Difficulty of access into the forest canopy has hindered the study of canopy bird assemblages as well as attempts to unify concepts on their structure and organization. Our study takes several steps to make such an attempt possible. First, we describe for the first time a canopy bird assemblage from northern Middle America, thereby broadening our perspective of canopy birds in lowland neotropical rainforests. Second, we compare canopy bird assemblages of distant neotropical forests, on the basis of similar canopy-based censuses in Honduras and Brazil. These sites are particularly useful for such a comparison because they share a similar climate and forest structure yet offer relatively independent biogeographic histories, being separated by over 2000 km. Most importantly, we took three steps designed to clarify previously unresolved arguments on the composition of canopy bird assemblages: (1) by categorizing bird observations in Honduras into distinct strata we achieved a quantitative definition of the core canopy birds at that site; (2) we assessed the relative importance of dietary guilds and of edge species and migrants in Honduras and Brazil by comparing the observed composition of these groups with expectations based on species drawn randomly from each regional pool; and (3) by reclassifying data sets from Costa Rica and Panama by similar criteria, we are able to broadly characterize bird assemblages of lowland neotropical rainforest canopies.
Species Richness and Abundance
The species diversity of forest birds is notably higher in Amazonia than in Middle America at levels of both local (alpha) and regional (gamma) diversity (Karr et al. 1990, Terborgh et al. 1990, Blake 2007). Although we observed 27 more species in the canopy in Brazil than in Honduras, the overall richness of the canopy assemblages of the two sites did not differ significantly. Furthermore, if we include six species extirpated from the study site in Honduras (Ara macao, Amazona farinosa, A. autumnalis) or reduced below detectable levels by persecution (Spizaetus ornatus, S. tyrannus, S. melanoleucus), the difference between the sites in observed richness is 14%. Given that overall richness of forest birds is approximately 35% higher in Brazil, the similarity in species richness in the forest canopy is notable. The higher richness of migrants in Honduras, roughly three times that in Brazil, partially explains this finding. Especially notable was the greater richness of the family Parulidae, represented by 11 more species in Honduras than in Brazil, and the families Tyrannidae, Vireonidae, and Cardinalidae, which together contributed 15 species of migrants. Additionally, the number of hummingbird species observed in the canopy in Honduras was twice that observed in Brazil (14 vs. 7), despite equal species richness at the regional level. We suspect that in Honduras the high number of individual flowering trees in the canopy (principally Vochysia guatemalensis and Symphonia globulifera) may have attracted a greater diversity of hummingbird species into the canopy. During peak flowering, as many as eight individuals of seven species were observed in the canopy during a single 3-hr census, whereas in Brazil the median number of both species and individuals observed per census was one. This difference may be due to factors intrinsic to the Honduras site, because a similarly high richness of hummingbirds was not observed in Costa Rica (five species) or Panama (seven species).
Nonetheless, the additional migrants and hummingbirds in Honduras do not completely account for the similarity to Brazil in species richness. Clearly, in Brazil a smaller proportion of the regional forest species occurs in the canopy stratum. One plausible explanation is that differences in stratification result from the difference in taxonomic composition of the local avifaunas. At the Brazil site, the families Tinamidae, Cracidae, Furnariidae, Thamnophilidae, Formicariidae, Pipridae, and Troglodytidae, are all dominated by species of lower and middle forest strata, and richness of these families is more than double their richness in Honduras. In contrast, the few families with similar (Trochilidae, Vireonidae) or greater (Parulidae, Cardinalidae) species richness in Honduras are weighted with species of the upper forest strata, or with migrants, which seem to be more important locally in the canopy than at lower levels. Therefore, we suggest that taxonomic differences in the regional avifaunas explain in large part the similarity of richness in the canopy in Honduras and Brazil.
Despite this similarity in species richness, patterns of species abundance differed markedly. Our findings that the Middle American canopy was dominated by a few superabundant species and that species’ abundances were distributed more evenly in Amazonia are consistent with the findings reported by Robinson et al. (2000) for sites in Panama and Amazonian Peru. Robinson et al. described an “oligarchy” of eight common species, six from the understory and two from the canopy (Hylophilus decurtatus and Zimmerius vilissimus), that accounted for a disproportionate number (36%) of individuals at the Panama site. Likewise, in Honduras the seven most abundant canopy species each accounted for ≥4% of all detections and a combined 46% of all detections in the canopy, whereas in Brazil only a single species reached comparable abundance. Finally, the overall pattern of more rare species than common ones observed in both the Honduras and Brazil canopies mirrors results from other lowland sites in neotropical forest (Pearson 1977, Karr et al. 1990, Terborgh et al. 1990, Thiollay 1994b, Robinson et al. 2000).
Richness of Migrants
The importance of migrants in the canopies of lowland neotropical forests deserves special recognition. Even in Brazil, where the proportion of austral to nearctic migrants is greater, migrants are observed disproportionately more often in the canopy than in the understory (Bierregaard 1990, Stotz et al. 1992). Moreover, few migrant forest birds are true ground-dwellers, and the majority inhabits mid- to upper forest strata. Of 36 passeriform species that are long-distance migrants to the Honduras study site, only six occur principally on the ground or in the understory (Seiurus aurocapilla, Parkesia noveboracencis, P. motacilla, Oporornis formosus, Wilsonia citrina, Hylocichla mustelina). Therefore, the annual influx of migrants to tropical forests adds disproportionately to the midstory and canopy. Finally, the pool of migrant species available to colonize the canopy is substantially greater in Honduras than in Brazil (Kelly and Hutto 2005), thereby disproportionately weighting the Honduras canopy with this group.
