BIOLOGY AND NATURAL HISTORY OF BRAZILIAN ATLANTIC FOREST SMALL MAMMALS Gustavo A. B. da Fonseca and Maria Cecilia M. Kierule ABSTRACT During a 17-month study, 20 small mammal species were recorded in the forests of the western slopes of the Brazilian Atlantic coastal region in the state of Minas Gerais. A trapping effort of 57,120 trap nights resulted in 1366 captures of 692 individuals belonging to 8 marsupial and 12 rodent species. Both terrestrial and arboreal traps were used. The common opossum Didelphis marsupialis was the most frequently trapped small mammal, followed by Mannosa incana. The spiny rat P~oechimys setosus was the third most common small mammal and the most abundant rodent species. One marsupial and five rodent species were considered rare in this study, being represented in the sample by two or fewer individuals. Population turnover at the traplines was high for all species, with most individuals only appearing once in the total study sample. Average body weight for the largest species, Didephis marsupialis, was close to 1000 g, while for the smallest species, 0,yzomys nigripes, it was approximately 20 g. Sexual dimorphism was common among marsupial species, but rare in rodents. Male marsupials, on average, tended to be larger than females. Male 0,yzomys trinimtis, on the other hand, were slightly smaller than females. Most marsupials are seasonal breeders, while rodents tend to reproduce throughout the year. The western slopes of the Atlantic forest are characterized by a pronounced dry season, and reproduction is concentrated in late dry and the early and midwet season. Seasonality of maI'supial reproduction was recognized not only by the presence of breeding males and females, but also by juvenile recruitment into the population in subsequent months. The vast majority of small mammal species have demonstrated a high degree of overlap in substrate use. A large number of small mammals have some level of scansorial ability, even the ones which are more frequently caught at the forest floor. It is also suspected that most species have mixed diets. Therefore, potential competition for resources is a definite possibility. * The senior author, Dr. Fonseca, is in the Departamento de Zoologia, 1nstituto de Ciencias Biologicas, Universidade Federal de Minas Gerais, Belo Horizonte, M.G. 30000, Brazil. da Fonseca, GA.B., and M.C.M. Kierulff. 1989. Biology and Natural Histoiy of Brazilian Atlantic Forest Small Mammals. Bull. Florida State Mus., Biol. Sci. 34(3):99-152. 100 BULLETIN FLORIDA STATE MUSEUM 34(3) RESUMO Durante um perfodo de 17 meses consecutivos de levantamento, vinte esp6cies de pequenos mamfferos foram registradas nas florestas da regiao 6cidental da Mata Atiantica Brasileira, no estado de Minas Gerais. Um esforgo de captura de 57.120 armadilhas-noite resultou em um total de 1.366 capturas de 692 individuos de oito esp6cies de marsupiais e doze,de roedores. Fez-se uso tanto de armadilhas ferrestres'quanto de arb6reas. 0 gambd comum Didelphis marsupialis toi o pequeno mamffero mais frequentemente capturado durante o estudo, seguido por Mannosa incana. 0 rato-de-espinho Proechimys setosus foi a esp6cie mais comum de roedor capturada neste estudo. Uma esp6cie de marsupial e cinco de roedores foram consideradas raras, sendo representadas na amostra por, n6 mdximo, dois indivfduos. A rotatividade de indivfduos nas linhas de captura foi bastante alta para quase todas as esp6cies, sendo que a maioria dos,individuos foram registrados uma 56 vez no total da amostra deste estudo. A m6dia de peso corporal para a maior das esp6cies, Dide*his marsupialis, foi de aproximadamente 1.000 gramas, sendo que para a menor esp6cie, 0*omys nignpes, a media de peso esteve pr6xima de 20 gramas. A maioria das esp6cies de marsupiais mostrou dimorfismosexual, uma caracteristica rara entre roedores. Os machos das espdcies de marsupiais tendem a ser, em m6dia, mais pesados e maiores do que as ftmeas. Por outro lado, os machos de 00'zomystnnitatis sao um pouco menores do que as fameas. As atividades reprodutivas da maioria dos marsupiais sao sazonais, enquanto que os roedores se reproduzem ao longo de todo o ano. As encostas ocidentais da Mata Atlantica sao caracterizadas por uma estagao seca pronunciada, e a reprodutao de pequenos mamfferos esta concentrada no final da estaqao seca e no infcio e meio da estagao chuvosa. A ocorrancia de reproduqao sazonal em marsupiais foi determinada pela observagao de machos e femeas em condigao reprodutiva e pelo recrutamento de jovens na populagao em meses subsequentes. A grande maioria das esp6cies de pequenos mamfferos mostrou superposigao no uso de substratos para locomoqao, utilizando-se tanto de substratos arb6reos quanto do solo das matas. Um grande numero de esp6cies demonstra habilidade escansorial, mesmo aquelas que locomovem-se mais frequentemente no so16. Suspeita-se tamb6m que a maioria dos pequenos mamfferos possuem dietas mixtas, ocorrendo superposigao de vdrios itens alimenticios. Existe a possibilidade, portanto, de ocorrencia de competivao por recursos entre estas esp6cies de pequenos mamiferos. TABLE OF CONTENTS Introduction.............................................................................................................................................. 101 Acknowledgement 102 Materials and Methods........................................................................................................................... 103 Study Siteq 103 Climpt, 106 Trapping 107 Results and Species Accounts .........................................'..................................................................... 108 Discu.inn 125 Seasonality and Resource T » 125 Life History PattprnQ 128 Population Turnnver 129 Appendiv I 134 FONSECA & KIERULFF: BRAZILIAN ATLANTIC FORESr SMALL MAMMALS 101 INTRODUCTION The neotropical region encompasses some of the most threatened ecosystems in the world, and yet the mammalian fauna of this area still remains very poorly known (Mares and Genoways 1982). In South America, the Brazilian Atlantic forest has been the vegetation formation subjected to the highest rate of destruction, with less than 5% of the region retaining some form of forest cover (Fonseca 1985a), and probably less than 1% in undisturbed state (Mittermeier et al. 1982). Nonetheless, the few natural history studies conducted on its fauna have indicated that species diversity is quite high, with many faunal elements being unique to the region (Mello- Leitao 1946; Muller 1973; Fonseca, unpublished data). The pioneer works of Moojen (1952), Vieira (1955), and Cabrera (1957; 1961) provide a general data base indicating that there are at least 129 species of non-volant mammals in the Atlantic forest region, about 40% of which are endemic. Were the taxonomy of most of the groups better understood, other species would probably be described. There are at least 23 marsupial and 57 rodent species described for the Atlantic forest, of which 39% and 53 %, respectively, are endemic to the region. Because of past and present habitat destruction, the fauna is increasingly isolated into small patches, and many members of this unique ecosystem are now endangered. This situation has been documented for the most conspicuous elements of the mammalian fauna, the primates (Coimbra-Filho and Mittermeier 1977; Mittermeier et al. 1982; Fonseca 1985b). Nonetheless, the rich and highly endemic small mammal fauna of the Brazilian Atlantic forest region has been the subject of very few long-term studies, especially when compared to the Amazon region (e.g. Pine 1973; Lovejoy et al. 1984; 1986; Terborgh et al. 1984; Malcolm 1987) or the Cerrado (Alho 1981; Alho et al. 1986; Fonseca and Redford 1984; Lacher et al., in press; Nitikman and Mares 1987). The only detailed study of Atlantic forest small mammals is now over 40-years-old (Davis 1946). Historically, the Brazilian Atlantic forest extended from the coast to the eastern and portions of the western slopes of the coastal mountains (Hueck 1972; Alonso 1977). This vegetation formation was originally distributed over an area of approximately 700,000 square km. Where rainfall permits the presence of tall, evergreens, this type of forest extends into the western slopes of the coastal formation. Due to a rain shadow, the vegetation of the western slopes, where this study was conducted, possesses a number of deciduous tree species, which lose