Braz and Morini2020). Substrates for capturing arthropods were grouped into ‘air’, ‘ground’, ‘leaf’, ‘flower’, ‘bare branches’, ‘covered branches’ (mosses/lichens) and ‘trunk’. Vanillosmopsis erythropappa (DC.) Sch. Bip. (Asteraceae) is a typical and abundant plant species in the study area; therefore, I considered its inflorescences as a separate substrate from other flowers (‘Vinfl’). I did not calculate diet similarity based on arthropod taxa because the number of prey identified, even at the order level, was very low. The similarity in the frugivorous diet was based on the weighted bird–plant matrix (frequency of interactions), with 50 of the 52 fruit species consumed by Thraupidae in Ibitipoca, whose diet was largely dominated by plants from the families Melastomataceae, Myrtaceae and Cecropia glaziovii Snethl. (currently Urticaceae). Vitex montevidensis Cham. (currently Lamiaceae) and Guatteria nigrescens Mart. (Annonaceae) were excluded because behaviours to consume these fruits were not recorded. However, there was only one record of consumption for each of these species (see details in Manhães Reference Manhães2003a). I was interested in the ecological aspects of the interactions, and for simplification, I kept the nomenclature described in Manhães (Reference Manhães2003a) for plant species despite some possible taxonomic revisions. The Mann–Whitney permutation test was applied to the observed values of Horn–Morisita similarity per pair of species to test the differences between foraging behaviours in fruit consumption and the composition of the frugivorous diet, as well as between behaviours to capture arthropods and the use of the substrate where they were captured. The conditional null distribution to the test was obtained by 10,000 Monte Carlo resamples.
The complementary specialisation index (H 2’) was calculated to characterise the degree of specialisation or partitioning of the entire fruit–frugivore network. The H 2’ metric can be calculated for weighted matrices and is independent of the network size (Blüthgen et al. Reference Blüthgen, Menzel and Blüthgen2006). For the ‘frugivorous portion’ of the feeding resources, I also calculated the specialisation asymmetry (SA) to evaluate trends towards higher or lower trophic levels. Positive values of SA indicate a higher specialisation of the higher trophic level (consumers), while negative values indicate a higher specialisation of the lower trophic level (resources) (Dormann et al. Reference Dormann, Fründ, Blüthgen and Gruber2009). For each tanager species, the d’ index (standardised Kullback-Leibler distance), the degree of interaction specialisation at the species level (Blüthgen et al. Reference Blüthgen, Menzel and Blüthgen2006), was obtained because SA is based on it. The d’ index is based on frequency data and is insensitive to the dimensions of the network; it ranges from 0 (no specialisation) to 1 (perfect specialist). To assess the significance of frugivorous diet/substrate use similarities and H 2’, the observed values of these metrics were compared with 95% confidence intervals calculated from 10,000 randomisations using the Vaznull and Patefield algorithms, including the standardised effect size (SES). The Vaznull algorithm, proposed by Vázquez et al. (Reference Vázquez, Melian, Williams, Blüthgen, Krasnov and Poulin2007), keeps the network connectance constant but changes the values of the cells; consequently, the marginal totals are different from those of the original matrices. On the other hand, Patefield maintains the marginal totals, but the connectance value is different from the original (Dormann et al. Reference Dormann, Fründ, Blüthgen and Gruber2009, Perez-Lamarque et al. Reference Perez-Lamarque, Petrolli, Strullu-Derrien, Strasberg, Morlon, Selosse and Martos2022). Significant observed values fall above or below the 95% confidence interval limits. The SES provides information about the magnitude and direction (positive/negative) of the difference between two groups (Durlak Reference Durlak2009) and was calculated as SES = (Iobs-Isim)/σsim following Gotelli and McCabe (Reference Gotelli and McCabe2002), where Iobs is the observed index and Isim and σsim are the mean and standard deviation of the 10,000 simulated indices, respectively.
The analyses were performed using the basic R packages (confidence intervals) (R Core Team 2023) and R packages coin (Hothorn et al. Reference Hothorn, Hornik, van de Wiel and Zeileis2008) (Mann–Whitney tests), bipartite (Dormann et al. Reference Dormann, Fründ, Blüthgen and Gruber2009) (SA, d’ index and H 2’) and vegan (Oksanen et al. Reference Oksanen, Simpson, Blanchet, Kindt, Legendre, Minchin, O’Hara, Solymos, Stevens, Szoecs, Wagner, Barbour, Bedward, Bolker, Borcard, Carvalho, Chirico, De Caceres, Durand, Evangelista, FitzJohn, Friendly, Furneaux, Hannigan, Hill, Lahti, McGlinn, Ouellette, Ribeiro Cunha, Smith, Stier, Ter Braak and Weedon2022) (Horn–Morisita index).
