Jaguarundi ( Puma yagouaroundi ) ( Geoffroy , 1803 ) ( Carnivora , Felidae ) food habits in a mosaic of Atlantic Rainforest and eucalypt plantations of southeastern Brazil

Food habits of jaguarundi (Puma yagouaroundi) (Geoffroy, 1803) (Carnivora, Felidae) were studied between November 2000 and November 2001, in a 24.9 km area of secondary Atlantic Rainforest and eucalypt plantation, in the Serra de Paranapiacaba, São Paulo State, Brazil. Analyses of 26 fecal and regurgitate samples, obtained over a stretch of 570.1 km, showed the consumption of 19 prey items and 74 prey occurrences. Small mammals were the most frequent food item (42.5%), followed by birds (21%), reptiles (14%) and medium-sized mammals (3%). The percent occurrence (PO) suggests that the diet consisted mainly of small rodents (30%) and birds (21%). We recorded for the first time the predation of Viperidae snakes by P. yagouaroundi. Although having a large list of items and range of dietary niche breadths (B sta = 0.76), our data show that jaguarundi prey mainly on small vertebrates (mammals, birds or reptiles), and even in tall tropical forests or eucalypt plantations, it preys mostly on animals that come to, or live on, the ground.


Introduction
The mesopredator release hypothesis predicted that the decline of large predators would result in the ecological release of mesopredators (small to medium-sized carnivores and omnivores), and that increased predation would result in higher mortality and local extinction rates of some prey species (Fonseca and Robinson, 1990;Crooks and Soulé, 1999).Large carnivores also play a decisive role in the maintenance of forest biodiversity controlling predation on plants; decreases on top predator populations cause an ecological instability, affecting biodiversity in the ecosystem (Terborgh, 1990;Terborgh et al., 1999;Terborgh et al., 2001).
Jaguarundi scats were identified as such by its hair ingested while grooming.Under a stereomicroscope, hair microstructure (hair medulla and cuticular scales) was compared to a reference collection from museums (Emmons, 1987;Olmos, 1993;Facure and Giaretta, 1996).
Mammal prey were identified by comparing hair found in the scats with hair and specimens previously collected in the study site and from museums; hair microstructure was also compared.Marsupial teeth were compared to museum specimens; rodents were identified through molar teeth analysis by A. Percequilho, a specialist researcher from MZUSP (University of São Paulo Zoology Museum).Except for one species, bird claws and feet helped identification, by comparison, as Passeriformes and non-passerines.Scales were used to identify reptiles by a specialist.
Scat and regurgitation contents were expressed in terms of frequency of occurrence (FO), the number of times that each item was found divided by the total number of scats, which shows how frequent a species is in the diet.Percent occurrence (PO) is frequency of occurrence divided by the amount of all occurrences of the prey items, and shows the importance of each item in the jaguarundi diet (Bisbal, 1986;Facure and Giaretta, 1996;Wang, 1999;Guerrero et al., 2002).
Differences in number of occurrences and prey occurrences in the wet and dry season were compared by chi-square tests using BioEstat 3.0 (Ayres et al., 2003) and the comparison between different studies was done with Cluster Analysis using Euclidean Distances, using STATISTICA 6.0 (StatSoft, 2001).Invertebrate preys were not included in the Cluster Analyses.
Average prey weights were obtained from the literature (Gans, 1980;Emmons and Feer, 1997;Eisenberg and Redford, 1999;Magalhães, 1999).Identified preys were classified according to weight in two groups: 1) below 1 kg and 2) more than 1 kg.Since the largest identified bird was a tinamou (Crypturellus obsoletus) (Temminck, 1815) (Tinamidae), which can weigh almost 0.5 kg, all birds, as well as snakes, were included in the first group.
Here we describe the food habits of jaguarundi in a secondary Atlantic Rainforest and mosaic of eucalypt plantation, in southeast Brazil.