their leaves during the approximately 6 months of the dry season. During the wet season, however, the forests of both the eastern and western regions are physiognomically undistinguishable. In addition, the 102 BULLEnN FLORIDA SrATE MUSEUM.34(3) faunal elements are mostly the same, generally belonging to the same biogeographical region (Muller 1973). The objective of this study was to investigate several aspects of the biology and natural history of eastern Brazilian non-volant small mammals. During a 17-months period, data on the general aspects of population dynamics, breeding, substrate use and movement patterns of small mammals (marsupials and rodents) were collected in six forest plots, at three main sites in the state of Minas Gerais, Brazil. Morphometric data were also collected. The present study was part of a larger research project on small mammal community structure, in which it has been shown that primary forests are comparably poorer in species richness and diversity when compared to secondary vegetation in mid-stages of succession. In addition, it was also suggested that predation may exert a high degree of influence on the patterns of community composition and structure in the forest fragments of the Atlantic coastal region of Brazil. Patterns of species composition, relative abundances, habitat preferences, predation pressure and other community structure param are presented in Fonseca (1988). ACKNOWLEDGEMENTS First we would like to thank John G. Robinson for his advice during planning, field work, analysis and writing of this study. John Eisenberg, Larry Harris, Kent Redford, Nigel Smith, Melvin Sunquist, and Charles Woods also provided valuable support. John Eisenberg and Oliver Pearson improved the manuscript. This research also benefited from discussions with Drs. Thomas Lacher and Michael Mares. Russell A. Mittermeier supported our research by advocating our cause with the World Wildlife Fund-US, which financed the most substantial part of the costs. Additional financial support was provided by the Program for Studies in Tropical Conservation, of the University 6f Florida, and by the Research Council of the Federal University of Minas Gerais, Brazil. The National Research Council of Brazil (CNPq) awarded one of us (G. A. B. da Fonseca) a doctoral fellowship. Celio Valle encouraged the field work with his enthusiasm, vision, and eternal optimism. Several people helped during field work. I would like to thank Ludmilla Aguiar, Ilmar Bastos, Sonia Rigueira, Carlos Alberto Pinto, Ederson Machado (sometimes), Silverio Machado, Ney Carnevalli, Luiz Fernando Mello, Gisela Herrmann, Jairo Vieira, Eduardo Veado, Eduardo Sabato, Luiz Paulo Pinto, and Maria Cristina Alves. Our small mammal data from the Rio Doce Park was partially collected by Jody Stallings and his crew (see Stallings 1989) Mr. Feliciano Abdalla and Dr. Antonio Cupertino kindly allowed us to work on their farms and provided housing. "Santinho" let us stay in his house at the expense of family problems, and we are thankful to him. The State Forest Institute of Minas Gerais (IEF) provided us with accommodations and gasoline at the Rio Doce State Park. We are very grateful to the staff of the Park, especially Ademir Camara Lopes, Jose Lourenco Ladeira and Hermogenes Ferreira Neto. M. Carleton, K Creighton, L. Emmons, G. Musser, P. Myers, and J. Patton generously identified the small mammal voucher specimens. FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREM SMALL MAMMALS 103 MATERIALS AND METHODS Study Sites Small mammal communities of six forest plots, two at each of three sites in the state of Minas Gerais, were the subject of this study (Fig. 1). At each site, a primary and a secondary forest plot was selected. The objective was to cover the widest possible spectrum of habitats. A set of three parallel transect trapping lines was established in each. The primary forest plots showed vertical stratification, with an average canopy height of 19 m. Tall buttressed emergents were a common occurrence (Fig. 2a). The herbaceous stratum was somewhat sparse, while the midstory tas generally well developed. The secondary forests, on the other hand, were mostly in their mid-stages of succession (Fig. 2b). Average canopy height was approximately 12 m; herbaceous cover was frequently extensive, with masses of tangled vines being a common occurrence. Epiphytes and emergents were conspicuously absent from the secondary forests. The first site was Fazenda Esmeralda, located in the county of Rio Casca (Fig. 1). The farm is under extensive agricultural use, with very little remaining under forest cover. Located 4,0 00 450 5 BRAZIL 00 G ES 8 A HIA Sp RJ O15 MINAS GERAIS -20 ESPIRITO SANTOSAO 20- PAULO RIO DE 0 150 300 KM JANEIRO 0 -25 250 40045 1 51 1 1 Figure 1. Locations of study sites in the state of Minas Gerais' Brazilian Atlantic forest (1 = Fazenda Esmeralda; 2 = Rio Doce State Park; 3 = Fazenda Montes Claros) 104 BULLEI'IN FLORIDA STATE MUSEUM 34(3) along the plains of the Rio Doce River, the farm was covered almost completely by pristine forest as recently as 1964. Wood extracting rights were then sold to the largest steel industry of the state of Minas Gerais and by 1970 most of the farm was deforested. Forest patches located on the top of two hills were selected for the study, one with a 60 ha second growth, and the other with 80 ha of primary forest, known locally as "Lagoa Fria." The second site was the Rio Doce State Park (also treated in detail by Stallings 1988), which has portions of its area within the county of Marlieria (Fig. 1). The park, with its 35,973 ha, constitutes the largest continuous area under tropical forest in the state of Minas Gerais. It was created in 1944 and has been under the jurisdiction of the Government of the state of Minas Gerais since that time. Since the creation of the State of Minas Gerais Forest Institute (IEF) in the late 19605, the park has been under its administration. It has recently become one of the best maintained and protected areas under the Brazilian State Parks system. Because of several extensive fires in the 1960s, a considerable area of the park is second growth. One of these, "Mata do Hotel," was selected as a study area. A pristine primary forest, known locally as "Campolina," constituted the second patch selected for study within the Rio Doce Park. Fazenda Montes Claros was selected as the third site for this study. It is a coffee and cattle farm located within the counties of Ipanema and Caratinga (Fig. 1). The total area of the farm is about 1200 ha, 860 ha of which remain under forest cover. A research station under the administration of the Brazilian Foundation for Conservation of Nature (FBCN) and the Federal University of Minas Gerais (UFMG) was established on the farm in 1983. A second growth forest patch at Fazenda Montes Claros, known as "Jao," was selected, and another under primary forest, "Matao," was also used. Figure 2. (a) Secondary forest of Fazenda Esmeralda, one of the study sites. (b - on facing page) Buttressed tree characteristic of the primary forests that were surveyed in this study (photo by Kent H. Redford); Jody Stallings is in the foreground. FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 105 106 BULLEI'IN FLORIDA STATE MUSEUM 34(3) Climate Climatological data were collected at all three sites, but because of the extreme similarity in temperature and rainfall regimes among study sites, only information collected at Fazenda Montes Claros are presented here. Hueck (1972) states that rainfall for the western slopes of the Atlantic forest region is always below 1600 mm annually. In some areas, it can achieve a little over 1000 mm. A climatogram of Walter (1971), constructed with data collected at this study, can be found in Figure 3. During the study period, the region experienced an unusually dry period; total rainfall for the first 12 months of trapping was 850 mm, while for the last 12 months rainfall totaled 931 mm. Rainfall is highly seasonal, being concentrated between the months of September and February (Fig. 3) Average monthly precipitation for the rainy season was 128 mm, while dry season rainfall averaged only 30 mm.monthly. Mean minimum annual temperature during the study period was about 18¤ C, close to the average for the region (Hueck 1972). Average daily differences between minimum and maximum temperatures are quite constant