Diet of Didelphis albiventris Lund, 1840 (Didelphimorphia, Didelphidae) in two periurban areas in southern Brazil

Didelphis albiventris Lund, 1840 is one of the least known Brazilian marsupials with respect to its feeding habits. Since the white-eared opossum is omnivorous, it occupies several niches and may be found in urban areas as well as in forest fragments. Current analysis aimed at determining and comparing the diet of D. albiventris in two periurban areas of the municipality of Viamão, a conserved forest remnant and a degraded area resulting from agricultural activities. Captures of D. albiventris were carried out for a period of 30 months by a sampling effort of 504 trap-nights. Forty fecal samples were collected from 45 captures corresponding to 18 specimens. Diet was determined by fecal analysis. Fruit was the most representative item, at an occurrence of 77.5%, comprising six species of plants, followed by invertebrates (62.5%) and vertebrates (37.5%). Didelphis albiventris displayed a frugivorous-omnivorous habit, with no difference between the sexes with respect to composition and frequency of the items consumed. Since there was a greater diversity of food items and fruits in the more conserved area, seasonal variations in the diet could be detected. Generalist species, such as D. albiventris, may have an important role in the recovery of degraded areas through the dispersal of seeds of species of initial successional stages.


Introduction
Marsupials currently exist in only two regions of the world, on the Australian and American continents (REDFORD; EISENBERG, 1992). They are considered small-to medium-sized animals which may occupy a variety of habitats. In fact, they are widely represented and distributed throughout the Neotropical region, mainly in forest environments (EMMONS;FEER, 1997). Marsupials are extremely diversified animals with regard to their habits, with a predominantly omnivorous diet, although many researchers attribute the feeding tendencies of frugivory, insectivory and carnivory to the group (SANTORI; ASTÚA DE MORAES, 2006). Thus, they perform important ecological functions, such as controlling populations of vertebrates and invertebrates, and dispersing seeds (BARBOSA et al., 2012). In tropical forests, endozoochory is the principal form of seed dispersal, which reveals an important interaction between plants and animals (CHARLES-DOMINIQUE, 1986). According to Uhl (1997) and Jordano et al. (2006), the conservation and regeneration of different types of vegetation are highly dependent on seed dispersal by mammals.
Didelphid marsupials are traditionally known as small mammals with nocturnal, solitary and generalist diet habits, consuming small vertebrates, invertebrates and fruits (REIS et al., 2006). Among the didelphids, D. albiventris is among the least studied species with regard to its feeding habits (SANTORI; ASTÚA DE MORAES, 2006). D. albiventris has a wide geographic distribution in South America, where it occurs in various biomes in Brazil, such as the Atlantic Rain Forest, Cerrado, Pantanal, Caatinga and Campos Sulinos (REIS et al., 2009). Since they may be found in small forest fragments and in places with a high degree of ecological disturbances, the white-eared opossums are capable of adapting themselves to fragmented habitats and to those modified by anthropogenic activities (FERNANDEZ;PIRES, 2006). Owing to their frugivorous-omnivorous and generalist habits, they are directly involved in the dispersal of seeds of many plant species (CÁCERES; MONTEIRO-FILHO, 2007), which, according to Cantor et al. (2010), are swallowed whole, with a viability of up to 71% after defecation.
So that knowledge on the still poorly known feeding habits of one species of Didelphis could be deepened, current study determined and compared the diet of D. albiventris in two periurban areas in southern Brazil, or rather, a more preserved habitat and another altered by agricultural activities. The white-eared opossum is a widespread species which presents a generalist diet, an opportunistic habit, as well as the capacity to occupy disturbed environments. Knowledge on diet of D. albiventris might help understand the importance of the species for seed dispersal and, consequently, its role in the restoration of degraded areas.