Results
Of the 1,229 foraging events for which behaviours were recorded, 749 were for fruit consumption and 243 for arthropod predation, totalling 992 events included in the analyses. The remaining 237 were for consumption of various items such as leaves, flowers, stamen, galls, Müllerian bodies, climbing stems, nectar and anthropogenic food wastes. The largest number of bouts observed was for S. ruficapillus (n = 336), followed by T. desmaresti (n = 241), which together corresponded to 58.1% of all records. The species with the lowest number of records was D. cayana (n = 31) (Table 1). Thraupidae species performed 17 foraging manoeuvres, of which 13 were near-perch manoeuvres and 4 were aerial manoeuvres. The most used manoeuvre was ‘glean’ (n = 291), followed by ‘reach down’ (n = 142) and ‘hang down’ (n = 134), which together correspond to 57.2% of foraging manoeuvres (Table 1). T. desmaresti was the species with the highest number of fruit species consumed (n = 29), followed by S. diadematus (n = 24). The most common plant species in the Thraupidae diet were C. glaziovii, Myrcia venulosa DC. (Myrtaceae) and Myrcia rostrata DC. (Myrtaceae) (n = 139, n = 59 and n = 58, respectively), corresponding to 34.2% of fruit foraging records. The most common substrates for capturing arthropods were leaves, air and bare branches (n = 68, n = 50 and n = 40, respectively), accounting for 65.0% of all substrates used (Table 2).
Table 1. Frequency of the foraging behaviours used by tanagers to feed on fruits (fr) and arthropods (art) in Ibitipoca. Classification scheme based on Remsen and Robinson (Reference Remsen and Robinson1990). Sr: Schistochlamys ruficapillus; Td: Tangara desmaresti; Sc: Stilpnia cayana; Sd: Stephanophorus diadematus; Ts: Thraupis sayaca; Dc: Dacnis cayana
Table 2. Substrates used by tanagers to prey on arthropods in Ibitipoca. Barebr: bare branches; Vinfl: Vanillosmopsis erythropappa inflorescences; covebr: covered branches (mosses/lichens)
The similarity in the frugivorous diet was lower than the behaviour for obtaining fruits (Z = 4.0856, P < 0.001), while there was no difference between the behaviours for foraging on arthropods and the substrates used to consume them (Z = 0.8503, P = 0.2071; Figure 1, Supplementary Material). Furthermore, the observed similarity (simijk = 0.228) of the frugivorous diet was lower than that expected by chance for the two algorithms considered, suggesting a partitioning of resources between the Thraupidae (Table 3). On the other hand, substrate utilisation (simijk = 0.620) did not differ from the null model for the Vaznull algorithm and was lower for the Patefield model. However, the SES of this last case was the lowest compared to the SES for frugivorous diet overlap and H 2’, which suggests that the differentiation of niches in substrate use is less relevant among the tanagers (Table 3). Niche similarity values for all factors were generally high, often above 0.500, except in the case of the frugivorous diet (Figure 1; Supplementary Information 1).
Figure 1. Comparisons between tanagers’ foraging behaviour x fruits in the diet and substrate to prey on arthropods. Box plots represent the 15 values of Horn–Morisita similarity index calculated by a pair of tanager species (Supplementary Information 1). Points are extreme values. NS: non-significant; *** P < 0.001 (Mann-Whitney permutation test after 10,000 Monte Carlo resamples).
Table 3. Values of observed trophic and network metrics compared to the 95% confidence limits from the two null models, including the standardised effect size (SES)
The value for bird–fruit complementary specialisation (H 2’ = 0.475) was much higher than expected by chance for both algorithms (Table 3), and both cases had high SES values. The positive value found for specialisation asymmetry (SA = 0.310) indicates that the greatest species specialisations in the network are higher at the highest trophic level; that is, among the Thraupidae. Tangara desmaresti, S. diadematus and D. cayana had the largest d’ indices (0.602, 0.550 and 0.526, respectively), probably due to interactions with less abundant plant species. The other three tanager species, T. sayaca (d’ = 0.472), S. ruficapillus (d’ = 0.466) and S. cayana (d’ = 0.297), concentrated their interactions on the most common plants, especially C. glaziovii (Figure 2; Supplementary Information 2), despite multiple realised interactions between all tanagers and fruits (Figure 2).
Figure 2. Interaction network between tanagers and plant species in the Ibitipoca State Park, including species-level index (d’). Grey links represent frequencies of fruit consumption (Supplementary Information 2). Sruf: Schistochlamys ruficapillus; Tdes: Tangara desmaresti; Scay: Stilpnia cayana; Sdia: Stephanophorus diadematus; Tsay: Thraupis sayaca; Dcay: Dacnis cayana.
