Study Area

The present study was conducted in an area of ca. 78 270 ha in the central Amazonian rainforest (2°36.78′S, 56°40.85′W), in a settlement project of the Brazilian Federal Government (Instituto Nacional de Colonização e Reforma Agrária—INCRA) in Parintins, Amazonas state, Brazil (Fig. 1), with 54 communities containing 3115 people (Silva and Oliveira 2010). The region is covered by primary lowland tropical rainforest and seasonally flooded forests. The vegetation is a mosaic resulting from different land uses, dominated by secondary forests and areas of slash-and-burn agriculture and livestock pasture (Silva and Oliveira 2010). In spite of human activities, the region still has intact primary forest, due to Brazil's Forest Code, which stipulates that 80% of the forest should be preserved (Brasil 1965).

Figure 1.

Location of study area in the central Amazonian forest, Parintins, Amazonas state, Brazil. Harpy Eagle diet was described from five active nests (stars) between September 2003 to November 2005.

The topography of the region is typical of the central Amazonian forest with clay and drained soils on the plateaus and sandy soils in valleys. The area is drained by a dense system of small streams and creeks. Altitude ranges from 40–150 m above sea level. The climate is tropical humid, average annual rainfall is 2223 mm, and mean annual temperature is about 26°C (INMET 2005).

In 2013, there were 15 known Harpy Eagle nests in the study area, averaging 1.5 km (0.6 km–2.5 km) away from local communities. With this proximity, as in many forest regions, there are human-wildlife conflicts in the study area because there is a predominant belief that large forest animals, such as jaguar (Panthera onca) and Harpy Eagle, prey heavily on livestock or domestic animals and represent a threat to their community or family (Valdez 2002, Lees 2006, Trinca et al. 2008, Curti and Valdez 2009, DeLuca 2012).

Diet Assessment

We collected all prey remains and pellets in the nests and on the ground around five of the seven known nesting trees (through 2005) of Harpy Eagles. We collected from the ground monthly from September 2003 to November 2005 at the five nests, and at four of these, a climber collected prey remains from in the nest one time, when the juvenile was 2 yr old, by which age the young spend much of their time away from the nest.

Prey Identification and Analysis

We separated bone remains into cranial and postcranial elements. We used cranial bones, mandibles, teeth, claws, hair, feathers, and beaks to identify prey to the most specific taxonomic level possible. We compared bone remains with reference material deposited in the mammal and bird collections of National Institute for Amazonian Research (INPA/Amazonas state) and in the Museu Paraense Emilio Goeldi (MPEG/Para state), both in Brazil.

The number of prey from each nest was estimated from the remains. When skulls and mandibles were fragmented, we attempted to find and fit the complementary bone parts of all samples to quantify the minimum number of individuals consumed. We used the number of pelvises to identify and quantify primates.

We measured skulls and mandibles of specimens from reference collections and used regression analyses to estimate body mass for each prey species. Body masses by family of unidentified birds and primates were estimated from available literature (Sick 1997, Emmons and Feer 1997, Paglia et al. 2012) and were represented by the average weight of an adult of that family. The occurrence (number of times a prey species was found as a percentage of all species sampled) represents the importance of a prey in the diet and was utilized to estimate niche breadth. Prey items were classified by locomotor habits (arboreal and terrestrial), and by guild: carnivore, frugivore, folivore, insectivore, omnivore, and seed-eater. The geographical distribution and taxonomy of the prey species were based on Wilson and Reeder (1993), Voss and Emmons (1996), Emmons and Feer (1997), Rylands et al. (2000), Alfaro et al. (2012) and Paglia et al. (2012). We examined the conservation status of each Harpy Eagle prey species in the IUCN Red List of Threatened Species (IUCN 2013). We used a Kruskal-Wallis test to compare the number of individuals and estimated body mass for each species consumed (Zar 1999).

Diet Niche Breadth

We analyzed niche breadth using the standardized Levins index (Colwell and Futuyma 1971):

where B is the Levins index (B  =  1/Σp_j²), p_j is the percentage of occurrence of each prey species, and n the number of prey species. Standardized Levins index (Bsta) values range between 0 (minimum niche breadth and, consequently, maximum selectivity) and 1 (maximum niche breadth, minimum selectivity; Krebs 1999).

