Food items consumed by sympatric maned wolves (Chrysocyon brachyurus), crab-eating foxes (Cerdocyon thous), and hoary foxes (Lycalopex vetulus) were investigated in the Cerrado of central Brazil to determine overlap among their diets. Home ranges and spatial segregation among these canids were also investigated. Overlap among diets was greatest for maned wolves and crab-eating foxes, which had generalist diets, although maned wolves fed on larger prey than did crab-eating foxes. Hoary foxes were frugivore–insectivores, with Syntermes termites being the most important food item. In relation to habitat use, hoary foxes were also the most selective canid, preferring more open habitats.
The Cerrado is the second largest biome of South America, covering an area of 2 million km2 in central Brazil (Alho and Martins 1995). In most of the Cerrado, 3 canids occur sympatrically: maned wolves (Chrysocyon brachyurus), crab-eating foxes (Cerdocyon thous), and hoary foxes (Lycalopex vetulus). Current knowledge concerning these species consists primarily of their feeding habits (Bisbal and Ojasti 1980; Dalponte 1997; Dietz 1984; Facure and Monteiro-Filho 1996; Montgomery and Lubin 1978; Motta-Junior et al. 1994, 1996; Schaller 1983). Few studies focus on ecological aspects, such as home range and activity patterns (Brady 1979; Dietz 1984; Macdonald and Courtenay 1996; Sunquist et al. 1989), and no study has investigated the ecological relationships among these canids.
Both maned wolves and crab-eating foxes are territorial animals, with crepuscular and nocturnal periods of activity (Brady 1979; Dietz 1984; Macdonald and Courtenay 1996; Sunquist et al. 1989). The maned wolf is a large canid (20–25 kg), distributed over central Brazil and reaching eastern Bolivia, Paraguay, and northeastern Argentina (Cabrera 1961; Dietz 1984; Langguth 1975; Medel and Jaksic 1988). It feeds mainly on small mammals and fruit, primarily on lobeira, Solanum lycocarpum (Carvalho 1976; Dietz 1984; Motta-Junior et al. 1996; Schaller 1983). The crab-eating fox is a medium-sized canid (4–7 kg) that occurs in tropical and subtropical forests and grasslands in Colombia and Venezuela, southward into Brazil, Paraguay, northern Argentina, and Uruguay (Berta 1982). Like the maned wolf, it is a dietary generalist, eating fruits, small mammals, and insects (Bisbal and Ojasti 1980; Brady 1979; Facure and Monteiro-Filho 1996; Motta-Junior et al. 1994). The hoary fox is one of the less-studied neotropical canids (Cabrera 1961; Langguth 1975). It is a small-sized canid (4 kg) that inhabits the Cerrado grasslands in central Brazil (Coimbra-Filho 1966; Langguth 1975). It is omnivorous, feeding on small mammals, birds, and insects, especially termites and grasshoppers (Cabrera and Yepes 1960; Langguth 1975; Nowak 1991). Dalponte (1997) studied the diet of hoary foxes in the wild and verified that the most important food item was, indeed, termites (Syntermes), present in 90% of their feces.
Of a possible total of 55 pairs of canids in South America, only 12 (18%) exhibit some degree of sympatry (Medel and Jaksic 1988). Such low degrees of sympatry found for these species may be related to interspecific territoriality, serving as a mechanism to reduce competition (Medel and Jaksic 1988). Segregation of species along 1 or more niche dimensions, such as differences in diet, activity patterns, and habitat use, reduces competition, thereby allowing coexistence (Pianka 1981). Studies of closely related species show that they frequently differ in their use of available resources (Bothma et al. 1984; Jaksic et al. 1980, 1983; Johnson and Franklin 1994; Mills 1989). Rosenzweig (1966) suggests that the occurrence of sympatric carnivores is facilitated by the size differences among them because they specialize in prey of different sizes.
In this study, we quantified and compared the diet of 3 sympatric canid species in the Cerrado through an analysis of their feces. We also report on aspects of home ranges and habitat use.
Materials and Methods
We conducted this study on the Rio Pratudão Ranch, in the municipal district of Jaborandi, Bahia, near the border of the states of Bahia and Goiás, in Brazil (14°14′S, 45°56′W; Fig. 1). The ranch comprises approximately 70,000 ha, located on an extensive plateau with an average altitude of 900 m. The study area occupied approximately 20,000 ha, of which about 5,000 ha are used for corn and soy plantation. The predominant habitats are grasslands (campo) and wooded savannas (cerrado sensu stricto). Semideciduous forests occur as irregular patches within the wooded savannas (Eiten 1979). The climate is tropical with rainy and dry seasons. The dry season is about 6-month duration, from mid-May to mid-October. Ninety-four percent of the rainfall occurs during the rainy season, with the greatest precipitation during January (<190 mm) and the least precipitation during June and July (<9 mm). The average annual precipitation and temperature are 1,300 mm and 21.0°C, respectively.