throughout the year. Rainfall maxima coincide with the warmest months of the year, while winters are usually very dry. For the purpose of this analysis, the dry season is considered to occur between the months of March and August, and the wet season between September and February. MONTHLY RAINFALL (mm) MEAN TEMPERATURE ( C)300 150 135 250 120 90 150 DROUGHT 75 t 60 100 --1~1 -- - - 45 050 / 15 0 Jun Jul Aug Sep Oct NovDec Jan FebMar Apr May Jun Jul Aug'SepOct MONTHS Figure 3. Climate diagram of Walter (1971), constructed using climatological data collected at Fazenda Montes Claros between June 1985 and October 1986. The diagram indicates periods of drought and water surplus. The di~¥ season can be defined between the months of March and August. Temperatures are given in C, and rainfall in mm. FONSECA & KIERULFF: BRAZILIAN ATLANI'IC FOREST SMALL MAMMALS 107 Trapping Three transect lines 300 m Iong were established in each plot at each of the three sites. These transects were as parallel to each other as possible and separated by 100 m. Each transect line possessed 16 trapping stations 20 m apart. Traps were placed in suitable locations within a 3.5 meter radius measured around center of station. A squirrel-size Tomahawk live trap (one- door folding trap, size 203, Tomahawk Live Trap Co., Tomahawk, WI) was placed on the ground at each station. At every other station, a second Tomahawk trap of the same size was wired either to a branch or vine at heights from 1 m to 4 m high. In addition to these traps, every other station possessed a mouse-sized collapsible Sherman trap (H. B. Sherman Traps, Inc., Tallahassee, FL), with altemation of. ground and tree traps. Moreover, the two outermost transect lines had, at every other station, a large 80 x 30 x 30 cm wire home-crafted live trap. Therefore, each outer line possessed 16 ground Tomahawk traps, 8 tree-bound Tomahawk traps, 4 ground and 4 tree-bound Sherman traps, and 8 ground large wire traps. The total for each outer line was 40 traps. The mid-line did not have large traps, but a total of 24 Tomahawk and 8 Sherman traps. In summary, each forest plot had 48 trap stations disposed into 3 transects of 16 stations each, and a total of 112 permanent based traps. With the exception of Sherman traps, all traps were closed at the end of each five-night trapping session and left in place. Sherman traps were removed, washed, and replaced each month. Trapping took place between June 1985 and October 1986, i.e. for 17 consecutive months. Each forest plot was trapped for five consecutive nights every month. During the course of the study, a total of 9520 trap nights was accumulated for each plot. Therefore, trapping effort for all six forest plots together totaled 57,120 trap nights. Fresh pineapples, oatmeal, and a cotton ball soaked with a commercial codfish oil solution were used as baits. Trap5 were checked every morning for captures and for adequacy of bait, which was replaced as needed. For each individual, the following information was recorded: 1. Species. If not readily identifiable, the individual was preserved for later taxonomic identification. The only individuals that were consistently prepared as skins and skulls were representatives of Ogzomys trinitatis and O,yzomys nignpes Specimens of the former species show a high degree of variability in skin color and pattern and were considered, during field work, as belonging to more than one species. Whenever possible, animals that died in the traps were also preserved. Moreover, live trapping and snap-trapping were conducted on other locations within the same forest type for preparation of voucher specimens and determination of stomach contents. Taxonomic identification of voucher specimens was provided through the kindness of the following specialists: Drs. M. Carleton, K. Creighton, L. Emmons, G Musser, P. Myers, and J. Patton. 2. Location in grid. 3. Individual identification. If already tagged with metal fish tags (fish and small animal tag, size 1, National Band and Tag Co., Newport, KY), the animal was promptly released; otherwise the animal was tagged. Individuals with positive taxonomic identification were released at the same station where captured. 4. Sex. 5. Weight. 6. Body length. 7. Tail length. 8. Ear length. 9. Hind foot length. 10. Reproductive condition. Female rodents were checked for perforated vaginas. I also noticed whether individuals were pregnant (in late stages) or lactating. Male rodents were considered in breeding condition if testes were descended. Female marsupials were considered in breeding condition if neonates were found in either pouch or attached to nipples, or if the nipple area showed signs of recent nursing by previous litter. Male marsupials with testes of reduced size were considered pre-pubertal, while 108 BULLErIN FLORIDA SI'ATE MUSEUM 34(3) for some species (Memchims nudicaudams, Marmosa incang Marmosa cinere4 Marmosa agilis, and PhUander opossum) abdominal or sternal scent gland activity indicated breeding condition. We also recorded any pouch young old enough to be sexed. 11. Behavior upon release. Analysis of valiance (Procedure :GLM: PC-SAS, SAS Institute, Cary, NC) was used to test the null hypothesis of absence of sexual dimorphism, using the following morphological characters: weight, body length, tail length, hind foot length, and ear length. Kruskal-Wallis tests were performed to test for differences in persistence times in traplines among sexes. Since the number of arboreal traps in each set of three traplines (N=36; 32 %) was different from that of terrestrial ones (N= 76; 68 %), analysis of differential trapping success of arboreal versus terrestrial traps was conducted with corrections to account for the differential probability of trapping in the different strata. Where sample size permitted, chi-square tests were performed to test the null hypothesis of lack of difference in substrate use among sexes. For this analysis, all ~ proportions were arc-sin transformed (Sohkal and Rohlf 1981). RESULTS AND SPECIES ACCOUNTS During the course of this study, a total of 692 individuals of 8 marsupial and 11 rodent species were caught 1366 times in all six forest plots (Table 1). Table 1. Trapping results for all six forest plots together during 17 months of trapping. Number of Total Number of Species First Captures Captures Caturomys philander 6 7 Didelphis marsupialis 144 404 Marmosa agilis 7 11 Marmosa cinerea 43 120 Marmosa incana 127 220 Marmosa microtarsus 1 1 Metachirus nudicaudatus 105 156 Phitander opossum 14 29 Abrawayomys ntschi 1 1 Akodon cursor 29 53 Nectomys squamipes 10 14 002Omys capito 1 2 On)zomys nigripes 5 5 Oryzomys subflavus 1 1 01?zomys trinitatis 84 99 Oxymyctems robmi 2 2 Rhipidomys mastacalis 1 1 Echimys sp. 239 1 1 Proechimys setosus 110 TOTAL 692 1366 FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 109 One individual of the species Monodelphis amencana was trapped on one of the auxiliary lines at the primary forest of Fazenda Montes Claros. Mean trapping success, averaged over 17 months of trapping at all sites, was 2.4 %. For most species, trapping success was much higher during the cold, dry months, dropping sharply around October with the onset of rains. Caluromys philander. Woolly opossums (Fig. 4) were only registered for the Rio Doce State Park. Although occasionally trapped on the ground, this species is mostly arboreal (Davis 1947; Nowak and Paradise 1983). In this study, only one individual was captured on the ground. Upon release, all individuals climbed into canopies, disappearing from sight at heights of about 10 m. Both the small number of arboreal traps (32%) relative to terrestrial ones, and the absence of traps in higher strata of the forest may have accounted for the low trapping success for this species. Only three males and three females were captured during the course of the study. One female was recaptured once. Morphometric data on these individuals are presented in Appendix lA. Didelphis marsupialis. Opossums occurred at all sites and represented the most common small mammal species captured in the majority of forest plots. This is a generalist species, found in all habitat types (Hunsaker 1977; Miles et al. 1981; Alho et al. 1986). Adult males were significantly heavier than females (Appendix lB). Hind feet were also sexually dimorphic. Other biometric param did not vary between sexes. Opossums have been