Diet of crab-eating fox (Cerdocyon thous) in two conservation units of the Amazon rainforest, Brazil

The present study aimed to evaluate the diet of the free-living crab-eating fox by identifying the stomach contents of the 17 crab-eating foxes ( Cerdocyon thous ) roadkilled in two conservation units, both located in the Amazon rainforest. The food items were quantified by frequency of occurrence (FO) and percentage of occurrence (PO). The stomach contents were analysed for dry matter (DM), crude protein (CP), crude fibre (CF), ether extract (EE), and mineral matter (MM). Nitrogen-free extractives (NFE), metabolisable energy (ME) values, as well as the energy need for maintenance were estimated. The composition of the diet for the crab-eating fox presented 29 food items from the different taxonomic groups, with a greater diversity of items of animal origin (n=22), although the highest frequency of occurrence was gramineae (Poaceae) (41.18%). Among the items of animal origin, 21% were mammals, 18% reptiles, 10% amphibians, 9% invertebrates and 3% birds. A high content of CF (62.76%) were determined. Nitrogen-free extractive and dry matter averages were 5.91% and 141.82 kcal/100g, respectively. The average energy was 447.01 kcal/day. These findings suggesting that the foxes have a generalist diet with an omnivorous diet in the Amazon basin, feeding on gramineae, fruits, amphibians, birds and small mammals and have the same feeding habit that present in other Brazilian biomes.


Diet of crab-eating fox (Cerdocyon thous) in two conservation units of the Amazon rainforest, Brazil
Dieta do cachorro-do-mato (Cerdocyon thous) em duas unidades de conservação da Floresta Amazônica, Brasil predominates on the tops of the plateaus and, specifically in the National Forest of Carajás, the ferruginous rupestrian field occurs (IBAMA, 2006;ICMBio, 2018). The climate of the region is humid tropical climate, with dry winter and with average precipitation of the driest month below 60 mm. The climatic seasons are well defined, with the period of greatest precipitation occurring between the months of December to April and the driest period between July and August (IBAMA, 2006;Raiol, 2010).

Animals
A total of 17 stomach content samples were taken from roadkilled adult crab-eating foxes (Cerdocyon thous), 12 males and 5 females, on two roads that cross the National Forests of Carajás and Tapirapé Aquiri National Forest. This study was authorized by the Committee on the Use of Animals of School of Veterinary Medicine and Animal Science, UNESP (CEUA 0227/2017), and by Brazilian Institute of the Environment and Renewable Natural Resources (SISBIO 66964).
The collection of run over animals was part of programs for monitoring of roadkill fauna from 2008 to 2018. For each animal collected, the date, time of collection, kilometer and road were recorded. In addition, body size, tail size and body mass of animals were verified according to Emmons and Feer (1997). Subsequently, the carcasses were taken to Anatomy and Zoology laboratory of the Universidade Federal Rural da Amazônia (UFRA), Parauapebas Campus, where they were dissected and the contents of each stomach were removed, frozen and destined for studies of food and bromatological items.
The analysis of stomach contents was performed with stereoscopic and optical microscopes, where the and present great ecological importance, regulating the population of natural prey and, thus, influencing the entire dynamics of the ecosystem in which they live (Pitman et al., 2002).
The crab-eating fox (Cerdocyon thous, Linnaeus, 1766) is a medium-sized Neotropical canid (average body mass of 6 kg), which lives in pairs or small groups and is considered widely distributed and is common in South America (Eisenberg and Redford, 1999). Although not a typical species in the Amazon rainforest, there are records of the presence of this animal in the north of the State of Mato Grosso (southern region of the Amazon Forest, in an area of forest impacted by deforestation) (Michalski and Peres, 2007), northeast of Pará (east of the biome, in forest remnants) (Stone et al., 2009), and southeast of the state of Pará (in areas of environmental preservation) (Carvalho et al., 2014).
The crab-eating fox is considered of little concern, not appearing in the Red Book of Endangered Brazilian Fauna (ICMBio, 2018) and in none of the Brazilian state lists of threatened species. Despite this, many populations are impacted by the roadkilling of individuals on the highways, since this is one of the species of carnivores with the highest occurrence of deaths of this type (Vieira, 1996;Cândido-Junior et al., 2002;Gumier-Costa and Sperber, 2009;Lemos and Facure, 2011). Although death by roadkill of wild species is one of the challenges for Conservation Biology (Forman and Alexander, 1998), roadkilled animals can provide important biological information (Vieira, 1996).
One of the possibilities for scientific use of roadkilled animals is the study of the diet based on stomach contents (Aguiar et al., 2011). The knowledge of food habits is essential to understand the trophic structures of an ecosystem and can show the organization of communities and how species overlap in the use of food resources, since these patterns of food exploitation are important for the study of ecology species (Poulin et al., 1994). Although this is one of the most studied aspects in research with carnivores, there is little research that addresses the bromatological composition of food items consumed. Thus, this study aimed to describe the diet of crab-eating fox roadkilled in two conservation units in the Southeast of the State of Pará, Brazil, by identifying the food items present in the stomach content and bromatological analysis of the food items.

