Diet of Didelphis aurita and Micoureus paraguayanus and the fruit availability in a semideciduous Atlantic forest in Southern Brazil

This study analysed the effects of diets and resource partitioning among Didelphis aurita and Micoureus paraguayanus, and their relationships with food availability in a semideciduous Atlantic Forest of southern Brazil. Species were separated by sex, age and reproductive condition to compare with dietary data. The fruit availability in the area was quantified bimonthly through collection of fleshy fruits and flowers available on the ground of the study area. Individuals of D. aurita trapped in the canopy vs. those trapped on the ground, or even regarding age and sex classes, did not differ in regards to their diet composition. Animals studied were heterogeneous in relation to their diet composition, being significantly different from each other (p = 0.03). The largest diversity indexes observed for both species in faeces and in the environment occurred during the rainy season. Both species presented insects in their diets, but with natural tendency to consume fruits or vertebrates (D. aurita). Rhythms of fructification of Piper sp. correlated with consumption by D. aurita (rs = 0.90, p = 0.03) and not by M. paraguayanus (rs = 0.40, p = 0.60). There were tendencies where independent young and lactiferous females of D. aurita correlated to the peaks of fruit availability during the rainy season.


Introduction
The neotropical marsupials are generally known as small omnivore mammals, consuming all food types, and these animals are not specialists in any food category (Santori et al., 1995;Astúa de Moraes et al., 2003).Nevertheless, studies have shown didelphid marsupials as presenting divergent diets from the traditional omnivory, such as trends to frugivory (Caluromys philander), carnivory (Lutreolina crassicaudata) or insetivory (Metachirus nudicaudatus) (Charles-Dominique et al., 1981;Atramentowicz, 1988;Leite et al., 1996;Fonseca et al., 1996;Carvalho et al., 1999;Vieira and Astúa de Moraes, 2003;Astúa de Moraes et al., 2003).However, invertebrates are currently the main food resources for didelphid marsupials (Fonseca and Kierulff, 1989;Busch and Kravetz, 1991;Carvalho et al., 1999;Talamoni et al., 1999;Pinheiro et al., 2002), classified as being in the majority insectivore-omnivore (Robinson and Redford, 1986;Fonseca and Kierulff, 1989;Fonseca et al., 1996).Didelphis aurita and Micoureus paraguayanus were more insectivoreomnivore, consuming food resources according to their proportion and availability in the environment (Leite et al., 1996;Cáceres et al., 1999;Cáceres, 2003;Carvalho et al., 2005).However, D. aurita has been considered as mostly omnivorous (Santori et al., 1995;Astúa de Moraes et al., 2003), but sometimes it is characterized as being mostly frugivore as well (Stallings, 1989;Cáceres and Monteiro-Filho, 2001;Cáceres et al., 2009).On the other hand, a more detailed analysis of the diet of M. Paraguayanus have indicated that this specie consumes, in general, a few vertebrates prey, a reasonable number of insects and sometimes large quantities of a select group of fruits (Cáceres et al., 2002;Casella and Cáceres, 2006).One of the factors that differentiate niches in a guild is the foraging strategy, which may be determined by body size and form.Differences among species in niches they occupy may allow the coexistence of a great number of species in communities (Medellín, 1991;Leite et al., 1996;Vieira, 2006).Besides, the vertical use of space by marsupials has considerable importance in the spatial separation of sympatric species.This increases the possibility of resource partitioning among them rather than competition (Charles-Dominique et al., 1981;Julien-Laferrière, 1991).Most studies carried out in Brazil referring to diet or food partitioning among marsupials are related to biomes other than semideciduous forest, such as restinga (Santori and Astúa de Moraes, 2006), mixed ombrophyll forest (Cáceres and Monteiro-filho, 2001) and dense ombrophyll forest (Leite et al., 1996;Carvalho et al., 1999;Cáceres et al., 2002;Pinheiro et al., 2002;Cáceres et al., 2006).Data on diet for Didelphis and Micoureus, for example, are scarce and normally available for the dense Atlantic forest only (Leite et al., 1996;Carvalho et al., 1999).So, it has been mentioned the need of studying interactions among the diets of marsupials for better comprehension of their strategies of resource partitioning.Therefore, this study aims to analyze the diet of the black-eared opossum Didelphis aurita (Wied-Neuwied 1826) and of the woolly mouse opossum Micoureus paraguayanus (Tate, 1931) and it also intends to obtain information on relationships between the diet and the food availability.