Dietary Patterns Across the Harpy Eagle's Range

Differences in diet composition were evaluated using the diet niche breadth and through a nonmetric scaling ordination (NMDS) using the Bray-Curtis dissimilarity matrix of genus-level composition (presence-absence). We compiled literature reports of the diet of 17 Harpy Eagles from nine countries in the neotropics: Argentina (Chebez et al. 1990, Anfuso et al. 2008), Belize (Matola 2006, Rotenberg et al. 2012), Brazil (Galetti and Carvalho, Jr. 2000, Sanaiotti et al. 2001, this study), Ecuador (Muñiz-López et al. 2007), Guiana (Fowler and Cope 1964, Rettig 1978, Izor 1985), Panama (Álvarez-Cordero 1996, Touchton et al. 2002), Peru (Piana 2007), Suriname (Ford and Boinski 2007) and Venezuela (Álvarez-Cordero 1996). The NMDS is a procedure for fitting a set of points (sites) in a space such that the distance between points corresponds as closely as possible to a given set of dissimilarities between a set of objects (prey genus-level). Comparisons among sites were made by multivariate analysis of variance for repeated measures for sites with replicated data (Argentina, Belize, Panama, Venezuela, Guiana, and Brazil). Multivariate analyses were carried out using the packages “Vegan,” “Permute,” and “Mass” in the R 2.15.2 statistical program (R Development Core Team 2013).

Prey Species

Two species of sloths, the three-toed sloth (Bradypus variegatus) and two-toed sloth (Choloepus didactylus; Xenarthra), were the most frequent prey and represented the majority of the body mass consumed at five Harpy Eagle nests in Parintins, Amazonas state, Brazil (Table 1). Twelve other species, including primates, rodents, carnivores, and birds, also contributed to the Harpy Eagle diet; most of these (98%) were species with arboreal locomotor habits (Table 1).

Table 1.

The relative abundance and estimated body mass of prey species at five nesting sites of Harpy Eagle from September 2003 to November 2005 in the central Amazonian forest, Brazil. Locomotor: A  =  Arboreal; T  =  Terrestrial.

Sloths represented 78% of the diet in terms of number of individuals (Kruskal-Wallis H  =  33.08, df  =  11, P  =  0.001), whereas primates made up 11% of individuals consumed. The most common species of primate were red-handed howler monkey (Alouatta belzebul) at 4% and Hoffmann's titi monkey (Callicebus hoffmannsi) at 4%, followed by brown capuchin monkey (Sapajus apella) at 2%, white-nosed bearded saki monkey (Chiropotes albinasus) at 1%, and gray saki monkey (Pithecia irrorata) at 1%. All other nonprimate species added together equaled 5% (Table 1). Folivores were the dominant dietary items at 82%, followed by frugivores (9%), seed predators (3%), and omnivores (1%; Table 1). Sloths also accounted for 85% (436 kg; Kruskal-Wallis H  =  68.78, df  =  11, P  =  0.001) of body mass of prey in the Harpy Eagle diet at the five nests during three years (Table 1). All prey weighed estimated between 0.3–6.5 kg (mean 2.6 ± 0.82 kg, n  =  252; Fig. 2).

Figure 2.

Frequency distribution of the estimated body mass of prey consumed (n  =  252) by Harpy Eagles in five nests over 3 yr in Parintins, central Amazonian forest, Brazil.

Diet Niche Breadth

There was an evident diet specialization of the Harpy Eagle, based on the niche breadth. Bsta  =  0.171, which was relatively low (Table 2).

Table 2.

Richness, standardized Levins diversity index (Bsta), and percentage of arboreal and folivore species from seventeen Harpy Eagle diet studies from Central and South America. n  =  number of individuals.

Dietary Patterns Across the Harpy Eagle's Range

The richness of prey species consumed by Harpy Eagles differed significantly, ranging from 3 to 20 species (R²  =  0.78, P  =  0.04); therefore, Levins diversity indices were low and similar between regions (Table 2). Prey species with arboreal habits dominated the Harpy Eagle diet (Table 1), and all seventeen studies revealed the same trend (Table 2).

The community structure of Harpy Eagle prey (genus-level) throughout its geographic range, based on presence/absence (NMDS r²  =  0.86; stress  =  0.1) was different among sites (P  =  0.04; MANOVA - Pillai-Trace  =  1432; Fig. 3). The high r² and low stress mean that the ordination was able to summarize and represent the pattern of data.

Figure 3.

