OUP user menu

Semelparity and Factors Affecting the Reproductive Activity of the Brazilian Slender Opossum (Marmosops paulensis) in Southeastern Brazil

Natalia O. Leiner, Eleonore Z. F. Setz, Wesley R. Silva
DOI: http://dx.doi.org/10.1644/07-MAMM-A-083.1 153-158 First published online: 19 February 2008


Data on the reproductive patterns of the Brazilian slender opossum (Marmosops paulensis) were collected in an area of Montane Atlantic forest, southeastern Brazil, from August 2002 to July 2004. Reproduction occurred from September to March in both years, a period of high food supply, probably as a way to maximize survival of juveniles. There was nearly zero postmating survival, thus, no individual took part in more than 1 breeding event. This pattern characterizes a semelparous life history, which has been described in other small didelphids and dasyurids. Females were reproductively active during months with longer day lengths and abundant fruit supply. Breeding seems to be initiated by a 12L: 12D photoperiod and a rapid rate of change in day length, as demonstrated in semelparous dasyurids. Hence, the effect of photoperiodic cues on the onset of reproduction also may stand for other semelparous didelphids. We suggest that fruit availability controlled the length of breeding activity in M. paulensis, and it could play a role in the occurrence of semelparity in this species. However, semelparity may occur only due to phylogenetic constraints, whereas food supply works as a selective force maintaining this trait.

Key words
  • Brazil
  • day length
  • food availability
  • Marmosops paulensis
  • opossum
  • reproductive activity
  • semelparity

For most mammals, reproduction is usually associated with favorable periods of the year, in order to maximize reproductive success (Flowerdew 1987). Many mammals thus reproduce opportunistically, such as cricetid rodents in the Brazilian Caatinga that reproduce whenever rainfall reaches a threshold (Cerqueira and Lara 1991). Other mammals show seasonally marked reproductive periods, including marsupials, which generally start to reproduce when environmental conditions are still unfavorable (Bronson 1989). In this case, both endogenous and exogenous factors are responsible for the onset and duration of reproduction (Bronson 1989; McAllan 2003). The presence of males can act as a trigger leading females to estrus due to pheromonal cues (Fadem 1985), ensuring synchronization of reproduction (Perret and Ben M'Barek 1991). Among exogenous factors, reliable cues such as photoperiod seem to play a role in the regulation of reproduction (McAllan 2003; McAllan et al. 2006), whereas the availability of resources can determine the amount of energy invested in reproduction, hence, the duration of the reproductive period (e.g., Julien-Laferrière and Atramentowicz 1990).

There are several studies describing the reproductive patterns of Australian marsupials, including carnivorous dasyurids, which are considered distantly related to neotropical didelphid species (Cockburn 1997). Among dasyurids, mating occurs during a highly synchronized and short breeding season (Bradley 1997; Oakwood et al. 2001). In the best-studied genera, timing of reproduction seems to be determined by photoperiodic cues, especially changing photoperiod, and by other factors of secondary importance such as pheromones and rainfall (McAllan 2003). Moreover, males die after mating due to stress-related pathologies (Bradley et al. 1980; Oakwood et al. 2001), whereas females survive to breed a 2nd time, with each generation separated in age by 1 year. Although females can breed a 2nd time, they are usually subject to “exaggerated senescence” (Cockburn 1997). In Didelphis virginiana, exaggerated senescence occurs when reproductive performance of females is higher during their 1st breeding season, whereas in Antechinus stuartii it occurs when females reduce their reproductive investment and are more susceptible to mortality during the 2nd breeding season due to severe infestation of parasites (Cockburn 1997). The occurrence of male die-off and female decline in fecundity after 1st reproduction has been described as a semelparous reproductive strategy by Braithwaite and Lee (1979).