We determined diets of the canids through an analysis of fecal samples collected along highways, along trails, and inside traps of captured animals from July to December 1995. A greater number of fecal samples were collected during the dry season than during the rainy season because they easily disintegrate during the rainy season. We identified feces to species through their form, size, presence of hairs, and presence of footprints in the immediate area. Feces of doubtful origin were discarded. We immersed fecal samples in 70% alcohol for 2–3 days, washed them with running water using fine-mesh filters (0.5 and 1.0 mm), and dried them in an oven at 40°C. We identified food items, comparing undigested material (teeth, feathers, seeds, hair, and others) with specimens in the collections of the Zoology and Botany Departments of the Universidade de Brasília and with a reference collection of fruits from the area. A total of 70 maned-wolf feces, 39 crab-eating fox feces, and 37 hoary-fox feces were analyzed.
The importance of each food item was determined by its relative frequency: number of scats containing the item as a function of the number of occurrences of all items (Dietz 1984), expressed as a percentage. Because frequency tends to overestimate the importance of smaller items (Corbett 1989), we also estimated the biomass consumed. All material identified in each fecal sample were counted and multiplied by the average weight of the consumed item. To avoid counting certain prey more than once, we only considered certain structures for each taxon. Thus, for small rodents, maxillaries containing molar teeth were counted, which provides the best estimate for the proportion of prey consumed (Lockie 1959). We obtained numbers of termites and ants through number of cephalic capsules in the feces but did not count fragments of other insects. To estimate number of worker and soldier termites, we used the proportion of 9.6 workers to 1 soldier reported for Syntermes dyrus (Negret and Redford 1982).
To obtain number of fruits consumed, we counted number of seeds present in feces. Whenever number of seeds in a fecal sample was less than that normally found in a fruit, we considered that the animal had consumed a whole fruit. The average weight of each food item was obtained in the study area from the reference collections or, in some cases, from the literature.
Width of the dietary niche was calculated from Simpson's diversity index (Ricklefs 1990), B = 1/Σ (Pij2), where B is the width of the diet, and Pij is the proportion of the items i used by species j. We calculated B by grouping items into the following categories: fruits, arthropods, reptiles, birds, and mammals.
The measure of diet overlap among the species studied was calculated through the equation O = Σ (PijPik)/[Σ (Pij2) Σ (Pik2)], where Pij is the proportion of items i used by species j, and Pik is the proportion of the items i used by species k (Ricklefs 1990). This index ranges from 0 to 1, from complete dissimilarity to complete similarity.
We captured hoary and crab-eating foxes using bobcat model Tomahawk live traps (Tomahawk Live Trap Company, Tomahawk, Wisconsin). We fitted captured animals with TW-3 radio transmitters, set at a frequency of 173 MHz (Biotrack, Wareham, Dorset, United Kingdom) and set to be detected at a distance of ≤1.5 km. Three captured animals received radio transmitters. On 2 July 1995, a lactating hoary fox was captured. On 7 December 1995, a male hoary-fox pup was captured in the same area. Both were tracked until the beginning of January 1996. The crab-eating fox, captured on 20 October 1995, was tracked until 15 December of the same year. We located the animals by car, with an M57 receiver (Mariner Radar Ltd., Lowestoft, Suffolk, United Kingdom) and a Yagi 4-element directional antenna (Biotrack), and determined its position through triangulation (White and Garrot 1990) using a compass and a global positioning system (GPS) device.
We located animals daily and recorded habitat type where the animal was encountered. In addition to daily monitoring (1 or 2 locations for each animal), we also tracked them at night with sampling periods of 6 h/day, totaling 36 h of tracking for the 3 animals sampled (2 hoary foxes and 1 crab-eating fox). During sampling periods, positions of the animals were taken at 15-min intervals, and linear distances between consecutive points were summed to obtain total distance traveled by the animal.
Home ranges were determined using the minimum-convex-polygon method (Mohr 1947) with the McPaal program (Conservation and Research Center of the National Zoological Park, Smithsonian Institution, Washington, D.C.).