Axis 1 and 2 from the NMDS (nonmetric multidimensional scaling) ordination scores derived from a Bray-Curtis dissimilarity matrix of presence/absence of prey genus-level in the Harpy Eagle diet in its distribution range. Distances between points reflect a dissimilarity matrix created using the Bray-Curtis coefficient. Argentina – Arg1 (Chebez et al. 1990), Argentina - Arg2 (Anfuso et al. 2008), Belize – Bel1 (Matola 2006), Bel2 (Rotenberg et al. 2012), Brazil - Bra1 (Galetti and Carvalho, Jr. 2000), Brazil - Bra2 (Sanaiotti et al. 2001), Brazil - Bra3 (this study), Ecuador – Ecu (Muñiz-López et al. 2007), Guyana – Gui1 (Fowler and Cope 1964), Gui2 (Rettig 1978), Gui3 (Izor 1985), Panama – Pan1 (Touchton et al. 2002), Pan2 (Álvarez-Cordero 1996), Peru – Per (Piana 2007), Venezuela – Ven1, Ven2 (Álvarez-Cordero 1996).

Prey Species

Sixty-nine species of mammals, birds, and reptiles (twenty-eight mammal genera) have been described as part of the Harpy Eagle diet from Central to South America (Fowler and Cope 1964, Rettig 1978, Izor 1985, Chebez et al. 1990, Álvarez-Cordero 1996, Galetti and Carvalho, Jr. 2000, Sanaiotti et al. 2001, Matola 2006, Touchton et al. 2002, Muñiz-López et al. 2007, Piana 2007, Anfuso et al. 2008, Rotenberg et al. 2012, including this study). Several other records of Harpy Eagle predation on mammal species have been reported (Eason 1989, Peres 1990, Sherman 1991, Ferrari and Port-Carvalho 2003, Martins et al. 2005, Benchimol and Venticinque 2010, Lenz and Reis 2011, Barnett et al. 2011), including black-headed uacari monkey (Cacajao ouakary; Barnett et al. 2011) that have not yet been documented in Harpy Eagle diet based on prey remains.

The prey brought to the nests in Parintins weighed an average of 2.6 kg (±0.82 kg, n  =  252). Analyzing the range of estimated body mass of sloths recorded from this study and comparing with the adult weights, (5.5 kg for three-toed sloth and 8.5 kg for two-toed sloth; Emmons and Feer 1997), we assume that the major predation frequency was on juveniles and subadults. This assumption was supported based on analysis of cranial sutures, which were not completely fused in subadults and juveniles (Hoson et al. 2012).

A Harpy Eagle preyed on an infant monkey (Sapajus apella) at Balbina Hydroelectric Reservoir within the Uatumã Biological Reserve in the central Amazonian forest, by capturing the infant from the back of an adult female (Benchimol and Venticinque 2010). Similarly, we observed a baby sloth (Bradypus variegatus) of approximately 260 g being carried to a nest in the Carajas National Forest (T. Sanaiotti unpubl. data). Large prey can be more difficult to catch and handle, which reduces the probability of a successful attack. In some areas, a Harpy Eagle successfully attacked prey that could not be carried in flight and lost the prey between the ground and the canopy (Martins et al. 2005). In 2005, W. Magnusson (pers. comm.) observed a dead howler monkey (6 kg) dropped by a Harpy Eagle at the Adolpho Ducke Forest Reserve, Manaus, Amazonas state, Brazil.

We also found a bias among the type of prey remains collected in the nests, in that we more frequently found the upper body parts of sloths (skulls and forelimbs) but the lower body parts (thoracic, sacral and caudal vertebrae, pelvis and leg bones) of primates. Primate skulls were rarely recorded in the remains. We believe that the adults fed on the prey before bringing the remainder to the nest. The same pattern was reported in other Harpy Eagle diet studies (Fowler and Cope 1964, Rettig 1978, Izor 1985, Muñiz-López et al. 2007).

In contrast to local beliefs, there was no evidence in the prey remains that Harpy Eagles fed on livestock or on domestic animals such as chickens or ducks, even where Harpy Eagle nesting territories overlapped human settlements. This was verified by testing mitochondrial DNA in birds not identified by bone samples (F.H. Aguiar-Silva unpubl. data). The biomass of wild animals in the study area is probably sufficient to support the population of Harpy Eagles, suggesting that there is enough forest cover to allow Harpy Eagles and arboreal mammals to coexist. In Acre (southwestern Amazonia) and Mato Grosso (central South America) states, humans killed adult Harpy Eagles due to perceived risk to livestock and domestic animals or to satisfy their curiosity (Lees 2006, Trinca et al. 2008, DeLuca 2012). Thus, effective conservation to maintain large areas of habitat for the Harpy Eagle are needed to prevent what has happened in Asia, where the Philippine Eagle (Pithecophaga jefferyi)—the most endangered eagle in the world (Salvador and Ibañez 2006) due to habitat loss—consumes chickens (Gallus gallus domesticus), cats (Felis catus), and dogs (Canis lupus familiaris; Concepción et al. 2006).