Reproductive patterns of the neotropical marsupials, including the didelphids, are poorly understood. Studies suggest that rainfall should play a role in the onset of didelphid reproduction (Fleming 1973; Julien-Laferrière and Atramento-wicz 1990), leading to a pattern in which lactation or weaning periods or both are usually correlated with periods of high fruit and insect abundance. Bergallo and Cerqueira (1994), on the other hand, demonstrated that photoperiod and not rainfall controls the reproductive activity of Monodelphis domestica, as occurs in Australian marsupials. Moreover, similar to some dasyurid species, some small opossums exhibit partial semel-parity, such as Marmosops incanus (Lorini et al. 1994), Gracilinanus microtarsus (Martins et al. 2006), and Monodelphis dimidiata (Pine et al. 1985). Notwithstanding, females that breed a 2nd time also show a decline in fecundity, resembling dasyurids. In this way, the presence of a semelparous reproductive strategy also seems to be common in didelphids, especially among smaller species.

The Brazilian slender opossum, Marmosops paulensis (Didelphimorphia, Didelphidae), is a small (20- to 70-g), nocturnal marsupial occurring in the southeastern part of the Brazilian Atlantic forest, restricted to montane forests above 800 m (Mustrangi and Patton 1997). Fruits, especially of the genus Piper, and insects are the major food items of this species (Leiner and Silva 2007). There are no data on the reproductive patterns of M. paulensis, but Lorini et al. (1994) presented data on the reproduction of M. incanus, a larger but closely related species that has previously been mistaken for M. paulensis. Lorini et al. (1994) suggested that the reproductive activity of M. incanus occurs between October and December, and that each individual participates in only 1 reproductive event, indicating a semelparous life history. Moreover, because data used in Lorini et al. (1994) came from museum specimens collected over 5 Brazilian states (Paraná, São Paulo, Rio de Janeiro, Espírito Santo, and Bahia), semelparity seems to be a rather widespread phenomenon in M. incanus. The objectives of our study were to describe the reproductive patterns of M. paulensis, and to analyze the factors influencing its reproductive activity.

Materials and Methods

Study site and data collection.—Our study was conducted from August 2002 to July 2004 in Parque Estadual Intervales, municipality of Ribeirão Grande, state of São Paulo, southeastern Brazil (24°16′S, 48°25′W). The study site is one of the largest conservation units in São Paulo state, covering an area of 49,000 ha of protected Atlantic forest. The climate is cold and wet and there is a rainy season from October to February and a less-rainy and colder season, hereafter called the dry season, from March to September. During the dry season average monthly temperature was about 14.3°C (SD = 1.97°C) and average monthly rainfall was about 44.4 mm (SD = 36.52 mm), whereas in the rainy season average monthly temperature was about 18.7°C (SD = 2.41°C) and average monthly rainfall was 196.1 mm (SD = 93.64 mm).

Within this conservation unit, sampling was carried out in a montane Atlantic forest area known as “Sede,” which is located at approximately 850 m above sea level. This area is covered with old secondary-growth vegetation, with the understory dominated by species of Piperaceae (Piper), Rubiaceae (Psychotria), and Melastomataceae (Leandro). Trees of Lauraceae (e.g., Ocotea) and Meliaceae (e.g., Cedrela fissilis) were common in the subcanopy.

A trapping grid composed of 5 parallel line-transects spaced 50 m apart was established to capture the animals. Each transect had 8 trapping stations 20 m apart, and in each trapping station we placed 2 Sherman live traps (model XLF15, 10.5 × 12 × 37.6 cm; H. B. Sherman Traps, Inc., Tallahassee, Florida). One trap was placed on the ground, whereas the other was placed at approximately 1.5 m height, on tree branches. The bait used was a mixture of banana, peanut butter, oatmeal, and bacon. Trapping was conducted on 5 consecutive days each month. Each day, we checked the traps early in the morning, and renewed baits if necessary.