Study area
The roadkill animals were collected in two conservation units, Carajás National Forest (6º4'14.972" S, 50º4'6.886" W) and Tapirapé Aquiri National Forest (5°35'52" and 5º57'13" S, 50°01'57" and 51°04'20" W), both located in the Amazon rainforest, Southeastern of the State of Pará, Brazil. These are part of a mosaic of protected areas, in the municipality of Parauapebas, state of Pará, Brazil ( Figure 1). The predominant vegetation is classified as open ombrophilous forest, with local variations, associated with changes in relief. The dense ombrophilous forest different food items were separated and identified at the least inclusive taxonomic level whenever possible. The identification of plant items was carried out by comparison with the reference material deposited in the scientific collection of the herbarium of Vale Zoo and Botanical Park. The items of animal origin were identified using identification keys and also compared to reference collection of the zoological collections available at Vale Zoo and Botanical Park and UFRA, both in Parauapebas, Pará, Brazil.

Bromatological analysis
The analyzis were performed in the Animal Nutrition Laboratory of the UFRA, Parauapebas, state of Pará, Brazil. Stomach content samples were thawed and then pre-dried in an oven at 55 o C for 72 hours. Then they were sent to grinding, in a Willey mill using a sieve with 1 mm holes. The samples were then placed in closed pots and identified for further analysis to determine the composition of dry matter (DM), mineral matter (MM), crude protein (CP), ether extract (EE) and crude fiber (CF) (Silva and Queiroz, 2006).

Data analysis
Food items were quantified according to their frequency of occurrence (FO) and percentage of occurrence (PO). The presence of a certain item in a stomach was considered an occurrence and the FO was calculated using the formula: FO = (ni / N) x 100, where FO is the frequency of occurrence of food item i in the sample; ni indicates the number of stomachs in the sample that contains the food item i; N corresponds to the total number of stomachs with content in the sample. The percentage of occurrence (PO) was calculated using the formula: PO = (Ni / ΣNi) x 100; where Ni represents the number of occurrences of each item, and Σni represents the sum of the occurrence of all items, which indicates the importance of each item in the diet (Tófoli et al., 2009).
The determination of dry matter (DM), crude fiber (CF), ether extract (EE), crude protein (CP) and mineral matter (MM) were submitted to descriptive statistics. The nitrogenfree extractive (NFE) was obtained by subtracting the sum of CP, CF, EE and MM from 100 (expressed as a percentage of DM). To predict the metabolisable energy (ME) of the diet, the modified Atwater factors for dogs and cats were used (NRC, 2006), according to the formula ME (Kcal/100g The maintenance energy requirement was calculated according to the NRC (2006), using the formula K x (BM) 0.75; where K represents the correction factor for different types of activity, and BM the body mass in Kg. The correction factor used was active adult dogs with an average need of 130.

Results
The body size, tail size and body mass averages of the crab-eating foxes studies were 68.8 cm ± 7.3, 29.3 cm ± 3.5 and 5.19 kg ± 1.25, respectively.

Description of food items
The stomach contents presented exclusive items of animal origin (24%), plant items (18%), and both animal and plant items (59%). A total of 29 food items was found in the crab-eating fox diet, 07 of which were of plant origin and 22 of animal origin, totaling 57 occurrences (Tables 1 and 2, Figure 2). Among the food items identified, gramineae (Poaceae) were the most consumed by the animals studied, with a FO of 41.18%, and PO of 12.28%. Within the items of plant origin, species such as Byrsonima crassifolia, Physalis angulata and Mangifera indica were observed. Byrsonima crassifolia and Physalis angulata presented a FO of 11.76%. In addition to these plants, species such as Bellucia grossularioides and Cecropia sp. were found with a FO of 5.88% each.
Among animal items found in the stomach contents, invertebrates and vertebrates represented 9% and 52%, respectively, of the total items. Among invertebrates, specimens of the order Odonata and family Scolopendridae were the most common with FO of 11.76% and 5.88%, respectively. Mammals, reptiles, amphibians and birds were examples of vertebrates found in the stomach content of the Amazon crab-eating fox (Figure 2).