Little consensus has been reached on the trophic organization of bird assemblages in the canopy of lowland neotropical rainforests. One confounding factor is that assignments to dietary guilds are not consistent among studies. In Honduras, omnivores and insectivores dominated the canopy in terms of species richness, whereas omnivores dominated in terms of numerical abundance. When we reclassified species’ assignments to dietary guilds from previous studies by similar criteria, and restricted analyses to true forest species by eliminating aerial foragers and scavengers, the pattern that emerged was similar at all four neotropical sites studied thus far. In Costa Rica, Panama, and Brazil, omnivores and insectivores were the most species-rich groups, with slightly higher richness in the omnivore guild for three of four forests. Similarly, omnivores predominated in numerical abundance at all sites. In the understory of lowland neotropical rainforests this patterns seems to be reversed. In Costa Rica species richness of insectivores was three times greater than that of omnivores, and abundance of individual insectivores twice as high (Blake and Loiselle 2001). In Brazil, 80% of the abundance of individuals and 69% of the biomass of understory species were of insectivores (Bierregaard 1990). The greater importance of omnivory in rainforest canopies contrasts with the greater predominance of insectivory in rainforest understories. This phenomenon may be related to the greater temporal and spatial unpredictability of canopy resources, which could favor diet generalists or agile species like migrants able to take advantage of diverse resources over broad areas (Martin 1985).
Another topic that has remained unsettled is whether the canopy of lowland neotropical rainforests is dominated by scrub species, as reported for Panama by Greenberg (1981), or forest species, as observed by Loiselle (1988) and Naka (2004). Our analyses for Honduras and Brazil showed that edge species are underrepresented in the forest canopy at both sites in comparison to the regional pool of species available to colonize the canopy. Although differences in observed and expected values for Honduras were low, we emphasize that the pattern of low importance of edge species found in both assemblages contrasts with previous suggestions that edge species should dominate the canopy stratum (Greenberg 1981, Walther 2002, Burney and Brumfield 2009). We propose that the occurrence of scrub species in the canopy at the site studied in Panama was due to the proximity of secondary forest and open habitats to the canopy tower. By comparison, the proportion of scrub species Naka (2004) observed in the canopy surrounding the tower at Reserva Ducke, situated on the outskirts of Manaus and surrounded on three sides by open, agricultural and human-disturbed habitats, was greater than at two towers in the midst of uninterrupted primary forest. Of further note, some of the most common species Greenberg (1981) reported from the canopy in Panama are commonly associated with gardens or forest edges, notably Coereba flaveola, which was only rarely observed in the canopy in Brazil and never in Costa Rica or Honduras, despite being found in neighboring secondary forests and open habitats.
Core Species of Lowland Neotropical Rainforest Canopies
Many species occur in the forest canopy as occasional visitors from lower forest strata or as vagrants from nonforest habitats, thus complicating a characterization of the core canopy assemblage. Having quantified the core canopy species for Honduras and compensated for methodological differences in previous studies, we are able to present a broad-scale characterization of the assemblage’s constituent genera. The species in the 25 genera that, at least historically, occurred at all sites should be taken as the nucleus of the core species or all species most likely to be found in neotropical canopies. The remaining 24 genera observed at a minimum of three sites complete the roster of core canopy constituents. Some general observations on this group are worth noting. In terms of species richness, the Tyrannidae are the predominant family in the canopy, with twice as many species as any other family. Other important families include the Thraupidae (18 species in our sample) and Psittacidae (9 species). Extirpations notwithstanding, the canopy typically includes seed predators that are large (Ara), medium (Amazona), and small (Brotogeris, Pionus). Although neotropical forest raptors are diverse in size and diet, the medium-sized species in the closely related genera Leucopternis and Pseudastur (Raposo do Amaral et al. 2009) that prey largely on reptiles and amphibians (Thiollay 1994a) appear to represent the core carnivores of the canopy. The genus Euphonia is particularly well represented in the canopy, with approximately 30% of known species observed just in our sample.
In conclusion, although we were able to address previously unanswered questions about the structure and organization of canopy bird assemblages, much remains to be learned about this understudied group of birds. As previously shown by Anderson (2009), ground methods alone are not adequate for canopy birds. We argue that the use of canopy-based methods at tropical field stations and other sites of continuous scientific research are essential for the accurate representation of long-term population trends, especially of secretive and rare or declining canopy species. Data thus derived from across a broader geographical range of lowland neotropical forests should help solidify our understanding of canopy bird assemblages and allow intensive analyses designed to reveal intrinsic differences in the structure and organization of bird assemblages in the top of the forest with those in the forest interior.
David L. Anderson and Luciano N. Naka
This is a dissertation on canopy birds in Pico Bonito National Park in Honduras. This is the most intensive study ever on birds of the rain forest canopy. It is written by David L. Anderson and Luciano N. Naka and presented through the Department of Biological Sciences and the Museum of Natural Science, Louisiana State University, Baton Rouge, LA.
Editor’s Note: Another individual, Robert Gallardo, has been studying Honduras birds for quite some time. He offers specialized birding tours if you are interested in a guided bird tour. Additionally, Mr. Gallardo has recorded Bird Sounds of Honduras, available on Amazon.com, and coming out Mid-May, 2015, is his long-awaited publication, Guide to the Birds of Honduras. [The first dedicated field guide to Honduras, featuring 1,000 illustrations on 73 color plates. The book also illustrates the majority of the migratory species.]