Sex, reproductive condition, and age class were recorded for each captured individual of M. paulensis. Females were considered reproductive if they had swollen nipples, young in the pouch, or if they were pregnant, which could be observed in a few cases by palpation. Nipples were checked for the presence of milk, because we assumed that swollen nipples without milk indicated females that had already weaned their young and that were not receptive to further mates. Reproductive condition of males was not assessed; once the male's testes become scrotal when they reach sexual maturity, they stay in this position permanently, precluding an accurate evaluation of their reproductive activity (Quental et al. 2001). Age class was estimated based on patterns of tooth eruption, following the criteria defined by Tribe (1990). Before releasing the individuals, they were marked with numbered ear tags. Trapping and handling conformed to guidelines approved by the American Society of Mammalogists (Gannon et al. 2007).

Data on precipitation were gathered at the park's meteorological station, approximately 1 km from the study area. Day length was obtained from the Brazilian National Observatory (Rio de Janeiro, RJ) and the rate of change of photoperiod per day was calculated by subtracting each successive day length from the preceding to give the difference in photoperiod from one day to the next. From May 2003 to July 2004 we measured monthly availability of fruits and arthropods. To quantify fruit abundance, we counted the number of fruits from Piperaceae, Solanaceae, and Melastomataceae trees found inside 20 strip-transects of 30 × 1.5 m, and the total number of fruits was used as a fruit availability index. These plant families were chosen because their fruits compose a substantial part of the diet of M. paulensis (Leiner and Silva 2007). To measure arthropod abundance, we placed a small pitfall trap consisting of a plastic container (200 cm3) filled with 70% ethanol at each trapping station. After 3 days, all arthropods were removed from the pitfall traps, identified to the level of order, and counted. The sum of all orders, excluding those that were not consumed by M. paulensis based on Leiner and Silva (2007), was used as an arthropod availability index.

Data analysis.—To evaluate the influence of abiotic factors on the reproductive activity of M. paulensis, we performed 2 multiple logistic regressions. The dependent variable used in both regressions was the reproductive activity of female M. paulensis, coded as reproductive (1) or not reproductive (0). In the 1st multiple regression, we used monthly average rainfall and monthly average day length (in hours) as independent factors, whereas in the 2nd multiple regression we used the abundance of fruits and arthropods as independent factors. All analyses were done using the software STATISTICA 6.0 (Statsoft, Tulsa, Oklahoma).


We captured 51 individuals of M. paulensis, including 28 females and 23 males, during more than 10,000 trap-nights. In both years, females presenting swollen nipples were 1st captured in mid-October, and we were able to detect pregnancy in a few females in mid-September (Table 1). Reproductive activity of females lasted until March, and during this period all trapped females presented swollen nipples, indicating a high degree of synchrony among females in the population (Table 1). In March, females were no longer producing milk, indicating that by this time they had already weaned their young. In fact, the 1st juveniles were trapped in March, and in this month we found (by using a spool and line device) a female's nest that contained 4 young. This behavior of leaving the young in a nest seems to be common in M. paulensis, because we never captured a single female with young attached to the nipples.

View this table:
Table 1

Monthly percentages of nonreproductive, pregnant, and lactating female Marmosops paulensis at the Parque Estadual Intervales, Brazil, from August 2002 to July 2004.

MonthNonreproductive femalesPregnant femalesLactating females

All males disappeared from the population after January in the 1st year, whereas in the 2nd year the same phenomenon occurred after December. Thus, no adult males were trapped from January to July despite an effort of 2,400 trap-nights during this period. The only males present during these months were juveniles, followed by subadults. After April, all captured females belonged to juvenile and subadult age classes. Appearance of new adults of both sexes did not occur until August, near the beginning of the reproductive season. In this way, each generation was discrete and separated in age by approximately 1 year (Fig. 1).

Fig. 1

Monthly percentage of individuals belonging to the juvenile (hatched bars), subadult (gray bars), and adult (black bars) age classes of Marmosops paulensis at the Parque Estadual Intervales, Brazil, from August 2002 to July 2004. Months within rectangles represent the period when only adult females were found in the population.