Bromatological analysis
The nutritional contents found in the stomach content of Cerdocyon thous were: The dry matter (DM) ranged from   14.21% to 23.87%, with an average of 18.85% ± 3.62; crude protein (CP) varied between a minimum of 18.10% and a maximum of 35.12%, with an average of 27.75% ± 7.15; the ether extract (EE) between a minimum of 1.22% and a maximum of 7.28%, with an average of 3.52% ± 2.32, and the crude fiber (CF) between a minimum of 49.99% and a maximum of 76.89%, with an average of 62.76% ± 9.09. The mineral matter (MM) ranged from 0.032% to 0.080%, with an average of 0.059% ± 0.016. The nitrogen-free extractive (NFE) was 5.91%, and the metabolisable energy (ME) was 141.82 kcal/100g of food. Considering the average metabolic weight of the animals studied, it was possible to verify the maintenance energy requirement of 447.01 kcal/day.

Discussion
The diet of crab-eating fox (Cerdocyon thous) has been reported in several Brazilian geographic areas such as Cerrado (Juarez and Marinho-Filho, 2002), southeastern Brazil (Gatti et al., 2006), southern Brazil (Rocha et al., 2004;Pedó et al., 2006;Rocha et al., 2008), Pampa (Bossi et al., 2019), and Caatinga (Dias and Bocchiglieri, 2016;Souza et al., 2021). However, this is not specific to the Amazon basin, because, generally this animal is not common distributed in this biome (Facure et al., 2003), being this the first report on diet of the crab-eating fox that inhabits the Brazilian Amazon.
In the region analyzed in this study, the crab-eating fox showed a generalist feeding habits, with an omnivorous diet, without food preference, being composed of invertebrates, amphibians, reptiles, birds, small mammals, fruits and grasses. This feeding habit observed in the stomach content of C. thous corroborates with that reported for crab-eating foxes that inhabit the Atlantic Forest (Facure et al., 2003), southern Brazil (Rocha et al., 2004;Pedó et al., 2006;Rocha et al., 2008), Cerrado (Juarez and Marinho-Filho, 2002), and Caatinga (Dias and Bocchiglieri, 2016). This diversity in the feeding habit can be explained by the fact that the animal consumes the most abundant food resources (Courtenay and Maffei, 2004;Pedó et al., 2006).
In this study, the consumption of Poaceae leaves was higher than other kinds of leaf, and this may be related to the function of aid food digestion of the animal (Motta-Júnior et al., 1994;Rocha et al., 2004;Pedó et al., 2006). Although gramineae represent little or no importance for energy return (Dietz, 1984), they are usually common items found in the crab-eating foxes diet (MacDonald and Courtenay, 1996;Rocha et al., 2004;Pedó et al., 2006;Dias and Bocchiglieri, 2016).
The food content of vegetable origin was also characterized by the presence of fruits such as murici fruits (Byrsonima crassifolia), mess apple (Bellucia grossularioides) and cape gooseberry (Physalis angulata), demonstrating the species potential for seed dispersal, which emphasizes its ecological importance in the ecosystems in which it is inserted (Courtenay and Maffei, 2004). Dias and Bocchiglieri (2016) reported that the fruits are an important item in crab-eating fox diet. Cazetta and Galetti (2009) claimed that the crab-eating fox act as a secondary disperser of Eugenia umbelliflora (Myrtaceae), being important for the recruitment of plants in a forest of Restinga in southeastern Brazil. The crab-eating fox was also considered as seed disperser in the Brazilian Caatinga (Souza et al., 2021).
The presence of invertebrates was uncommong in the stomach contents of the animals in this study, and probably, it was associated more with accidental ingestion than a preference for this item, since the species feeds on insects, while looking for other foods. Facure et al. (2003) claimed that insects may be ingested accidentally with fruit or carrion.
Mammals and reptiles corresponded to the most constant groups of vertebrates in the stomach content of crab-eating fox of this study. The presence of mammals such as armadillo (Dasypus novemcinctus) and opossum (Didelphis marsupialis) in the crab-eating fox diet reflects an opportunistic habit of the species (Pedó et al., 2006;Lemos and Facure, 2011). Facure et al. (2003) reported that opossums, rabbits and cavies were the largest prey of the C. thous in the Mantiqueira Range. Moreover, in this study, wild rodents were the most frequent and important items in the C. thous diet. Small rodent species were also the most important food item in the diet of the crab-eating fox in a suburban area of southern Brazil (Pedó et al., 2006). Therefore, the presence of exotic rodent such as Rattus rattus found as a food items was observed in this study, which is not in agreement with previous observations to the crabeating fox (Gatti et al., 2006;Pedó et al., 2006;Rocha et al., 2008). Gatti et al. (2006) reported the presence of Rattus rattus as a food item in raccoon (Procyon cancrivorus) scats, which is other Neotropical carnivore. Probably, this is can be explained by the opportunistic habit of the crab-eating fox which in anthropogenic environments, it may feed on these rodents, as the area of this study was close to human communities.