Study area
The National Park of Iguaçu is located in southern Brazil, near Misiones, Argentina, at coordinates 25º39'10.95''Sand 54º26'14.53''W.The National Park of Iguaçu (PNI) is considered the largest Brazilian Conservation Unit in the domain of the Atlantic Rain Forest, with approximately 185 000 ha, and it is also one of the last reminiscences of that vegetation in the south of the country.It is represented by the semideciduous seasonal forest (in the most part of the area), mixed ombrophyll forest (north-eastern portion) and pioneer alluvial formation (IBAMA, 1999).The rainy period is from October to January and the dry period is from June to August.The mean annual temperature ranges between 18 and 20ºC, being 34 to 36º C the maximum monthly temperatures and -8 to 4ºC the minimum temperatures.The annual rainfall varies from 1500 to 1750 mm (data from SIMEPAR, the Meteorological System of Paraná).The hotter period is from November to April, and the colder period is from May to October.This study included a 4 ha-grid inserted in the semideciduous seasonal forest, with an altitude of approximately 400 m above the sea level.

Captures and marking
The capture of marsupials was done by using 150 traps: 130 of wire mesh (90 large of 40x16x16 cm and 40 small of 20x15x15 cm in size) and 20 Sherman type (40x12x12 cm).The 100 wire traps (90 large and 10 small) were set in the canopy and other ones were set on the ground.The traps were set M. paraguayanus (r s = 0.40, p = 0.60).Houve tendências onde jovens independentes e fêmeas lactantes de D. aurita correlacionaram com os picos de disponibilidade de frutos disponíveis durante a estação chuvosa.
Diet of Didelphis aurita and Micoureus paraguayanus and the fruit availability in a semideciduous Atlantic forest in Southern Brazil in November 2004, January, March, May and July 2005, being the ground traps activated in November.The captures occurred during 10 consecutive days in field phase, with a total of 50 days of sampling collection.Ten transect lines were set 20 m apart from each other in the continuous forest, with 15 traps in each transect line.In each transect line 10 traps were placed in the canopy (20 m apart) and five on the ground (40 m apart).The canopy traps were lifted by mean of ropes (10 to 20 m high), positioned in wood platforms connected to trees (following Charles-Dominique et al., 1981).The traps were baited randomly with bacon, banana, pumpkin or pineapple, mixed with codfish liver oil.They were checked every morning and new baits were placed when necessary.The captured animals were identified according to species, weighed (in grams), sexed, checked for reproductive condition (pouch youth and milk production), and aged (by verifying dentition in accordance to Tyndale-Biscoe and Mackenzie, 1976).For individual identification, animals were marked with combinations of holes in the ears (Monteiro-Filho, 1987) and released.

Faecal samples and diet analyses
The faeces of the animals captured were collected on the floor of the cages after the release of the animals.The same faeces were stored in plastic bags for a posterior analysis in laboratory.Samples were washed with water by using a mesh sieve of 1 mm and analyzed in a stereoscopy.Seeds contained in each sample were separated at the species level, identified by using a guide and compared to a reference collection.The vertebrate items were identified at the class level based on the presence of feather, bones, fur and/or teeth.Insects were identified up to the order level through fragments of their body and legs.To analyze possible differences between the diet compositions of the two species of marsupials, a multivariate analysis was used through the main coordinates (PCoA), the statistical program used was Past 1.89 (Hammer et al., 2001).The ordinations (matrix of the association) were done based on the presenceabsence of food items in samples and calculated by the similarity index of Sorensen, known as Coefficient of Community (Costa et al., 2005).This index gives more importance to rare items in the samples, considering that more abundant data usually occur in many localities having a small contribution to these differences.(Costa et al., 2005).This test was used to compare the diet among individuals from both species.Due to the higher amount of samplings, comparisons were also performed for diet variations in D. aurita, taking into account age classes (young vs. adults) and sexes (male vs. female).Faecal samples from the same individual in consecutive days of trapping were excluded from the analyses; avoiding pseudoreplication occasioned by recaptures in short time intervals (Cáceres, 2000).Thus, there was made an effort to the independence of the faecal samples when they occasionally came from the same animals.In the interspecific analyses, when there was the capture of only one of the species in a given field phase, the data were excluded in order to standardize comparisons between species.

Fruit availability in the environment
The fruit availability in the study area was quantified by the collection of zoochoric fruits available on the ground along the 10 transect lines set for animal captures, all 1 m width, during the periods of capture and monitoring of the animals.Fruits and flowers were counted and placed in plastic bags and then weighed (in grams).The number of species of fruits and flowers was also evaluated in each field phase.Bushes of Piper sp. were evaluated due to previously known importance of the marsupial diets (Medellín, 1994;Henry, 1999;Cáceres et al., 2002), and compared with the presence of respective seeds in faeces.Bushes were counted in the study area within the same transect lines used for fruit availability.For all statistical analysis, the alfa value was 0.05.