Average monthly day length and average monthly rainfall were related to the reproductive activity of female M. paulensis (G = 31.626, n = 24, d.f = 2; P < 0.0001). However, removal of rainfall from the model did not change the result (G = 1.229, P = 0.268), indicating that this factor did not influence the reproductive activity of M. paulensis. Breeding started when day length was about 12L:12D and the rate of change of photoperiod was most rapid, whereas reproduction ceased when day length was longest but the rate of change of photoperiod was slowest (Fig. 2), Abundances of fruits and arthropods also were related to reproductive activity of M. paulensis (G = 6.905, n = 15, d.f. = 2; P = 0.03), although removal of arthropod abundance from the model did not influence the model fit (G = 0.775, P = 0.37). Thus, photoperiodic cues (G = 30.39, n = 24; d.f. = l,P< 0.0001; Fig. 3) and fruit supply (G = 6.13, d.f. = 1, n = 15, P = 0.01; Fig. 4) appeared to be most strongly related to the reproductive activity of female M. paulensis.

Fig. 2

Reproductive activity of Marmosops paulensis plotted against the yearly cycle of photoperiodic change. Solid line = absolute day length in hours; dashed line = rate of change of photoperiod per day in seconds.

Fig. 3

Relationship between monthly average day length and reproductive activity of female Marmosops paulensis.

Fig. 4

Relationship between monthly fruit availability and reproductive activity of female Marmosops paulensis.


The reproductive activity of M. paulensis was markedly seasonal, occurring from September to March in both years. Although the beginning of the reproductive period of M. paulensis occurred in the end of the dry season, most lactating females were present during periods of high food abundance, which corresponded to the rainy season at our study site. By adjusting lactation to periods of food abundance, female M. paulensis can maintain their investment in reproduction (e.g., milk production), and increase the probability of survival of their young. Well-fed females are usually larger and are able to produce and wean larger litters and heavier heonates, which translates into higher reproductive success than females with a poorer body condition (Cothran et al. 1985). Examination of data on other didelphids also found a correlation between lactation and high food supply (Fleming 1973; Julien-Laferrière and Atramentowicz 1990; Lorini et al. 1994).

After the reproductive period, adult males disappeared from the population, whereas adult females disappeared after weaning their young (March and April). This leads to an annual replacement of generations, which could be due to dispersal by adults to another area; however, the lack of immigration of foreign adults argues against this hypothesis. Moreover, the occurrence of fur loss in the rump and parasite infestation in a few males after the mating season indicates a low probability of survival in these individuals. This decline in body condition also can enhance the susceptibility of these individuals to predation. Hence, the disappearance of adults from the population seems to be due to high adult mortality after the breeding season, such that each individual can take part in only 1 reproductive event, thus resulting in semelparity.

Semelparity was 1st defined by Cole (1954:105) as “reproducing only once in a lifetime,” whereas Braithwaite and Lee (1979) have used this word to describe the situation where males participate in 1 breeding season and females are iteroparous. Although semelparity is unusual among vertebrates, it evolved at least 5 different times among members of the families Dasyuridae and Didelphidae (Cockburn 1997). In addition to the postreproductive mortality of males in these groups, females show a strong decline in fecundity after 1st reproduction. Cockburn (1997) argued that there might be a phylogenetic predisposition in the species belonging to both clades (Dasyuridae and Didelphidae) toward postreproductive senescence, leading to die-off of males in one extreme. Martins et al. (2006), in a study on a population of G. microtarsus, found that although mortality was high after the reproductive period, some males and females were able to survive until the next breeding season, resulting in partial semelparity. In our study, adults of both sexes did not survive to reproduce a 2nd time in both years, resulting in complete semelparity.

Usually, semelparity evolved in species coping with major fluctuations in survival of juveniles or adults (Promislow and Harvey 1990). Assuming that survival of juveniles is consistently higher during 1 season of the year, selection should result in an intense, short, and synchronized reproductive period such that young are born during this season (Braithwaite and Lee 1979). As expected, in the semelparous didelphids known to date, females present a high reproductive investment, usually expressed as large litter sizes (mean of 10 young per litter). In contrast, didelphid species with longer life-spans and production of >1 litter per year (iteroparity), such as Philander opossum and Caluromys philander, generally have a mean of 4 young per litter that varies according to resource level (Julien-Laferrière and Atramentowicz 1990).