previously described as being mainly terrestrial, although they are able to climb opportunely while foraging (Miles et al. 1981). These generalizations were confirmed here. Overall, 73% of D. marsupiahs captures were from terrestrial traps, while 93% of individuals released also remained on the ground. However, young opossums tend to use aerial substrate more often than adults. Approximately 23% of the individuals released (11 out of 37) climbed into trees and/or onto vines. The same result was obtained by Davis (1947). There were no significant differences in substrate use between adult males and females. A total of 79 males and 64 females was captured 196 and 204 times, respectively, with sex ratios of both number of individuals and total captures deviating very little from 1:1. Based upon recaptures, males did not persist for long periods on averaging 1.74 months. Females stayed within the traplines an average of 2.6 months, but this difference is not significant (p = 0.14). Record persistence times were achieved by one female and one male, which were recorded for a total of 12 months. Fleming (1972) found similar low persistence times for Didelphis mampialis in Panama. If the conclusions of Sunquist et al. (1987) with D. marsupialis from Venezuela are extrapolated to 110 BULLEnN FLORIDA SrATE MUSEUM 34(3) Figure 4. Adult Caluromys philander captured at the Rio Doce State Park. FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 111 the present study, the low persistence times can probably be attributed to high mortality rates rather than to dispersal. The distance traveled between successive captures were not significantly different between males and females. The Overall mean was 143.4 m (maximum= 400 m), slightly higher than the figure given by Fleming (1972). However, similar to grid trapping methods, traplines underestimate actual distances traveled in a night, which Sunquist et al. (1987) found, using radio transmitters, to average one kilometer. Even though trapping success during the first three months of study was very high, population levels, as reflected by the number of captures, did not appear to vary between dry and rainy seasons (Appendix lB). Reproductive activity, on the 6ther hand, was highly seasonal, with most females breeding just prior and well into the rainy season. The number of juveniles recorded also followed this same pattern (Appendix lB). Two females were recorded as having two consecutive litters, the first in August, and the second in October. The same has been observed in Brazil by Collins (1973 in Nowak and Paradiso 1983). The occurrence of two consecutive litters has also been reported elsewhere for D. marsupiahs in Colombia (Tyndale-Biscoe and Mackenzie 1976) and southeastern Brazil (Davis 1946). An interesting result was the decrease in opossum abundan'ces during the second year's late dry season, especially in the small forest plots. Smaller plots are where D. marsupialis was found to be most abundant during the first year of trapping, and capture success was much lower than for the previous year (Appendix lB). The reason for this apparent population crash is not yet clear, but may be due to a reduction in food resources that occurred as a result of a severe dry season. Although the reduction in abundance was observed across all plots, smaller forests experienced the highest decrease, lending support to this hypothesis, since populations inhabiting smaller plots should be more susceptible to a decrease in food resources. Average litter size of pouch young was 8.6, with a M:F sex riftio of 1.7:1.0, i highly skewed towards males (Appendix lB). Overall, two-thirds of females produced male-biased litters. This average litter size is much higher than those obtained in Colombia by Tyndale-Biscoe and Mackenzie (1976) or by Fleming (1973) in Panama. Both studies reported averages of approximately 6.5 young. , In addition, these studies found that sex ratios of pouch young did not differ from a 1:1 ratio. Austad and Sunquist (1986) were able to experimentally induce male-biased sex ratios in D. marsupialis by providing dietary supplementation. Therefore, relative to the sites studied by Fleming (1973) and Tyndale-Biscoe and Mackenzie (1976), the forests sampled in the present research may be more productive for common opossums, which may have resulted in the observed skewed sex ratios. 112 BULLErIN FLORIDA STATE MUSEUM 34(3) Marmosa agilis . This small arboreal marsupial (Fig. 5) can be fairly common in gallery forests of the Cerrado region (Nitikman and Mares 1987), but was infrequently caught in this study. The first individual of the species was recorded only after 12 months of trapping. In total, three males and four females were caught 11 times in two forest plots. Of the 11 captures, 10 (91 %) were in traps placed on trees or vines, which is a capture rate on arboreal traps similar to that obtained by Nitikman and Mares (1987). Alho et al. (1986) also found M agilis to be preferentially arboreal in gallery forest habitat. Except for two males and one female, which were recorded in traplines for two months, no other animal was recaptured in subsequent trapping sessions. The only individual for which data on movement between trapping stations were obtained traveled 40 m in consecutive nights, a figure very close to that determined by Nitikman and Mares (1987). Biometric data on M. agilis are presented in Appendix 1C. Mannosa cinerea. This species (Fig. 6) is one of the largest within the genus (Nowak and Paradiso 1983) and was fairly common at the Rio Doce Park, but was conspicuously absent from both Fazenda Montes Claros and Fazenda Esmeralda. Most individuals were captured during the cold, dry months (Appendix lD), with trapping success during the rainy season being fairly low. Lactating females, on the other hand, were only trapped during the rainy season, indicating that the high trapping success of the following months reflected the addition of recently born young into the population. A total of 8 males and 35 females was caught 120 times (Table 1), making it the species with the highest recapture rate in this study (31% for males and 37% for females). While sex ratio for first captures was 1.6:1.0, total captures approximated a 1.0:1.0 ratio. This might indicate that females are captured more frequently than males, which may be due to the fact that males are more transient than females (chi-square=5.17; df= 1, p <0.02) and averaged only about 1.8 months in traplines. Females, on the other hand, persisted in the areas for an average of 3.8 months. The longest tenancy in traplines was also by a female that was recorded present for 14 months. In addition, males also traveled farther between successive captures than females (males = 142 m, maximum:380 m; females=80 m, maximum:180 m), which also supports the contention that transient males are a frequent occurrence. On average, adult males were found to be slightly heavier than females. However, the species did not show sexual dimorphism for any biometric parameter measured (Appendix lD). Although trapped frequently on the ground, M. cinerea appeared to be primarily arboreal, especially the females (Appendix lD). The behavior upon release also revealed that, while 56% of the males remained on the ground, only 27% of the females displayed a similar FONSECA & KIERULFF: BRAZILIAN ATLANI'IC FOREST SMALL MAMMALS 113 Figure 5. Mannosa agilis, recorded in Fazenda Esmeralda and Fazenda Montes Claros (Photo by Kent H. Redford). 