To study habitat use by the canids, each week we recorded locations and habitat types where feces were deposited along 100 km of roads and trails. Types of habitats considered in this analysis (and distance covered in each habitat) were shrub grasslands (25 km), wooded savannas (30 km), patches of dry forests (5 km), and cultivated areas (40 km). Distances traveled in each habitat roughly correspond to proportion in which they appear in the study area. We then compared the frequencies of use of habitat types by each species.
Thirty-three taxa were identified in the diet of the maned wolf, 13 of plant origin and 20 of animal origin. Plant material was more frequent, representing 50% of the 191 food items. Lobeira fruit, S. lycocarpum, was the most important item both in frequency and in estimated biomass (Tables 1 and 2), being found in 87% of the analyzed feces. Mammals were the vertebrates most consumed by the maned wolf, of which rodents from family Muridae were the most frequent (Table 1). Some taxa identified in this family were Bolomys, Calomys, and Pseudoryzomys. In relation to estimated biomass, however, small rodents had the same importance as armadillos had, of which fewer were consumed (Table 2). Among avifauna, Tinamidae was the most frequent (Table 1). A blue and yellow macaw (Ara ararauna), found in 1 of the samples, was the largest single item in this group (Table 2). Insects and reptiles were seldom consumed by the maned wolf (Table 1).
In diets of crab-eating foxes, 25 taxa were identified: 9 plant and 16 animal. From a total of 110 occurrences of food items, 73% were of animal origin. The most frequent items were murid rodents (Table 1). In relation to estimated biomass, fruits had the greatest representation, with lobeira being the most important item (Table 2). Reptiles, birds, and insects were less important. Among insects, families Gryllidae and Acrididae were the most frequent (Table 1). No estimate of biomass consumed was attempted for these insects because undigested parts in the feces were broken into fragments. These insects only occurred in low numbers in fecal samples, accounting for <1% of nondigestible remains.
In hoary fox diets, 22 taxa were identified: 9 plant and 13 animal. Insects represented 59% of 123 total occurrences of food items in samples. Isoptera was the insect group most consumed by the hoary fox, with termites from the genus Syntermes being the single most important insect item found in relation to both frequency and estimated biomass (Tables 1 and 2). Other important groups of insects in hoary fox diets were Orthoptera (Gryllidae, Acrididae, and Tettigoniidae), Coleoptera, and Hymenoptera (Formicidae). Although it was not possible to estimate numbers of these insects in feces, they occurred in 60% of analyzed samples and represented an average of 15% of total nondigestible remains. In relation to biomass, fruits were more important than items of animal origin (Table 2). Vertebrates in the diet of the hoary fox were birds and mammals (mainly murid rodents), which were infrequently represented and were low in biomass (Table 2).
Width and overlap of feeding niches
The hoary fox presented the narrowest dietary niche (B = 2.37), followed by the maned wolf (B = 3.13) and the crab-eating fox (B = 3.54). Diet overlap between these canids was greatest between the maned wolf and the crab-eating fox (O = 0.76) and lower between these and the hoary fox (hoary and crab-eating foxes, O = 0.30; hoary fox and maned wolf, O = 0.10).
The average distance traveled by the female hoary fox from 1900 to 0100 h during 3 sampling nights was 4.76 km. The distance traveled at night, from 0100 to 0700 h during 1 sampling night, was 4.55 km. The home-range size of the adult female, calculated from 60 locations, was 385 ha (minimum convex polygon), and the home-range size of the pup, calculated from 15 locations, was 208 ha (minimum convex polygon).
Average distance traveled by the crab-eating fox, from 1900 to 0100 h on 2 sampling nights, was 10.9 km. Size of home range of the crab-eating fox, calculated from 30 locations, was 1,280 ha (minimum convex polygon).
Feces collected in different habitat types indicated that maned wolves and crab-eating foxes can use all the habitats present in the study area (Table 3). But sampling in the shrub grassland and wooded savanna, which corresponded to 55% of the total sampling effort, resulted in 93% of all scats found for maned wolves and 94% of scats for crab-eating foxes, showing a clear preference for these habitats. Hoary foxes were even more selective, preferring more open areas—only 1 fecal sample of this species was found along the border of the wooded savanna. All others were in grassland habitats (Table 3).
Differences in diet
Although diets of maned wolves and crab-eating foxes overlapped significantly (O = 0.76), crab-eating foxes fed exclusively on small-sized prey, whereas maned wolves consumed a great number of medium-sized prey (Table 2). Bestelmeyer and Westbrook (1998) reported a maned wolf preying on pampas deer, indicating that it may feed on larger prey. During this study, we found a molar tooth from a brocket deer (Mazama) in one of the maned wolf's fecal samples. Vertebrates heavier than 1 kg, although consumed by maned wolves sporadically, were more representative relative to the biomass consumed (24%) than were small mammals (20%), a pattern that was also found by Motta-Junior et al. (1996).