Study Areas
The study was conducted in two areas in the municipality of Viamão, Rio Grande do Sul State, Brazil ( Figure 1). One of the areas, known as Morro do Coco, is a remnant of a semidecidual seasonal forest, part of the Atlantic Rain Forest (30°16'15"S and 51°02'54"W). The area, covering approximately 142 ha, lies on the edge of Lake Guaíba and is somewhat connected with other forest areas, such as the hills of São Pedro and Extrema, and with conservation units, such as the Reserva Biológica do Lami José Lutzenberger and Parque Estadual de Itapuã. Its surroundings comprise small farms and ranches and small urban centers.
The other area comprises a rural property close to the boundary of the APA do Banhado Grande (29°58'11.37"S and 50°58'1.34"W). The region is an ecotone characterized by a mosaic of forest formations of the Atlantic Rain Forest (semidecidual seasonal forest), costal pioneer formations and fields (TEIXEIRA et al., 1986). The area lies close to the river Gravataí, where marshes, fields and remnants of a semidecidual seasonal forest are predominant. It is an area highly disturbed by anthropic impacts, which include houses, ranches, farms, eucalyptus plantations and other city-connected activities, such as hunting, domestic animals and factories.
The region's macroclimate is Cfa type (subtropical humid), according to Koeppen´s classification, with mean annual temperature at 19.5°C, cold winters with minimum temperature at 3°C, and warm summers with maximum temperature at 41°C; annual rainfall 1347 mm, distributed uniformly throughout the year (MORENO, 1961). There are four distinct seasons (autumn, winter, spring and summer) in the southern State of Rio Grande do Sul, which differ by photoperiod (varying from about 10h on June 21, at the beginning of winter, to 14h on December 21, at the start of summer) and by rain regime and temperature variations (HASENACK; FERRARO, 2006;LIVI, 2006).

Sampling of Scats
Capture of marsupials of the species D. albiventris was carried out with 10 live traps placed on the ground, in the understorey and at the tree stratum (between 1.5 and 6 m high). The traps were set monthly, from one to two consecutive nights, close to fruiting trees and streams, on paths, and inside and on the border of forest remnants. They were checked daily in the morning. The captured animals were labeled with numbered tags and their sex determined (authorizations by SISBIO 12094-2 of 2008 and 24673-2 of 2010). In Morro do Coco, the expeditions were conducted from January 2009 to June 2010, and from April to June 2012, totaling 21 months. In APA do Banhado Grande, the samplings covered the period between March and November 2011, totaling nine months. Data collection on the diet of D. albiventris, in both areas, totaled 30 months. The capture effort corresponded to 390 trap-nights in Morro do Coco and 114 trap-nights in APA do Banhado Grande. Banana, peanut butter, bacon, sardines and pineapple were used for bait.

Diet Analysis
Samples of feces were collected from the floor of the live traps and placed in labeled plastic pots for preparation and laboratory analysis. Each fecal sample consisted of the total feces produced by one specimen in one night. The samples were taken to the laboratory and dried in an oven at 60°C. The fecal content was sorted out and the food items present in the feces were identified with the help of a stereomicroscope and specialized literature (LORENZI, 1992), as well as on the basis of comparison with reference material collected in the study area.
The frequency of occurrence, which resulted from the ratio between the frequency of the particular item and the total number of fecal samples, evaluated the representativeness of the different items found in the diet. Differences in diet or diet diversity between the two areas were compared by Levin's Standardized Index (B A ), which attributed greater weight to food items that were more abundant in the diet, according to Krebs (1998) and Caceres (2002), varying from 0 to 1. Values closer to 1 meant that the items recorded were close to maximum diversity. The significance of the differences found was tested by the non-parametric Kruskal-Wallis (H), Mann-Whitney (U) and chi-squared (χ 2 ) tests. Yates  (HAMMER et al., 2001).