Diet Niche Breadth

Although the Harpy Eagles in our study preyed on a large variety of species such as primates, rodents, kinkajous, marsupials, and birds, the sloths B. variegatus and C. didactylus were the most frequent prey in their diet; this may be due to the habits of sloths, including slow movement (Montgomery and Sunquist 1978). Sloths accounted for two-thirds of the Harpy Eagle diet in the five nests of Parintins, central Amazonian forest, Brazil. Together the two species of sloths contributed a maximum of 436 kg in biomass of prey consumed by the five Harpy Eagle pairs and delivered to the nestlings in the nests. The behavioral habits of sloths in the canopy (Montgomery and Sunquist 1978, Queiroz 1995) probably increased their probability of detection by the Harpy Eagle. The low number of primates in the Harpy Eagle diet may be due to antipredation behaviors of primates (Gil-da-Costa 2007), which reduce the risk of attacks, at least for the dependent young (Neal and Norconk 2009). Researchers reported unsuccessful attempts of predation on howler monkey (Alouatta seniculus; Eason 1989; Peres 1990; Sherman 1991) and on bearded saki (Chiropotes utahicki; Martins et al. 2005) all of which live in groups. These researchers asserted that the presence of a group, which made it impossible for the Harpy Eagle to carry away the prey, resulted in an unsuccessful predation attempt.

The Harpy Eagle primarily consumed arboreal mammals (86% of species consumed; Table 1, 2) like other large eagles, including the Philippine Eagle in Asia (Ibañez et al. 2003) and the Crowned Hawk-Eagle (Stephanoaetus coronatus) in Africa (Mitani et al. 2001). The same pattern was recorded for almost the entire Harpy Eagle range, with the exception of Argentina (Arg1  =  40% arboreal; Table 2). Success rates of observed capture attempts by Harpy Eagles in Panama seem to be higher on solitary arboreal rather than for social arboreal animals (Touchton et al. 2002).

The percentage of folivores in the Harpy Eagle diet recorded for all Harpy Eagle diets (Table 2) supports the hypothesis that a predator can “keep the world green” (Terborgh et al. 2006). By consuming the most abundant herbivores in the forest, the Harpy Eagle helps limit herbivore impact on primary production; herbivores and granivores can have an important effect on forest regeneration (Dirzo and Miranda 1991). For example, in Panama the absence of large predators was associated with decreased tree diversity (Terborgh et al. 2006).

Dietary Patterns Across the Harpy Eagle's Range

The dissimilarity in richness and in Levins diversity index recorded in the Harpy Eagle diet (Table 2, Fig. 3) may be partly explained by the sources of data, which, although temporally and spatially independent, rely on diverse methodological approaches and sample sizes (Table 2). The richness of prey consumed by Harpy Eagle in our study area (n  =  14 prey species) was low compared with the Harpy Eagle diet from Venezuela (n  =  20 prey species; Álvarez-Cordero 1996), which was surprising because the central Amazonian forest, where the five Harpy Eagle nests we studied were located, has high species diversity (Voss and Emmons 1996, Myers et al. 2000, Paglia et al. 2012, Jenkins et al. 2013). It is possible that the small sample size (five nests) in our study area was responsible for the lower diet diversity.

Although most of the studies indicated a highly specialized diet, some of them, such as those in Argentina (Arg1, Arg2), Belize (Bel2), Panama (Pan2), and Brazil (Bra1), revealed minimal selectivity (Table 2). According to optimal foraging theory, as the abundance of a preferred food type in the diet increases in the foraging habitats used, the number of less-preferred food types included in the optimal diet will decrease (Pyke et al. 1977). Harpy Eagle diets in Belize and Argentina, where the sloths were absent (Moraes-Barros et al. 2010, Rotenberg et al. 2012) included increased consumption of other species, resulting in a less specialized diet (Table 2).