Relationships of diet and fruit availability
In order to evaluate the relationship among the quantity of fruits (estimated by the seed number) consumed by D. aurita and M. paraguayanus and the total mass of zoochoric fruits available in the environment, the Spearman Correlation (r s ) was used.Such test was also used to verify the relationship between the number of seeds of Piper sp.present in faeces of the two marsupial species in relation with the fruiting plants of Piper sp.observed, besides the relationship between the diversity of seeds (calculated by the index of Shannon) and the diversity of fruit in the environment.

Diet
Field work resulted in 8500 trap nights and 50 faeces collected, being 42 samples of D. aurita (33 animals captured) and eight of M. paraguayanus (four animals captured).Both opossum species presented insects in their diets, but with a higher or lower consumption of fruits or vertebrate animals.Didelphis aurita was more omnivorous showing more similar frequencies of occurrence of the food categories among themselves.The consumption of vertebrates by D. aurita, such as birds and mammals, was also evidenced whereas for M. paraguayanus there was the main consumption of insects and some fruit species (Figure 1).The most consumed insect categories were Coleopterans which were present in 92.9 % of the samples of D. aurita and in 87.5 % of M. paraguayanus (Figure 1).A total of 29 753 seeds was found in faecal samples of D. aurita, including Piper sp., Cecropia pachystachya, and Jaracatia spinosa.For M. paraguayanus, the total was 1803 smaller seeds, including Piper sp.There were a high proportion of whole, not damaged seeds (> 90%) in comparison to a small proportion of damaged seeds, for both marsupial species.In a general view, marsupials showed heterogeneous diet compositions, differing significantly between both species (P = 0.03; F = 3.921; g.l.= 4 and 84; Pillai Trace = 0.238).Individuals of D. aurita captured in the canopy vs. those captured on the ground did not differ significantly in the diet composition (P = 0.68; F = 0.390; g.l.= 2 and 30; Pillai Trace = 0.025).Only young individuals were captured in the canopy (age classes 2 and 3).There were no significant differences in the diets between age classes (P = 0.290; F = 1.288; g.l.= 2 and 30; Pillai Trace = 0.079) or within sexes (P = 0.90; F = 0.106; g.l.= 2 and 30; Pillai Trace = 0.007) of D. aurita.

Frugivory and seasons
Avoiding pseudoreplication occasioned by recaptures in short time intervals a total of 33 faeces of D. aurita and eight faeces of M. paraguayanus were analyzed for seasonal variation in the diets.The highest number of seeds in faeces was registered in January for D. aurit and in March for M. paraguayanus (Table 1, Figures 2a, 2b).The mass of zoochoric fruits in the environment was higher in July, followed by March.The highest abundance of fruits was in March, with 12 plant species.The greatest peak of biomass in July is attributed to Citrus aurantium, an exotic species whose fruit is large, which was responsible for 51% of the total mass of fruits.When the C. aurantium was excluded, the peak of zoochoric fruits changed to March.The biomass of fruits was smaller in January.The total biomass of flowers in the environment was greater in November, being the flowers of the Sweet Palm Tree, Euterpe edulis, the main responsible for this peak (Table 2).The highest diversity indexes observed in faeces occurred during the wetter season (September to March), being 0.306 for seeds in faeces of D. aurita in March, and 0.43 in faeces of M. paraguayanus in January, whereas the highest index for the environment was 0.63 in November.The correla-tion analysis among diversity of fruits in the environment vs. diversity of fruits in faeces of D. aurita revealed a positive trend (r s = 0.90; Pp = 0.03) (Figure 3a), whereas for fruits in the environment vs. diversity of fruits in faeces of M. paraguayanus there was seen no relationship (r s = 0.40; P = 0.60) (Figure 3b).The highest number of fruiting Piper sp.observed was 73 in January, declining in the next months.The mean number of seeds of Piper sp. in faeces of D. aurita and M. paraguayanys was higher in January (1935) and March (203), respectively.The analysis of correlation between the number of fruiting plants of Piper sp. vs. the average number of respective seeds in faeces of D. aurita showed a positive trend (r s = 0.90; P = 0.04) (Figure 3c) whereas there was no relation seen for M. paraguayanus (r s = 0.60; P = 0.40) (Figure 3d).