Among fruits, lobeira was the most important item in the diet of C. brachyurus and C. thous. This fruit was overestimated in the diet of the crab-eating fox because, although few seeds were found in the fecal samples (average of 8), a typical fruit contains about 300–500 seeds. The method we used for estimating biomass consumed tends to overestimate the consumption of fruits because the animal does not always consume an entire fruit. Canids are important dispersers for plants of the Cerrado. They feed on a great variety and quantity of fruits, do not chew their food much, and seemingly do not present specific adaptations for the digestion of cellulose. The seeds remain intact after passing through the digestive tract of these animals.
Despite the relatively high frequency of insect remains (23%) in the feces of the crab-eating fox, these were consumed in small amounts considering the numbers of individuals per fecal sample. Insects were even less important in the diet of the maned wolf. It is possible that at least some of the insects were accidentally ingested while both species foraged for larger prey or fruits.
Crab-eating foxes and hoary foxes, similar in size and taxonomically closely related (Wayne and O'Brien 1987), exhibited low overlap between their diets (O = 30). Hoary foxes were frugivorous and insectivores and crab-eating foxes were omnivorous. Hoary foxes consumed high amounts of fruits and insects and a low number of vertebrates. Termites from the genus Syntermes were present in 92% of fecal samples (up to 5,000 individuals per sample), whereas small vertebrates (birds and mammals) were present in only 32% of fecal samples, and only one individual was found per sample. Dalponte (1997) found similar results for hoary foxes: Syntermes was present in 90% of samples, and small vertebrates (reptiles, birds, and mammals) were present in 25% of samples. Biomass of termites (Syntermes), which composed the diet of the hoary fox, was analogous to that of murid rodents in the diet of the crab-eating fox (23.9% and 24.1%, respectively). Other insects, such as crickets, grasshoppers, and beetles, were also important in the diet of the hoary fox. Considering the small amount of vertebrates consumed, termites and other insects are the primary source of protein in the diet of the hoary fox.
Differences in habitat use
Other studies with maned wolves and crab-eating foxes have already shown that these animals can use all types of habitats of the Cerrado. In the Serra da Canastra, maned wolves spend 34% of the time in grassland areas, 43% in wooded savanna, and 24% in forested habitats (Dietz 1984). The crab-eating fox, although poorly studied in the Cerrado region, is a habitat generalist (Macdonald and Courtenay 1996). In contrast, the hoary fox uses open areas of the Cerrado almost exclusively.
The preference of the hoary fox for more open habitats of the Cerrado may be related to distribution of important food resources. Negret and Redford (1982) reported that 90% of S. dyrus colonies are located in grassland areas. Despite the fact that canids are, in general, opportunistic animals (Kleiman 1967), hoary foxes may be considered as specialists in relation to both diet and habitat use. The little overlap in the use of food resources seems to facilitate the coexistence of these 3 canids of the Brazilian Cerrado.
Os itens alimentares consumidos pelo lobo-guará (Chrysocyon brachyurus), cachorro-do-mato (Cerdocyon thous), e raposa-do-campo (Lycalopex vetulus), canídeos simpátricos do Cerrado, foram analisados para investigação da sobreposição nas dietas desses animais. As áreas de vida e a segregação espacial desses canídeos também foram investigadas. A sobreposição nas dietas foi maior entre o lobo-guará e o cachorro-do-mato, os quais apresentaram dietas generalistas. No entanto, o lobo-guará se alimentou de presas maiores do que o cachorro-do-mato. A raposa-do-campo apresentou uma dieta frugívora-insetívora, sendo cupins do gênero Syntermes o item alimentar mais importante. Em relação ao uso do habitat, a raposa-do-campo também foi o canídeo mais seletivo, preferindo habitats mais abertos.
We thank M. M. Guimarães for assistance in fieldwork and the personnel of Fazenda Rio Pratudão for permission to study in the area and to use the excellent facilities. We also thank C. Proença for identification of botanical specimens, A. Percequillo for identification of small mammals, K. Kitayama for identification of termites, G. Colli for identification of reptiles, M. Bagno for identification of birds, and R. Macedo and an anonymous reviewer for review of the manuscript. Conselho Nacional de Desenvolvimento Cientifico e Tecnológico-CNPq provided a graduate scholarship to K. M. Juarez and financial support to J. Marinho-Filho (Proc. 300591/86-1).