Reptiles such as snakes were the second most frequently consumed vertebrates in the stomach contents of the Amazon crab-eating foxes. Snakes have been identified in the diet of the crab-eating foxes (Juarez and Marinho-Filho, 2002;Facure et al., 2003;Gatti et al., 2006;Rocha et al., 2008). In addition, there are studies that claimed that this Neotropical canid can prey on snakes. Gonzalez et al. (2016) reported a predatory event of C. thous upon the water snake Erythrolamprus miliaris. Silva et al. (2018) also described a cooperative predation by two crab-eating foxes upon an adult short-tailed boa (Boa constrictor amarali) at a Brazilian Cerrado.
Due to the fact that the crab-eating foxes withstand anthropic interference in their habitat, in some cases they take advantage of the greater supply of food provided by close contact with the human species (Courtenay and Maffei, 2004). On the other hand, Lima (2017) verified the food preference of Cerdocyon thous in remnant of the Atlantic Forest and demonstrated the effect that the nutritional quality of the resources and the competition for interference of a co-specific exercise on the selective feeding behavior and, despite being considered a generalist species in its feeding habit, the crab-eating fox has a preference for the item with the greatest nutritional value.
According to the National Research Council (NRC, 2006), to obtain the nutritional value of food, food composition tables can be used or estimated using bromatological analysis and the use of appropriate equations. However, studies with bromatological analysis of the stomach contents of wild animals are scarce, making comparisons difficult. Bromatological analyzes are necessary, as they provide subsidies to infer the nutritional requirements of species in free life, aiming to contribute to the conservation of species kept in captivity, through the formulation of specific and appropriate rations for each stage of development of the animal in captivity.
For a species of omnivorous habit, such as the crabeating fox, the need for animal protein, lipids and energy can be a determining factor for its survival and performance as a predator, since they are nutrients that are difficult to acquire in nature, similar to other carnivorous animals (Clauss et al., 2010).
In the present study, the values observed for crude fiber (CF) are considered high, when compared to the values for domestic dogs (NRC, 2006), in which diets must contain a maximum of 8% CF. Apparently, the high CF content may be associated with the constant presence of gramineae in the stomach content, combined with the low digestion capacity of the fiber in the stomach and the low nutritional value of tropical forages. These data show the influence of this low dry matter digestibility on the reduced content of crude protein (CP) and minerals (Euclides, 1995).
The stomach contents of the bush dogs (Spheotus venaticus) run over in the Carajás National Forest presented crude protein (CP) of 38.35%, ether extract (EE) of 6.21%, and mineral matter (MM) of 55.43% (Nobre, 2016). The lowest level of CP and MM obtained in the present study in relation to reported to the bush dogs, may be associated with the different feeding habit of the species, since the bush dog in nature is strictly carnivorous, thus presenting higher levels of CP and MM, while the crab-eating fox is an omnivorous canid, having a more diversified diet.
The composition of the diet of the free living crabeating fox showed to be diversified, presenting food items belonging to the different taxonomic groups, having a greater diversity of items of animal origin. Although the crab-eating fox of this study inhabits a characteristic biome, the Amazon rainforest, its diet were not very different from those reported for this species in other Brazilian geographic regions (Juarez and Marinho-Filho, 2002;Facure et al., 2003;Rocha et al., 2004;Gatti et al., 2006;Pedó et al., 2006;Rocha et al., 2008), suggesting that although the crab-eating foxes may inhabit different biomes, their diet is very similar and the crab-eating foxes found in the Amazon basin are generalists with an omnivorous diet. Furthermore, through these findings, it is also suggested which dietary components are pivotal for nutrition of animals kept in captivity, since in captivity, the crab-eating foxes receive a food similar to the domestic dog.