Frugivory and reproduction
Only D. aurita was captured with pouch young during the study.The largest litter size was with 12 individuals in November 2004, averaging 4.5 cm in length.There was not a trend to a relationship between the mean pouch young length and richness of seeds in faeces of D. aurita (r s = 0.41; P = 0.49) (Figure 4a).However, there was a significant relationship between the number of independent young of D. aurita and the richness of seeds species consumed during the study (r s = 0.95; P = 0.01) (Figure 4b), with peaks of independent young mainly in November and in January.

Discussion
Seasonal forests can have reasonable influence on animal diets, making them seasonally dependent of feeding resources.However, in a general view, the diets of marsupials seen here are very similar to those reported in more dense Atlantic forest sites (Carvalho et al., 1999; Cáceres and Monteiro-     219) 250 ( 1025) The largest values are in bold.The hyphen indicates despicable mass (under 0.1 g).The total number of individuals is given in parentheses., 2001;Cáceres et al., 2002).Probably, seasonal differences among the physiognomies present in the Atlantic Forest sensu lato (e.g.dense, mixed, and semideciduous forest) are not very large (M.D. Alive, pers.comm), providing similar dietary resources in terms of great taxonomic groups (e.g.Coleopterans).In spite of the most didelphid marsupials being commonly considered as insectivorous-omnivorous (Robinson and Redford, 1986;Fonseca and Kierulff, 1989;Fonseca et al., 1996), differences in this omnivory can be seen when more detailed analyses are done.As an example, D. aurita consumed more vertebrate prey than M. paraguayanus, and this last species consumed more Lepidopterans, and such differences had already been observed elsewhere (Cáceres et al., 2002).These differences were already emphasized in other communities of marsupials in more humid coastal places (Santori et al., 1995;Carvalho et al., 1999;Cáceres et al., 2002;Santori et al., 2004) or even in similar forests (Casella and Cáceres, 2006).The insectivorous habit seems to dominate among the neotropical small mammals (Fonseca and Kierulff, 1989), mainly in cursorial and escansorial ones (Vieira and Astúa de Moraes, 2003;Casella and Cáceres, 2006).The consumption of other items, such as fruits, seems to take place randomly, except for arboreal and frugivorous-omnivorous species which consume fruits more regularly (Charles-Dominique et al., 1981;Julien-Laferrière, 1999;Vieira and Astúa de Moraes, 2003;Vieira, 2006).Micoureus paraguayanus is known as an arboreal species and, despite of having been captured only in the canopy here, it consumed fruits in lower frequency than D. aurita.However, body size differences between these species can be one of the reasons for such differences in diet, because the larger the body size, the more fruits will be found in the diet of such animal (Cáceres et al., 2002).It is important to note that D. aurita is 10 times larger in body size than M. paraguayanus, or more (Fonseca et al., 1996).Another factor can be the habitat which could drive diets for some specialities (Cáceres et al., 2002).Ecological segregation of marsupials is being suggested to be regulated by use of space rather than competition (Astúa de Moraes et al., 2003).The trend of M. paraguayanus for insectivory was verified in a dense Atlantic Forest site (Leite et al., 1996), but trends for frugivory was reported for the same species elsewhere (Cáceres et al., 2002).