Discussion
Fruit consumption
Large overlaps in resource use by birds are common in plant–frugivore interaction systems (Silva and Melo Reference Silva and Melo2013, Terborgh and Diamond Reference Terborgh and Diamond1970) because the generalist habits of frugivores may lead to functional redundancy in network interactions (Chama et al. Reference Chama, Berens, Downs and Farwi2013). Among tanagers, Tangara species are important generalist frugivores performing a large proportion of fruit–frugivore interactions in tropical forests (Quitián et al. Reference Quitián, Santillán, Espinosa, Homeier, Böhning-Gaese, Schleuning and Neuschulz2018), showing high feeding overlaps in the use of fleshy-fruit plants in these ecosystems (Buitrón–Jurado and Sanz Reference Buitrón-Jurado and Sanz2021). However, because low specificity seems to be common in bird–plant seed dispersal (Fuentes Reference Fuentes1995), and generalist birds, such as some tanagers, have low dependence on frugivory (Snow and Snow Reference Snow and Snow1971), competition should not be an important factor in bird coteries formation, as observed by Silva and Melo (Reference Silva and Melo2013). Thus, the low similarity values in frugivory observed for the tanagers in Ibitipoca, despite not being congeneric species, and higher than expected specialisation for the network as well for bird species, are remarkable.
Some behavioural skills probably influenced the reduction of similarities in the frugivorous diet among tanagers. For example, S. ruficapillus acted as a cut feeder (Foster Reference Foster1987) by separating the seed pulp from the relatively hard fruits of Byrsonima variabilis A. Juss. (Malpighiaceae) and consuming only the green pulp and was the only species to consume it. The behavioural response of reaching for a fruit must also be linked to its presentation to birds (Moermond and Denslow Reference Moermond and Denslow1985). Thus, fruits arranged as infructescences on peduncles of C. glaziovii or hanging from branches like the legumes of Senna bicapsularis L. Roxb. (Fabaceae) have always forced ‘hang’ movements, predominantly ‘hang down’, a movement widely used by Thraupis species, as was also observed by Silva (Reference Silva1980) and Gonçalves and Vitorino (Reference Gonçalves and Vitorino2014) in Cecropia spp., favouring the consumption of those two plant species by Thraupis sayaca in Ibitipoca. However, most fruit species in the diet of the Thraupidae produce small berries exposed at the ends of the branches, as occurs in Myrtaceae and Melastomataceae, the families most represented in foraging observations, making the use of some tactics repetitively, such as ‘glean’ or ‘reach’, favouring higher values of behavioural similarity. I also observed aggressive encounters among tanagers during the consumption of C. glaziovii and, sometimes, aggressive encounters also occurred between S. ruficapillus and S. diadematus when consuming Periandra mediterranea (Vell.) Taub. (Fabaceae) flowers, despite being a locally abundant resource. The dietary breadth of tanagers may have allowed birds to coexist in the environment by using alternative food when individuals of a particular species were chased away during agonistic encounters while foraging on some resources.
The switching behaviour hypothesis predicts that species will seasonally switch the proportion of alternative resources in the diet, such as fruits and invertebrates, due to seasonal availability, and changes in the proportion of fruits in the diet will affect some interaction network properties (Carnicer et al. Reference Carnicer, Jordano and Melián2009). Based on the same data used in this study, I recorded seasonal variation in the diet composition of S. ruficapillus, which consumed more fruits with increasing abundance, but not for T. desmaresti, which maintained higher fruit consumption throughout the year (Manhães Reference Manhães2003b). In a montane forest in Kenya, Borghesio and Laiolo (Reference Borghesio and Laiolo2004) also observed omnivorous birds seasonally changing their feeding habits and consuming more fruits when this resource was more abundant, an apparently common strategy among omnivorous birds. In addition, the tanager species in Ibitipoca consumed other food items in addition to fruits and arthropods. For example, leaves and flowers were consumed in different proportions by some of the studied species (see Manhães Reference Manhães2003a), remarkably S. diadematus, S. ruficapillus and T. sayaca, for which the consumption of P. mediterranea flowers and C. glaziovii fruits corresponded to 71% of the foraging bouts. These and other alternative resources may have allowed the birds to partition fruit species.