114 BULLETIN FLORIDA SrATE MUSEUM 34(3) behavior . Terborgh et al . (1984) list M. cinerea as being able to exploit several forest strata, from understory to canopy. Miles et al. (1981) determined M. cinerea to be mostly arboreal, with all nests also located above ground. Marmosa incana. This was the second most frequently caught species in this study (Fig. 7). ]t was present in all six forest plots sampled. We captured 80 males and 44 females a total of 220 times. Three individuals escaped before their sex could be determined. Sex ratio for first captures was 1.8:1.0, and 2.9:1.0 for all captures. Males also persisted on the traplines more than females (chi-square = 8.1, df = 1, p < 0.005), on average 1.9 and 1.2 months, respectively. Therefore, in contrast with M. cinerea, M. incana males appear to be more sedentary than females . M. incana males traveled an average of 64.7 m between successive captures (maximum = 200 m), while the only female recaptured in the same trapping session traveled 40 m. 0 1 Figure 6. Marmosa cinerea, recorded only at the Rio Doce State Park. FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 115 Marmosa incana adult males were sexually dimorphic for all morphometric characters measured. with males being, on average, 20% heavier than adult females (Appendix lE). Body length of males was also 14% larger than of females. Adult females also lack the sternal gland, which is functional in males during the breeding season. As with M. cinerea, trapping success increased markedly with the end of the rainy season (Appendix lE). This population growth is accounted for by the increase in the number of juveniles trapped (Appendix lE). The hypothesis that increased trapping success is due to juvenile recruitment is further supported by the observation that the occurrence of breeding males and especially breeding females is tied tightly to mid to late rainy season (Appendix lE). Mannosa incana can be characterized as a scansorial species, as its use of arboreal and terrestrial traps was approximately evenly split (Appendix lE). Figure 7. Marmosa incana, trapped at all study sites. 116 BULLETIN FLORIDA SrATE MUSEUM 34(3) The same result was obtained by using behavior upon release as a measure of differential substrate use. Approximately 53% of individuals released climbed trees or vines, while 47% remained on the ground. There were no statistical differences in substrate use between males and females (chi- square= 3.19; p > 0.05). Mannosa incanq appears highly insectivorous. Three stomachs analyzed in this study contained only insects, mostly belonging to the orders Coleoptera and Orthoptera. Nowak and Paradiso (1983) reported that most members of the genus Mannosa are insect and fruit eaters, although vertebrates are also occasionally consumed. Mannosa microtarsus. During the last month of the survey, October 1986, an adult male of this species was caught in an arboreal trap in the secondary forest of the Rio Doce Park. The species was previously described as being quite abundant in both secondary and primary forests of the Atlantic forest region (Davis 1947). Mannosa microtarsus is similar.in morphology to Mannosa agilis, and it is usually difficult to distinguish them. Morphometric data from this individual are as follows: weight = 31 g; body length = 106 mm; tail length= 148 mm; ear length= 14 mm; hind foot= 17 mm. Memchims nudicaudatus. This relatively large-bodied terrestrial didelphid (Fig. 8) was the third most common marsupial species trapped in this study (Table 1). The brown four-eyed opossum was present in all six forest plots surveyed. A total of 60 males and 45 females were caught 88 and 68 times, respectively. The sex ratio for first captures was equal to that of recaptures (1.3:1.0). Adult males were, on average, larger than females, with most biometric param measured proving sexually dimorphic (Appendix lF). Males were, on average, 25% heavier than adult females. Even though the species was relatively common throughout the year, a trapping success peak was observed following the rainy season (Appendix lF). This peak probably coincides with the onset of breeding in mid rainy season and the beginning of the dry months. Both lactating females and males with functional abdominal glands were frequently caught at this time (Appendix lF). Memchims nudicaudatus may also be able to produce a second litter in the same year. One female had a litter in March and the second in October. The number of pouch young ranged from 5 to 9, with an average of 7.2 young (Appendix lF). For the two females in which young were old enough to be sexed, there was a biased sex ratio towards females. Nowak and Paradiso (1983) regard brown four-eyed opossums as being arboreal, but the species caught in an arboreal trap was only once in 156 captures. Furthermore, only once was an individual observed to use aerial substrate upon being released. In fact, the large and non-graspable hind feet and clumsy behavior on above ground support do suggest a complete FONSECA & KIERULFF: BRAZILIAN ATLANTIC FORESr SMALL MAMMALS 117 terrestrial life for the species. Using a spool-line device, Miles et al. (1981) also found Metachirus nudicaudatus to be completely terrestrial . Terborgh et al. (1984) also regard gray four-eyed opossums as a species confined to the ground. Male and female brown four-eyed opossums did not differ in persistence times in traplines, with both sex classes averaging 1.7 months. Two males and two females were also recorded in traplines over a nine-month period. Distances traveled between successive trapping were among the lowest for marsupials, averaging 36.7 m (maximum = 120 m; N= 6) for males and 40 m (maximum = 80 m; N = 7) for females. Monodelphis americana. Short-tailed opossums have a fairly wide distribution in Brazil (Streilein 1982). Only one female was caught in an auxiliary line established in the primary forest at Fazenda Montes Claros, for Figure 8. Metachims nudicaudams, trapped at all study sites. 118 BULLErIN FLORIDA STATE MUSEUM 34(3) Figure 9. Philander opossum, captured at all sites , except at the Rio Doce State Park. the purpose of collecting voucher specimens. The capture site did not differ physiognomically from the regular traplines, and therefore the species might also occur consistently at this site. However, it is felt that trapping methods were not proper to adequately represent the species. Its small body size and foraging habits on the forest litter may have accounted for its low representation in the sample (Davis 1947). The biometric data on this individual are as follows: weight = 19 g; body length = 92 mm; tail length = 46 mm; hindfoot = 16 mm; ear length = 13 mm. Comparisons with data provided by Nowak and Paradiso (1983) indicate that this individual was probably a juvenile. Philander opossum. Gray four-eyed opossums (Fig. 9) were trapped in all forest plots, except at the Rio Doce Park. Fourteen individuals were caught during the course of this study. The occurrence of this large didelphid marsupial is apparently tied to the presence of standing or running water (Davis 1947; Handley 1976; Nowak and Paradiso 1983; Alho et al. 1986). As only a few transects occurred close to streams, this may explain the low trapping success for this species. As with other species, trapping success was much higher during the dry season (Appendix 1G). While only three individual females were recorded, these had a much higher recapture rate; 11 males were caught 17 times, while 3 females were captured in 12 different occasions. The longest persistence times on the traplines were 4 and 5 months, achieved by two adult females. All females caught were lactating, and two still had pouch young. Nowak and Paradiso ( 1983), based on several studies, concluded that P. opossuni breeds aseasonally. However, the smalI sample size of the present study does not allow any conclusions as to seasonality of reproduction. One female had pouch young in February, while the remaining were caught lactating in August and September. Each had 5 young attached to the nipples. The figure given by FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 119 Davis (1947) in the Atlantic forest was an average number of pouch young of 4.5, with a maximum of seven, while litter sizes in Nicaragua were found to be slightly larger (Phillips and Jones 1965). For one litter which sex could be determined, there were four female and one male young. Even though the small sample size precluded statistical analysis of sexual dimorphism, adult males were on average over 30% heavier than females (Appendix 16). As noticed by Nowak and Paradiso (1983), in this study the species also proved to be primarily terrestrial with only 17% of captures in arboreal traps, and 7% of individuals climbing trees after being released. Based on field observations, however, it is felt that gray four-eyed opossums are able, if needed, to efficiently use arboreal substrate. The same was suggested by Miles et al. (1981) and Crespo (1982). Abrawayomys mschi. This rare monotypic murid rodent, endemic to the Atlantic forest of eastern Brazil, is only known through its type specimen, collected in the state of Espirito Santo (Nowak and Paradiso 1983). A single adult male, with descended testes, was collected at the secondary forest of the Rio Doce