Filho
The proportion of frugivory increases as the arboreal habit of species is dominant (Vieira and Astúa de Moraes, 2003).We would expect M. paraguayanus as more frugivore since the species is arboreal, but this was not seen here probably due to the low number of faeces collected here.On the other hand, scansorial species tend to exhibit a more balanced diet between arthropods and fruits (50: 50 %), such as the species of Didelphis (Atramentowicz, 1988;Cáceres and Monteiro-Filho, 2001).Hence, D. aurita was more omnivorous than M. paraguayanus in the study area, showing a tendency to consume all kinds of items.This should be provided by its scansorial mode of locomotion, larger body size and opportunism.This dietary pattern has been observed in Brazil (Stallings, 1989;Santori et al., 1995;Cáceres and Monteiro-Filho, 2001, Cáceres et al., 2009), French Guiana (Charles-Dominique et al., 1981;Atramentowicz, 1988) and Venezuela (Cordeiro and Nicolas, 1987).
In spite of the simpatry of the animals studied, they are not syntopic most of the time, one dwelling mostly in the canopy (M.paraguayanus: Leite et al., 1996;Graipel et al., 2003) and the other on the ground (D. aurita: Leite et al., 1996;Graipel et al., 2003).Such differences, on the view of D. aurita, would also happen due to the exploration of resources found in the litter (Cáceres and Monteiro-Filho, 2001).Even young D. aurita captured substantially in the canopy did not overlap in feeding habits with the arboreal M. paraguayanus.Young D. aurita should make rapid incursions in the canopy, exploring more resources on the ground (Miles et al., 1981;Vieira, 2006).However, fruits and nectar could be consumed by D. aurita when in the arboreal stratum (Vieira et al., 1991;Vieira and Izar, 1999).On the other hand, M. paraguayanus seems to use the ground level during certain times of fruit rarefaction in the canopy (Cáceres et al., 2002).
Feeding differences among individuals of different ages were not observed in this study and neither in other place, although some differences in consumption of snakes have been observed (Cáceres, 2002).Nonetheless, differences for the gray four-eyed opossum Philander frenatus have been noticed (Santori et al., 1997).The occurrence of strategies in which animals reproduce in periods when there is the highest amount of fruits is related to species that face seasonal fluctuations in the availability of resources.Normally, there is a synchronism between reproductive periods and fruit availability, which is markedly important in semideciduous forests.Marsupial species stop reproduction in times when there is a decrease in food availability, by exhibiting an anoestrous phase (Julien-Laferrière and Atramentowicz, 1990;Martins, 2004;Leiner and Wesley, 2007;Monteiro-Filho and Cáceres, 2006).The production on flowers and fleshy fruits were more intense in the rainy season, as is expected for semideciduous seasonal forest (Smallwood and Howe, 1982;Terborgh, 1986;Van Schaik et al., 1993).The abundance of fruits can influence the reproductive success of vertebrates (Howe, 1986;Powlesland et al., 1997;Henry, 1999).In this study, during the wetter months (e.g.November to January), there were large quantities of fruits in the diet of D. aurita, which is thought to be associated with the effort and energy demands in reproduction of the species during this period (Julien-Laferrière and Atramentowicz, 1990;Monteiro-Filho and Cáceres, 2006).This opossum species spends high energy demand with lactation rather than pregnancy, and obligatory lactating phase and release of young from the pouch occur in favourable times, which is fruit and flower richest.The intensity of these plant resources evidenced here are thought to be very important nutritionally (such as sugars) for these mammals during this time, as have been observed for the Piper in the study area.This strategy of reprodution in season with much fruit is related with species that face seasonal fluctuations in resource availability, there is synchrony between reproductive periods and fruit availability in the environment, even in less seasonal Forest (evergreen), with interruption of reproduction when there is a less food availability (Julien-Laferrière and Atramentowicz, 1990;Monteiro-Filho and Cáceres, 2006).Many species of marsupials didelphids behave this way, such as Marmosops paulensis (Leine, 2005), D. marsupialis, P. opossum (Julien-Laferrière and Atramentowicz, 1990) and Gracilinanus microtarsus (Martins, 2004).

Figure 1 .
Figure 1.Percentage of food items in faeces of the marsupials Didelphis aurita and Micoureus paraguayanus in a seasonl semideciduos forest of the Iguaçu National Park, southern Brazil.

Figure 2 .
Figure 2. Abundance of seed fruit species in faeces of the marsupials (circle): (a) Didelphis aurita and (b) Micoureus paraguayanus compared to the fruit abundance of the environment (triangule) in the Iguaçu National Park, southern Brazil.

Figure 3 .
Figure 3. (a) Correlation of Spearman among diversity for zoochoric fruits present in the environment and diversity of fruits in faeces of the marsupials (b) Didelphis aurita and (c) Micoureus paraguayanus.Correlation of Spearman among number of fruiting individuals of Piper sp.observed in the study area and mean number of seeds of Piper sp. in faeces of (d) Didelphis aurita and (e) Micoureus paraguayanus in the Iguaçu National Park, southern Brazil.

Figure 4 .
Figure 4. (a) Correlation of Spearman among richness of seeds species in faeces of Didelphis aurita and mean pouch young length (cm) of the opossum Didelphis aurita.(b) Correlation of Spearman among richness of seeds species in faeces of Didelphis aurita and number of independent young of Didelphis aurita sampled in the Iguaçu National Park, southern Brazil.

Table 1 .
Mean number of seeds present in faeces of didelphid marsupials in the Iguaçu National Park, southern Brazil.Higher numbers of Piper sp.seeds were responsible for peaks observed for both species.
Note: Standard deviation is given and the coefficient of variation is in parentheses.

Table 2 :
Biomass of fleshy fruits and flowers (in grams) between November 2004 and July 2005 in the Iguaçu National Park, southern Brazil.