Results and discussion
In the two areas under analysis, there were 45 captures of D. albiventris, of which 33 occurred in Morro do Coco and 12 in APA do Banhado Grande. Fourteen specimens, 8 females and 6 males, among the 45 captures, were recorded in Morro do Coco, and 4 specimens, 2 males and 2 females, in APA do Banhado Grande. Thus, out of the total captures, 27 corresponded to recaptures on different expeditions. As a result, 28 samples of feces were collected in Morro do Coco and 12 samples in APA do Banhado Grande, totaling 40 samples. Therefore, five samples were discarded to ensure independence of data since they were obtained from specimens recaptured in the same expedition.
Didelphis albiventris showed a varied diet, composed of vertebrates, invertebrates and fruits, although fruits and arthropods were the main items in the diet, when the combined samples of the two areas are taken into consideration (Table 1) Glassman, representing items of plant origin, were the most frequent food items in the diet and comprised 77.5% of the samples. Invertebrates were the second most found item in 62.5% of the fecal samples, followed by vertebrates (37.5%). Cáceres (2002) obtained a similar result in forest fragments with araucaria in a periurban region in the State of Paraná, Brazil, although invertebrates were the main food item found in the feces of D. albiventris, followed by fruits and vertebrates. Consequently, the diet of D. albiventris was predominantly frugivorous-omnivorous, as observed in other studies (CÁCERES, 2002;CASSELA;CANTOR et al., 2010). The method of analysis of fecal content proved to be efficient to identify most items in the feces of D. albiventris, corroborating results obtained by other authors (CÁCERES, 2002;CÁCERES;MONTEIRO-FILHO, 2001, 2007CÁCERES et al., 2009). The diversity of food items differed between the two study areas and was higher in Morro do Coco (B A = 0.6208), the more preserved area, when compared to that of APA do Banhado Grande (B A = 0.3223). In fact, there was a high intake of fruits (82%) and arthropods (68%), whereas the occurrence of these items in the degraded area was lower. The landscape of Morro do Coco comprised more forest environments, surrounded by conservation units and better conserved areas, which implied the availability of a wider variety of resources and justified a more diverse and predominantly frugivorous diet. The area inside APA had received a greater impact of anthropogenic activities, leading to a matrix with more fields than forests and, consequently, a less frugivorous diet. Nonetheless, differences in sampling effort could also have contributed to this result. Although fruit diversity in the fecal samples from Morro do Coco (B A = 0.6286) was about three times greater than in the impacted area (B A = 0.1951) belonging to APA do Banhado Grande, the occurrence of fruits, in relation to occurrence of other different items in the scats, did not differ significantly between the two areas (χ 2 = 2.212, df = 1; p = 0.1369). Opossums fed mainly on fruits in both areas (86 and 58% occurrence in scats, respectively, according to Table 1).
In Morro do Coco, seeds of five plant species were found in 75% of the samples. Seeds of Ficus cestrifolia Schott ex Spreng. and Coussapoa microcarpa (Schott) Rizzini were the most frequent in samples from Morro do Coco (29% occurrence for both species). Cereus hildimannianus K. Schum. was the second highest occurrence (11%), while Solanum sp. and Ficus luschnathiana (Miq.) Miq. seeds were the least frequent in the samples analyzed (both with 4% occurrence). The large seeds of S. romanzoffiana were recorded only through the fibers present in the feces, with a 29% frequency. Perhaps the seeds had been expelled near the parent tree, as reported by Charles-Dominique et al. (1981) in a study on the diet of marsupials. In their evaluation on the interaction between S. romanzoffiana and local fauna, Silva et al. (2011) suggested that Didelphis aurita could disperse palm seeds by moving them at least over short distances during pulp removal. In APA do Banhado Grande, arthropods and fibers of the fruit of S. romanzoffiana were the most common items, occurring in 50% of the samples. Seeds had only 25% frequency where the only seeds recorded were those of F. luschnathiana and C. microcarpa (17 and 8% occurrence, respectively). Vegetative and reproductive parts of Poaceae and Asteraceae were found exclusively in this area, a fact also cited by Caceres (2002) for both D. albiventris and D. aurita, in fragmented areas of mixed ombrophilous forest in southern Brazil.