State Park in January 1986, for which measurements are as follows: weight= 63 g; body= 128 mm; tail length= 146 mm; hind foot=31 mm; ear length= 20 mm. Akodon cursor. The genus Akodon comprises over 40 species, and A. cursor is among the largest. Sexual dimorphism is lacking in A. cursor (Appendix 1H), although males weighed more than females in another study (Nitikman and Mares 1987). This species occurred frequently in the Rio Doce Park secondary forest, but also was caught occasionally at other sites, especially in some auxiliary lines that were located in humid grasslands. Within the traplines, a total of 29 individuals was caught 53 times (Table 1), 27 of which were trapped at the Rio Doce Park. Sex ratios for first captures and all captures were, respectively, 2.3:1.0 and 3.4:1.0. Akodon cursor was previously described as being completely terrestrial (Crespo 1982; Alho et al. 1986; Nitikman and Mares 1987). However, the species demonstrated scansorial ability in this study, with approximately one- third of trapping success being obtained at arboreal traps. A distinct peak in population density was found between May and July. Very few individuals were recorded during the rainy season (Appendix 1-H). This was attributed to recruitment of young into the population just at the end of the wet and in the dry season. This conclusion is supported by the observation that the number of males with descended testes closely follows that of trapping success, and coincides with mid dry season (Appendix 1H). Gestation and subsequent weaning are close to five weeks (Nowak and 120 BULLETIN FLORIDA SrATE MUSEUM 34(3) Paradiso 1983), and the influence of reproduction activity on population density was readily noticeable in terms of increased trapping success. Akodon cursor appears to be primarily insectivorous. Individuals trapped in both grasslands and forests had high amounts of insects in their stomachs, which possibly indicates that insects are a part of their diets in all habitats. The species also makes use of seeds, fruits, and vegetative parts, especially those of the Graminae (Appendix 1H). Akodon cu,sor turnover rates appear very high, with 78% of individuals only being recorded during one trapping session. Maximum persistence was achieved by two individuals, but for only three and four months. The two individuals for which data on travel distances were available, moved 20 and 40 m between successive captures. Nectomys squam(pes. This semi-aquatic rat occasionally was trapped in the vicinities of small streams or flooded areas within forests. Alho et al. (1986) obtained 93% of all the captures of N. squam(pes in the Cerrado region on flooded areas. A few individuals in the present study were found over 500 m from any source of water, suggesting that the species occasionally may exploit non-aquatic habitats. The three stomachs analyzed contained only vegetative material, two exclusively fruit pulp. Nectomys squamipes is a fairly large-bodied rodent, with males attaining over 250 g (Appendix lI). A total of eight males and one female was recorded in tree plots, one at each of the study sites. Another individual escaped before sex could be determined. All but one individual was caught during the mid dry season, i.e. between May and August. Although adapted for semi-aquatic life, N. squamipes were trapped 38% of the times in arboreal traps. Two males persisted in the areas for three and four months, respectively. Oryzomys capito. A widespread habitat generalist, this rodent is extensively distributed in South America (Handley 1976). It is possibly mostly terrestrial, and in the Cerrado region was found more commonly at dense forests (Alho et al. 1986; Nitikman and Mares 1987). However, it was a rare species in the forests sampled in this study and was represented by only one male and one female at the Rio Doce Park, both caught in September 1986. All captures were on ground traps. The measurements of these individuals are, respectively: weight = 60 and 63 g; body length = 122 and 132 mm; tail length= 115 and 129 mm; hind foot= 32 and 34 mm; ear length=22 and 21 mm. FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 121 Oryzomys nigripes. This small cricetid rodent commonly occurred at auxiliary traplines located in grasslands, and it has wide distribution among Cerrado habitats (Alho et al. 1986; Nitikman and Mares 1987). It was, however, relatively uncommon in forests. Only five individuals, two males and three females, were caught in forest traplines. All trappings were dode in mid to late rainy season, i.e. between January and April. Two of these captures were on arboreal traps, although it is felt that the species is certainly more terrestrial and/or scansorial, as it was also observed by Crespo (1982). Alho and Pereira (1985) in Cerrado gallery forests and Veiga-Borgeaud (1982) in the Atlantic forest region determined that Oryzomys nignpes ( = eliums) makes extensive use of low shrubs, with nests located about 1 meter from ground. Stomach contents revealed high frequency of insects, complemented by fruit, seeds and leafy material (Appendix 1J). Barlow (1969 in Dalby 1975), and Crespo (1982) also found insects as part of O. nignpes' diet, even though in lower proportions. Biometric data are presented in Appendix 1J. Reproduction occurs throughout the year, albeit it may increase in frequency at some periods. Two females, one caught in February and the other in August, both had four fetuses, coinciding with two major reproductive peaks Observed by Veiga-Borgeaud (1982). A third female, trapped in April, had five implanted fetuses. Oryzomys subfavus. Only one female of this otherwise fairly common Cerrado species (Melo 1977; Alho and Pereira 1985; Alho et al. 1986) was caught in an arboreal trap in the forest traplines. The measurements of this individual are as follows: weight =92 g; body length= 165 mm; tail length= 173 mm; hind foot=34 mm; ear length=25 mm. Oryzomys trinitatis (=concolor). This cricetid (Fig. 10) was the most frequent rodent found in the forests sampled, being present at all sites. It is widely distributed, from Costa Rica to Paraguay (Nowak and Paradiso 1983), and found mostly in forested habitats (Alho and Pereira 1985; Nitikman and Mares 1987). A total of 84 individuals was captured 99 times (Table 1). Due to coat color variation, most individuals were prepared as skins. Therefore, recapture figures underestimate true recapture rates. Sex ratios at first captures were 1.35:1.00. In contrast to other small mammal species observed in this study, females were slightly heavier and longer than males. Other biometric param were found not to be significantly,limorphic (Appendix lK). The species appears to be mostly arboreal. Approximately 74% of males and 62% of females were captured in arboreal traps (Appendix lK). This figure is very close to that observed by Nitikman and Mares (1987) in a gallery forest of Central Brazil. Males and females did not differ in degree of arboreality (chi-square =2.40; df= 1; p>0.05). Since most individuals were prepared as voucher specimens, sample size was too small to evaluate the 122 BULLEnN FLORIDA SrATE MUSEUM 34(3) A Figure 10. Ogromys trinitatis , captured at all study sites . FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 123 differential use of substrate by analyzing behavior upon release. However, the few individuals that were released demonstrated climbing ability. Even though relatively uncommon in the forests of Manu National Park in Peru, O,yzomys trinitatis (=concolor) was listed by Terborgh et al. (1984) as being abie to use most all forest strata, from the understory to canopy. Although there were variations in trapping success on a monthly basis, there does not seem to exist any season characterized by high population, in contrast to most other species (Appendix lK). This might be a consequence of the absence of seasonal breeding. Reproduction seems to take place throughout the year, as breeding males and females were captured in 14 out of 17 months of study (Appendix lK). The two stomachs available for analysis indicated a high degree of insectivory. The stomach contents of an individual were 100% insects, while a second had 48% unidentified insects and coleoptera larvae, and 52% fruit pulp. Oxymycterus roberti. This genus includes semi-fossorial cricetids. The genus is represented in these forests by only one juvenile male and one