The composition of Harpy Eagle prey (by genera) was reduced to two axes by NMDS, with 78% of the total variation in diet composition explained. The configuration points (sites) pattern (Fig. 3) in the Harpy Eagle diet for each country reflects the presence/absence of some important species, such as the sloths. The Xenarthra, particularly the sloths, dominant in Neotropical forests (Eisenberg and Thorington 1973, Mascarenhas and Puerto 1988, Janson and Emmons 1990), decrease in diversity and abundance (Acevedo-Quintero et al. 2011, Catzeflis and Thoisy 2012) at the extreme limits of the Harpy Eagle distribution range (Belize and Argentina). Belize is at the northern edge of the range for sloths, and supports only two primate species (Rotenberg et al. 2012), which may explain the dissimilarity among diets reported in this study and those recorded in Argentina and Belize.

The first Harpy Eagle diet reported in Brazil (Bra1; Galetti and Carvalho, Jr. 2000) was separated from the other two Brazilian sites (Bra2, Bra3; Fig. 3) because the Harpy Eagle diet in that region exhibited low richness compared to this study and others (Table 2), which is likely explained by the sample size difference among these studies. Variations in the abundance and availability of prey species due to differences in demographic factors, weather and seasonal patterns (Montgomery and Sunquist 1978, Touchton et al. 2002), reproductive period (Dias et al. 2008), vegetation structure (Acevedo-Quintero et al. 2011), antipredator behavior of potential prey (Gil-da-Costa 2007), and exploitation of resources by human communities (Alvarez and Ellis 1994, Wright et al. 2000) may also explain variations in the diet of Harpy Eagles among locations.

Harpy Eagles tend to catch similar genera of prey species in different parts of their geographic range. The wide variety of prey species consumed by Harpy Eagles across their range may indicate opportunistic foraging behavior, as in other raptors (Skorupa 1989, Concepción et al. 2006) leading to a maximization of fitness (Pyke et al. 1977). This foraging behavior would allow the Harpy Eagle to use a wide range of prey species and to remain in areas surrounded by different kinds of land use.

The diversity of species taken by the Harpy Eagle throughout its geographic range suggests that this large predator might not be food-limited, but instead shows great feeding adaptability. This foraging behavior is a primary factor in the functional responses to maximize fitness (Pyke et al. 1977). There are breeding records from Venezuela (Alvarez and Ellis 1994, Álvarez-Cordero 1996) and Peru (Piana 2007), where Harpy Eagles have been breeding for years in areas dominated by human activities such as logging and manipulated forest, livestock grazing, and subsistence hunting. Optimal foraging theory (MacArthur and Pianka 1966, Pyke et al. 1977) suggests that dietary specialization can occur when a reliable and abundant food source is available. It seems likely that the abundance and accessibility of sloths, within their distribution range, has facilitated this level of dietary specialization. Even with the bias of different sample sizes in the studies throughout the Harpy Eagle's range, the patterns of resource use shown are valuable for understanding the dietary flexibility of the species.

Our analyses cannot show the extent to which a particular prey species was affected by Harpy Eagle predation because we do not have data on the relative population sizes of available prey. It seems reasonable that Harpy Eagle eat more sloths where sloths were common and eat more of other prey species where sloths are less common. Having a broader range of prey species reduces vulnerability from population fluctuations in any one prey species (Touchton et al. 2002).

Implications for Conservation

The patterns of prey species recorded in the Harpy Eagle diet are the result of various important factors acting on the foraging behavior and consequent predation success. Dietary studies of a top predator can show the interrelations between species, which can help researchers evaluate the carrying capacity in the remaining forests and in the protected areas in the near future. We found that in the central Amazonian forest, Harpy Eagle nesting territories that are near human development areas with different land uses (e.g., logging and small-scale agriculture) can support the reproduction of this large predator. Close proximity between Harpy Eagle nesting territories and human settlements has also been reported in Venezuela and Panama (Alvarez and Ellis 1994).

Conservation of the Harpy Eagle should include measures to protect nesting territories and prey to maintain the variety and availability of resources through its entire geographic range. We recommend outreach and environmental education within local communities to help reduce the belief that the Harpy Eagle is a threat to domestic animals (Curti and Valdez 2009). With traditional management focused on the species (Simberloff 1998), the Harpy Eagle can also be used as a charismatic keystone species (Simberloff 1998, Touchton et al. 2002, Curti and Valdez 2009), so that broader regions of these ecosystems may be protected.

The results from our study (1) extend our knowledge of food resources used by the Harpy Eagle, (2) suggest that the Harpy Eagle in this region does not include domestic animals in their diet, and (3) offer a comparison of the diet of the species throughout their range.