Although D. albiventris has a generalist diet that includes a great variety of food items, its importance in fruit intake (CHARLES-DOMINIQUE et al., 1981;CÁCERES;MONTEIRO-FILHO, 2001;CÁCERES, 2002) and its likely influence on seed dispersal (CANTOR et al., 2010) are gradually being acknowledged. However, little is known about the germination of seeds that pass through the intestinal tract. It is known that the germination of some seeds is facilitated by their passing through the intestinal tract of vertebrates, or rather, break of dormancy, reduction of pulp and fertilization by fecal matter (ROBERTSON et al., 2006;TRAVESET, 1998). Cáceres and Monteiro-Filho (2000) showed that most seeds remained intact after being ingested by D. aurita. Further, Cantor et al. (2010) found that most seeds were small sized and came from berry or drupaceous fruits with many seeds. In current analysis, seeds found in both study areas corroborated the above results, since most, besides being small-sized, remained whole after passing through the digestive tract. The smallest were the Ficus spp. seeds, which measured only 0.5 mm, while the largest, C. hildimannianus seed, measured approximately 4 mm. It should be noted, however, that breaking dormancy did not seem to be the greatest advantage for zoochoric plants. According to Jordano et al. (2006), the dispersal of seeds, a process that involves the removal and displacement of seeds from the vicinity of the parent plant to locations where predation and competition are lower, is fundamental in the life cycle of most plants, especially in tropical ecosystems.
Didelphis albiventris is the least studied of marsupial species with respect to its feeding habit.
Although it is known that its diet includes 21 fruit species , only S. romanzoffiana and Solanum sp., out of all the species of fruits mentioned in current study, are cited in the survey carried out by  for D. albiventris. Meanwhile, C. microcarpa and Ficus spp. were recorded in the diet of other marsupials by the same author. Cereus hildimannianus (Cactaceae), however, was not mentioned in any other study on the diet of marsupials and was recorded here for the first time. Cáceres (2002) demonstrated that more than half the fruits consumed by D. albiventris were from pioneer plants. Of the six species of fruits recorded in the samples from the two areas, Solanum sp., C. hildimannianus and Ficus spp. are considered pioneer (CHARLES-DOMINIQUE, 1986;FLEMING, 1986). Fig trees, particularly, are protected by law in Rio Grande do Sul, because of the numerous interactions that they establish with fauna and flora (especially epiphytes). Another relevant factor concerns the reporting of seeds of flora species only native to the state, although there are orchards with exotic fruit trees in the neighborhood of the areas under analysis, which revealed the ecological importance of D. albiventris as a disperser of native plant species.
At Morro do Coco, where sampling followed an entire annual cycle, a significant difference in the consumption of items between the seasons of the year (H = 12.71; p = 0.008267) was detected. Temporal variations in the diet were also found in other studies, generally associated with variations in food availability CANTOR et al., 2010;CEOTTO et al., 2009). Autumn differed from the other seasons with regard to the items in the fecal samples of D. albiventris and was the period with the highest number of samples collected (n = 13) and with the greatest variety of food items. Of the six fruit species recorded in the fecal content, five (C. microcarpa, F. cestrifolia, F. luschnathiana, C. hildimannianus and S. romanzoffiana) were present during this period, coupled to all invertebrate and vertebrate groups. Marsupials are considered opportunistic frugivores (CÁCERES, 2002;CÁCERES et al., 2009) since they consume resources available in the environment. This fact appears to indicate seasonality in the availability of fruits in the study area. In the semidecidual seasonal forest of the State of Paraná, Brazil, Mikich and Silva (2001) reported a fructification peak during the dry season (May and June), coinciding with autumn. According to Cáceres and Monteiro-Filho (1998), in a study conducted in a remnant of an altered mixed ombrophilous forest in the State of Paraná, D. albiventris and D. aurita accumulated body fat during autumn to endure winter, during which they lose weight. Although in current study winter was the period with fewer samples collected (n = 3), the samples included 50% of the fruits recorded in the area, including C. hildimannianus. Seeds of Solanum sp. were found exclusively in summer.

Conclusion
Didelphis albiventris showed a frugivorousomnivorous habit although fruits constituted an important part of their diet, both in diversity and frequency.
The highest diversity of food items in Morro do Coco, especially fruit diversity, indicates the good state of conservation of the area, comprising a more forested landscape, thereby offering a larger variety of resources. Accordingly, the frequency and biomass of other items such as seeds, arthropods and vertebrates were also higher in this area.
The greatest consumption of fruits and, consequently, the highest occurrence of seeds in the fecal samples of the individuals captured in Morro do Coco, revealed the importance of D. albiventris in the area, as a disperser of plant species of initial successional stages. Generalist species, such as D. albiventris, which are capable of adapting themselves to fragmented landscapes, might help in the regeneration of vegetation by the dispersal of seeds and thus have an important role in degraded areas, where specialized frugivores are absent.