adult female of the species roberti (Table 1). Members of this genus are mostly insectivorous (Borchert and Hansen 1983; Redford 1984); two stomachs analyzed in the present study yielded 100% insects, especially Coleoptera lame and ants. Species of the genus Oxymyctents art more commonly found in grasslands and inundated savannas (Borchert and Hansen 1983; Fonseca and Redford 1984; Redford 1984), a habit which can account for the rarity of the species in the forests of this study. Biometric data for the male and female are, respectively: weight=57 and 85 g; body length= 124 and 245 mm; tail length= 111 and 112 mm; hind foot=27 and 30 mm. R/:(pidomys mastacalis. This arboreal rat (Fig. 11) was only caught once in this study, in the month of September. It is found mostly in moist, forested habitats (Davis 1947; Dietz 1983; Fonseca and Redford 1984; Alho et al. 1986), even though it can also invade households (Nowak and Paradiso 1983; J. Stallings, pers. comm.). The adult male measurements are as follows: weight=72 g; body length= 137 mm; taiI length= 161 mm; hind foot= 27 mm; ear length = 18 mm. Echimys sp. Members of this genus are entirely arboreal (Hershkovitz 1969). This may explain the low success in recording this species in the forests sampled. It is possible that it preferentially occupies higher canopy strata. Davis (1947) never trapped Ecliimys below 5 m, while Miles et al . (1981) found nests of Echin:ys chgsums in tree cavities located at the canopy level. Only one adult male was caught on the primary f6rest of Fazenda Montes Claros in July 1986, in an arboreal trap. The voucher specimen could not yet be identified beyond genus, but may possibly either be Echimys brasiliensis 124 BULLETIN FLORIDA STATE MUSEUM 34(3) Figure 11. Rhipidomys mastacalis, a rare species in the forests trapped in this study. Gensu Moojen 1952) or a species which has not been described (L. Emmons, pers. comm.). This specimen had the following measurements: weight= 225 g; body length=215 mm; tail length= 205 mm; hind foot=38 mm; ear length= 14 mm. Proechimys setosus . Spiny rats were the third most abundant small mammal in this study (Table 1). Only opossums surpassed P. setosus in number of recaptures . Even though species of the genus Proechimys were also found to be quite common in other studies in the Atlantic forest (Davis 1947; Carvalho 1965; Avila-Pires and Gouvea 1977; Botelho and Linardi 1980; Miles et al. 1981; Fonseca et al. 1987), in the Cerrado (Fonseca and Redford 1984; Alho et al. 1986) and in the Amazon region (Bishop 1974; Emmons 1982; 1984; Terborgh et al. 1984; Malcolm 1987), spiny rats were surprisingly absent from FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 125 the Rio Doce Park traplines. This can possibly be attributed to higher predation rates by mammalian carnivores, owls and other predators at the Rio Doce Park (see Fonseca 1988). The park has a larger and richer carnivore fauna which exert a larger impact upon P. setoms populations than the comparatively depauperated predator community of the smaller forest plots. Sex ratios of first captures deviated little from a 1:1 sex ratio, with 56 male and 49 female individuals being trapped. Five individuals escaped before their sex could be determined. Females were recaptured at a slightly higher rate than males, yielding a sex ratio for all captures of 0.83:1.00. Although spiny rats were caught in all 17 months of the study, a distinct mid to late dry season peak in trapping success was notable (Appendix lL). This may result from the increase in the recruitment rate of juveniles into the population at this time (Appendix lL). However, reproduction does not seem to be strictly seasonal, as it was also observed by Bishop (1974) in Mato Grosso, Brazil. Pregnant and lactating females were present in all trapping sessions, although an increase in the frequency of breeding individuals could be observed both in mid rainy season and in mid dry season (Appendix lL). A few females had two litters in the same year, coinciding with the aboyementioned breeding peaks. There is no sexual dimorphism in this species of spiny rat, as none of the morphological characters measured significantly differ between sexes (Appendix lL). Proechimys setoms is entirely terrestrial, lacking any scansorial ability. All 239 spiny rat captures were conducted in terrestrial traps. Analysis of stomach contents indicate that spiny rats are primarily frugivorous, but also make opportune use of insects and seeds (Appendix lL). Emmons (1982) found Proechimys to be highly frugivorous, with insects being frequent in stomach samples. Although, on average, females persisted in traplines more than males (respectively, 2.7 and 2.0 months), there were no significant differences between sexes (chi-square=0.58, df= 1, p >0.05). The records for persistence times were obtained by two females who stayed, respectively, 10 and 12 months within traplines. Males and females also did not differ in mean traveled distances between successive captures with, respectively, 98.7 (maximum = 220 m) and 87.5 m (maximum=240 m). DISCUSSION Seasonality and Resource Use One of the most striking phenomena common to several small mammals of the western slopes of the Atlantic forest is the occurrence of seasonal 126 BULLETIN FLORIDA SrATE MUSEUM 34(3) reproduction. With the exception of the two most abundant rodents (Oryzomys t,initatis and Proechimys setosus), all four species with sufficient sample size to allow analysis proved to concentrate breeding in the late dry season and into the early to mid wet season. These were all marsupials. The evidence for these patterns derive from both the increase in trapping success starting in late wet season, represented mostly by the addition of juveniles, as well as the observation of adults in breeding condition just previous to that period. Furthermore, the number of first-time captures of all small mammals is significantly and negatively correlated with average monthly precipitation (r= 0.55;p <0.02), Which indicates that recently weaned juveniles are recruited into the population at the end of the rains. Akodon cursor is also a seasonal breeder, but unlike marsupials its reproduction is concentrated in the dry season. Davis (1946) and Laemmert et al. (1946) observed a pattern of cyclic reproduction among several Brazilian Atlantic forest small mammal species. Breeding was concentrated in the late dry winter and in the wet summer months, albeit less marked than in the present study. The possible reason for this may be that both these studies were in the eastern side of the Brazilian coastal mountains, where the dry season is shorter and less severe (Hueck 1972). These two studies have also revealed the lack of seasonality for some rodent species, especially ones from the genus Proechimys, which probably breed year-round throughout its range. A large number of African rodents seem also to closely tie the onset of reproductive activities to just before the end of the wet season (Delany 1986). Cerrado species appear to be equally divided between seasonaland year-round breeders (Dietz 1983; Alho 1982; Alho and Pereira 1985). In Panama, over 50% of mammals are seasonal breeders (Fleming 1973), although reproduction is concentrated in the three months of the dry season, which results in most young being weaned at the beginning of the rains. Other things being equal, reproductive output and juvenile survival of small mammals would be maximized if reproduction occurred when the environment is best in terms of resource availability (Fleming 1975). In fluctuating environments the energetic costs of pregnancy and lactation, coupled with the needs of adequate resources for newly weaned young, should pose constraints on the timing of reproduction. Lee and Cockburn (1985) state that all tropical didelphids are seasonal breeders, and the timing of reproduction is linked with availability of food. Marsupial food resources, in turn, are tied to the fluctuation in rainfall regimes of seasonal environments (Charles-Dominique 1983). Information on small mammal diets in this study are not sufficient to provide an accurate and longitudinal picture of diet composition. However, it is important to notice that all but the semi-aquatic Nectomys squamipes were observed to consume insects in variable quantities. All didelphid marsupials appear to be predominantly insectivorous (Nowak and Paradiso 