Materials and Methods
This study was conducted at João XXIII Farm (JF) (23° 53' 09''-23° 56' 29'' S and 47° 42' 30''-47° 40' 08'' W), a private property of Eucatex S.A., in the Pilar do Sul municipality, São Paulo State, southeast Brazil, on coastal Atlantic Forest.The area is contiguous to the largest rainforest conservation tract in São Paulo State, which comprises Conservation Units, such as Carlos Botelho, Intervales and Alto Ribeira (PETAR) State Parks.The 24.9 km 2 JF comprises 10.5 km 2 of secondary dense tropical rainforest, semi-deciduous rainforest and gallery forest, of which 6.1 km 2 is continuous forest, 13.1 km 2 is Eucalyptus saligna (Smith) (Myrtaceae) plantation, with native understory, and 1.3 km 2 is made up of deforested areas.The climate is Cfa (subtropical humid) according to Köppen's classification, with average temperatures around 22 °C in the hottest month and 15 °C in the coldest one.Annual rainfall during the study period was 1,722 mm.
Jaguarundi diet was studied by fecal and regurgitation analysis.Monthly, between November 2000 and November 2001, we collected felid scats and regurgitation samples along dirt roads, fire-breakers and trails.Collected scats were stored in plastic bags and dried.In the laboratory, the scats were rinsed with water over a 63 µm mesh screen and dried.Hair, bones, teeth, claws, The 26 jaguarundi samples had 19 identified food items, totalising 74 occurrences (Table 1).Grass appeared in three (12%) samples in the wet season but was not included in the occurrences.All scats contained mammals, 62% birds, 58% invertebrates and 27% reptiles.The number of occurrences in a sample ranged from 1 (only rodents) to 6. Once, two individuals of the same species (Brucepattersonius soricinus) (Hershkovitz, 1998) (Muridae) occurred in one scat.Small mammals comprised 42.5% of the occurrences, followed by birds (21%), reptiles (14%) and larger mammals (3%).Invertebrates made up 20% of occurrences (Table 1).

Discussion
The present study represents the largest sample of jaguarundi scats from Brazil, and the third from its geographic range (Table 2).Also, in Pilar do Sul, jaguarundi had the more diverse prey list (19 items), and at least three items recorded for the first time (brocket deer, naked-tailed armadillo and pit-vipers).
on fish (Table 2).Furthermore, by cluster analyses (Figure 2), we found that diet at Pilar do Sul is also closer to the study at CBFR, in south-central Belize (Konecny 1989).As in the four other studies, rodents were the most important item in Pilar do Sul (Table 2).Larger mammals, naked-tailed-armadillo and brocket deer had low frequency at Pilar do Sul.Medium-sized mammals such as paca (Agouti paca) (Linnaeus, 1766) (Agoutidae) and brown-throated-three-toed-sloth (Bradypus variegatus) (Schinz, 1825) (Bradypodidae) were recorded elsewhere (Wang, 1999).The brocket deer in the jaguarundi diet at Pilar do Sul could be predation of young or carrion consumption; in the same month we recorded two deer in puma's diet (Rohe, 2002).The predation of brocket deer was already mentioned by Cabrera and Yeppes (1940), but they did not register it in the diet.The largest prey recorded for jaguarundi was cattle (Bos taurus) (Linnaeus, 1758) (Bovidae) in Mexico (Guerrero et al., 2002).
Birds appear in the jaguarundi diet in eight studies (Table 2), but were the most important item only in Venezuela (Bisbal 1986).Reptiles are also frequent (seven studies, Table 2), and appeared as the most frequent item in Venezuela (Mondolfi 1986, Table 2).Only two studies documented snakes (Facure and Giaretta, 1996;Guerrero et al., 2002).This is the first record of pit viper (Family Viperidae) consumption by a jaguarundi and the first with snakes more frequent than lizards.
As in the study conducted by Mondolfi (1986) at different localities in Venezuela, jaguarundi at Pilar do Sul seem to have an intermediate diet (Table 2), more specialised than at CVRDFR (Facure and Giaretta, 1996) and at SCNP in northeastern Brazil (Olmos, 1993), and more generalised than at CBFR in Belize (Konecny 1989), MGAM in southeastern Brazil (Manzani and Monteiro-Filho, 1996), different localities in Venezuela (Bisbal, 1986) and at Tenacatita in Mexico (Guerrero et al., 2002).
Although having a diverse list of food items and range of dietary niche breaths, our data show that jaguarundi mainly prey on small vertebrates (mammals, birds or reptiles) and suggests that in tall tropical forests (or eucalypt plantations with native understory), it preys mostly on animals that come to the ground.Considering the fragmentation process in the Brazilian Atlantic Forest, more information about the diet of medium-sized carnivores, like P. yagouaroundi, and their role in this ecosystem are urgently needed.These studies are essen-

Figure 1 .
Figure 1.Seasonal occurrence of jaguarundi food items in Pilar do Sul, southeast Brazil.

Table 1 .
Percentage of occurrence, month and identification form of food items from jaguarundi scats and regurgitations at Pilar do Sul, southeast Brazil.