The reproductive activity of female M. paulensis occurred during a period of longer day lengths and abundant fruit supply, indicating an influence of these 2 factors on the breeding season. Although rainfall also is considered a useful predictor (e.g., Fleming 1973; Julien-Laferrière and Atramentowicz 1990), its relationship with reproductive activity might be due to a high correlation between rainfall and day length (Bergallo and Cerqueira 1994).

Considering photoperiodic cues, reproduction started at the spring equinox, when the photoperiod was about 12L:12D, and ceased when photoperiod was longest. In the genus Antechinus (Dasyuridae), which also is semelparous, females usually enter estrus at this same day length, although the rate of change in photoperiod seems to be a more important reproductive cue than the absolute day length (McAllan et al. 2006). Experimental evidence demonstrated that female Antechinus flavipes under artificial, unchanging 12L:12D do not enter estrus, whereas increasing daily day length in minutes triggered the breeding activity of these females (McAllan and Geiser 2006). Female M. paulensis started to breed when photoperiod was changing at its most rapid rate, similarly to members of the genus Antechinus (McAllan et al. 2006). In contrast, when the rate of change of photoperiod was at its slowest, young were weaned and reproductive activity ceased. To test the relative roles of absolute day length and rate of change of photoperiod as cues to the onset and ending of reproduction, we would need experimental evidence or evidence gathered in a broader timescale.

Resource availability also may function as a seasonal predictor of reproduction in didelphids. In our study, reproductive activity of female M. paulensis was highly correlated with food supply. Rademaker and Cerqueira (2006) observed a decrease in the duration of the reproductive period in the genus Didelphis at higher latitudes, which typically provide lower food supplies. This decrease could lead to semelparity in extreme cases. Earlier studies found that semelparity may be facultative, and may be linked to latitude and fluctuations in resource availability. Lorini et al. (1994) found that a population of M. incanus inhabiting an area of low latitude (14°50′S) exhibited an iteroparous reproductive strategy, whereas this species was semelparous at higher latitudes. Moreover, Mills and Bencini (2000) and Wolfe et al. (2004) found that different populations of Parantechinus apicalis exhibited semelparity when coping with variations in food supply, but were able to invest in a 2nd reproductive event when presented with a more constant food supply. Thus, it is possible that resource availability may affect the evolution and maintenance of life-history strategies such as semelparity and iteroparity.

Julien-Laferrière and Atramentowicz (1990) demonstrated discontinuation of reproductive activity and a decrease in litter size in Didelphis marsupialis, C. philander, and P. opossum when coping with scarcity of fruit. Because mouse opossums are more frugivorous than big-bodied opossums (Astúa de Moraes et al. 2003), it is reasonable to expect a tighter relationship between reproduction and fruit production in the former opossums, such as M. paulensis, which is highly frugivorous at the study site (Leiner and Silva 2007). Hence, fruit supply seems to control the duration of the reproductive period in M. paulensis, and may play a role in the occurrence of complete semelparity in this species. Comparisons of reproductive patterns among populations living in contrasting environments regarding resource availability, and experiments using food additions, should help to elucidate the role of sea-sonality in food resources in the occurrence of semelparity in M. paulensis and other didelphids. Future studies should take the phylogeny of the marsupials into account as well, because the evolution of semelparity may be due to phylogenetic constraints, and seasonality of food resources may act only as a selective force maintaining this trait.


We thank M. A. R. Mello, G. Machado, M. V. Vieira, and D. Broussard for improvements on earlier versions of the manuscript, and P. E. C. Peixoto for help with the logistic regressions. Idea Wild donated field equipment, and Parque Estadual Intervales and Departamento de Zoologia (Instituto de Biologia, Universidade Estadual de Campinas) provided logistical support. Fundação Mary Brown, Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, and Fundação de Amparo à Pesquisa do Estado de São Paulo (research grants 98/05090-6 and 03/03595-3) provided financial support.


  • Associate Editor was Craig L. Frank.

Literature Cited

View Abstract