A diversity of plant species in the diets of frugivorous birds may present some fruit traits that offer different feeding perspectives to birds. For example, fruit or seed size (Hazell et al. Reference Hazell, Sam, Sreekar, Yama, Koagouw, Stewart and Peck2023, Quintero et al. Reference Quintero, Pizo and Jordano2020), fruit presentation (Fleming and Kress Reference Fleming and Kress2013), colour variation (Wheelwright and Janson Reference Wheelwright and Janson1985), abundance (Ramos-Robles et al. Reference Ramos-Robles, Dáttilo, Díaz-Castelazo and Andresen2018) and energy/nutritional rewards (Howe Reference Howe1993, Sebastián-González Reference Sebastián-González2017) have all been shown to affect fruit consumption by birds and seed dispersal mutualism. Therefore, it is not surprising that a body of research has found higher than expected mutualistic network specialisations (e.g., Blüthgen et al. Reference Blüthgen, Menzel, Hovestadt, Fiala and Blüthgen2007, Crestani et al. Reference Crestani, Mello and Cazetta2019, Machado-de-Souza et al. Reference Machado-de-Souza, Campos, Devoto and Varassin2019), as I found here. Furthermore, the studied tanager species are known for spatial occupation according to vegetational type and density (Isler and Isler Reference Isler and Isler1987, Ridgely and Tudor Reference Ridgely and Tudor1994), and the trails sampled in Ibitipoca crossed a mosaic of open and forested physiognomies. Natural light conditions in particular site types may offer different possibilities for fruit detection (Cazetta et al. Reference Cazetta, Schaefer and Galetti2009), in addition to the occurrence of different plant species and their traits; therefore, those conditions possibly influenced the niche similarity and network specialisation of the tanagers in Ibitipoca, as well as species specialisation at a higher level (tanagers).
Use of substrate for arthropod consumption
Considering insectivorous birds, Moermond (Reference Moermond1990) argued that some species may differ substantially in the manoeuvres used to obtain prey among foliage, while Robinson and Holmes (Reference Robinson and Holmes1982) and Blancher and Robertson (Reference Blancher and Robertson1984) found few differences in the use of tactics to capture arthropods between sympatric species. For bird species with mixed diets, such as tanagers, Snow and Snow (Reference Snow and Snow1971) suggested important differences regarding arthropod predation techniques and substrate use. However, the tanagers in Ibitipoca consumed arthropods secondarily and should probably be less adapted in morphology and behaviours for insectivory than typical insectivores, possibly investing in prey with less anti-predator defences that are more conspicuous on flowers, leaves, branches, the ground or in the air.
Tanagers were rarely observed using more complex manoeuvres such as ‘gape’ or ‘flutter chase’ or looking for arthropods hidden in the mosses or lichens on branches, an abundant substrate in Ibitipoca. The tactics used to consume arthropods did not constitute a satisfactory parameter for separating the niches of the species, and frequently used movements such as ‘glean’, ‘reach down’ or ‘sally strike’ were used to consume arthropods in different substrates. The birds observed most in the open environment in Ibitipoca, such as S. ruficapillus and S. cayana, experienced other opportunities, such as being able to use the air as a substrate, with ‘sally’ movements. Even S. diadematus, which was found mainly in riparian forest areas, used this movement when in open areas, while T. desmaresti, common in closed forest environments, practically never used it. Other than this, no Ibitipoca tanager showed specific behavioural abilities to capture arthropods or use the substrate to do it, except on rare occasions, for example, the use of ‘gape’ by D. cayana. Thus, the behaviours (and possibly the morphologies) of tanagers appear to be sufficiently conservative and do not promote differences in obtaining arthropods as food resources, such as those observed in tropical insectivores (Lloyd Reference Lloyd2008, Rosenberg Reference Rosenberg1993).
In general, the results found here contrast with those found for the trophic ecology of Tangara species studied by Naoki (Reference Naoki2007) in Ecuador, who found a lack of segregation in fruit consumption and fine segregation in arthropod foraging. In contrast, Rodrigues (Reference Rodrigues1995) showed that the tanagers’ frugivorous diet was more important in reducing overlap between species than the use of substrate, as observed for tanagers in Ibitipoca. It is worth noting that the data from Rodrigues (Reference Rodrigues1995) were also based on birds from different genera, as well as those in the present study, and not just Tangara. Several ecological and behavioural factors seem to have been combined to determine the ways in which the Ibitipoca tanagers used fruits as food resources to reduce similarities in diet and reveal some degree of fruit–frugivore network specialisation. On the other hand, the analysis of predation behaviours indicates that, in general, the tanager species used similar tactics to reach arthropods randomly and opportunistically on different substrates. In general, frugivory seems to offer more relevant conditions in trophic partitioning than foraging on arthropods.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0266467425100072
Acknowledgements
I wish to thank IBAMA (Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renováveis) and IEF-MG (Instituto Estadual de Florestas do Estado de Minas Gerais) for licences and authorisations; CAPES for the MSc scholarship; Eduardo P. Brettas for the bird silhouettes in Figure 2; and Profs. Marco Aurélio Leite Fontes (UFLA), Júlio Lombardi (UNESP), Waldir Mantovani (UFC), Fátima R. G. Salimena (UFJF), Selma M. S. Verardo (UFJF) (in memoriam), Marco Antônio Batalha (UFSCar), Leonardo Dias Meireles (USP), and Maria de Fátima Freitas (Jardim Botânico – RJ) for identifying the plant species.
Financial support
This research received no specific grant from any funding agency or the commercial or not-for-profit sectors.
Competing interests
The author declares none.