1983), and FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 127 even large-bodied species, such as Didelphis marsupialis and Philander opossum, depend heavily on insect prey (Charles-Dominque 1983). Memchims nudicaudams, Mannosa spp., and Caluromys philander have also been listed as being consumers of ants and termites (Redford 1987), although the latter species may rely more heavily on fruits (Atramentowicz 1982). Results of studies on insect population fluctuations in the neotropics vary (see Elton 1975; Bigger 1976; Janzen and Schoener 1968; Janzen 1973; Wolda 1978), but most authors agree that samples obtained in wet season months are larger than comparable ones collected during the dry months (Davis 1946; Wolda 1978; Smythe 1982; Charles-Dominique 1983). Therefore, the trends of this study certainly support the notion that small mammal reproduction, especially that of marsupials, is influenced by availability of insect prey. However, until the variation in the cycles of prey populations across the year are described, explaining timing of reproduction of small mammals as a consequence of abundance of insect prey in the late wet to early rainy season will remain speculative. Fruits and seeds were the other frequent items present in the stomachs of small mammals in this study, although it is felt that rodents may depend more heavily on these resources than the more insectivorous marsupials. If the number of trees with fruits can serve as a measure of food resource availability for small mammals, the timing of reproduction of most small mammal species seems to track that of resources. In the Brazilian Atlantic forest asynchronously fruiting trees may be found at every month of the year (Davis 1946). Nonetheless, fruit production of several plant species of seasonal neotropical environments has been shown to be influenced by precipitation, usually peaking just prior to the start of the rainy season (Foster 1982; Smythe et al. 1982; Charles-Dominique 1983). A second peak in fruit productivity can also be present at the end of the rainy season (Davis 1946). A preliminary study conducted in the Rio Doce Park (CETEC 1981) also indicated that there are tree species flowering and fruiting throughout the year, and also that there is a slight peak in the number of fruiting species in September and October, i.e. in early rainy season. I expected marsupials, which use invertebrates to a larger extent than rodents, to display a higher degree of seasonality due the more nuctuating nature of their food resources. This was observed in the present study. Rodents, on the other hand, can probably rely on resources which are more seasonably stable in the Atlantic forest, such as fruits, seeds, and leaf tissue, and thus can reproduce throughout the year. The two most common rodents, Proechimys setosus and Ogzomys Mnitatis, do not show seasonality in their reproductive patterns. Since they also have been shown to use and during some periods heavily rely on insects, the consequent larger resource spectrum should place lower limits on the reproduction of rodent species. 128 BULLETIN FLORIDA SrATE MUSEUM 34(3) While some small mammals may exploit resources on the forest floor, most of the tropical forest productivity is located above ground (Eisenberg and Thorington 1973). Therefore, even for primarily terrestrial species, some level of scansorial ability should prove advantageous. There was a high degree of overlap in substrate use among small mammal species in this study. Twelve species are shown to use both aerial and ground substrate regularly, even though their relative degrees of arboreality varied. Among marsupials, only Metachims nudicaudams was confined to the ground level of the forest. Only 4 of 11 rodent species can be safely classified as terrestrial, while the remaining are at least marginally arboreal. This is consistent with descriptions of other neotropical small mammal fauna, where the majority of species show some level of climbing ability (see August 1984; Fonseca et al. 1987; Nitikman and Mares 1987). It should be stressed that these findings do not imply that arboreality is the dominant mode of life among Atlantic forest small mammals. Of the species caught in the present study, only Caluromys philander, Mannosa cinereg Marmosa agilis, Oryzomys trinitatis, Rhipidomys mastacalis, and Echimys sp. can be regarded as predominantly arboreal. It does, however, indicate the widespread ability of a greater fraction of the community in exploring a tri- dimensional environment. This may be especially important during certain periods of the year. Charles-Dominique (1983) provided data indicating that the insect faunas of the canopy and of the undergrowth can fluctuate asynchronously with each other. Therefore, if a particular food resource is undergoing a period of seasonal shortage, those species with versatile habits would be at an advantage. It has been demonstrated before that in the highly seasonal savanna region of central Brazil gallery forests play an important role in maintaining overall mammalian species diversity (Fonseca and Redford 1984) and become crucial during periods of stress. Part of the reason for this may be linked to the higher productivity of the tri-dimensional gallery forest environment, when compared, during periods of moisture deficit, to that of the cerrado savannas. Life History Patterns While the occurrence of pouch litters with skewed sex ratios in neotropical marsupial species has not often been reported, it was interesting to find Didelphis marsupialis Mth litters having a predominance of males. Skewed sex ratios have only been achieved experimentally under a regime of diet supplementation, a procedure which does not always produce the predicted male bias outcome (Austad and Sunquist 1986). Therefore, it is important that future investigations address this question. The female-biased sex ratio FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 129 obtained for Metachims nudicaudams pouch young is not based on a large enough sample size and therefore remains inconclusive. Three out of four marsupial species, for which sample size was large enough to allow statistical inference, proved sexually dimorphic. Male Didelphis marsupialis also has been found to be larger than females elsewhere in South America (0'Connell 1979). Other Mannosa species also have larger and heavier males (Nowak and Paradiso 1983), the same being true for Philander opossum. No references have been found to indicate size differences for Metachims nudicaudams, except for the present study. While larger males usually have been associated with polygynous species in which males strongly compete for females (Ralls 1976), there was no evidence of territorial defense by any of the marsupials studied here. However, this does not necessarily preclude males from actively competing for females at overlapping ranges. Fierce fighting for estrous females has been observed elsewhere in Didelphis ma,supiahs (Austad and Sunquist 1986). Since marsupial breeding in this study was usually confined to a certain season, the potential for male-male competition for estrous females is likely. Rodents, on the other hand, are seldom dimorphic (see Eisenberg 1981), especially the smaller species. In only one species, Oryzomys trinitatis, could a size difference be found in this study. In this instance, however, females proved to be larger on average than males. This could be attributed to extra energetic demand placed on pregnant and lactating females, a cost non- existent for males who do not provide parental care (see Rails 1976). Moreover, since rodent young do not undergo, unlike marsupials, a teat attachment phase, female placental mammals may increase litter size above that of the teats. The increase in rodent litter sizes may be achieved if the female is larger and better nourished. Marsupial litter sizes, on the other hand, due to the obligatory teat attachment phase, are constrained by the number of teats. Furthermore, marsupials can spontaneously terminate lactation, or more often in smaller species reduce litter size if energetically stressed (Lee and Cockburn 1985). Young are also born after a very short gestation period, making the reproductive investment minimal. For these reasons no added security would be achieved by genetically fixed propensity for larger marsupial females. Population Turnover Although the area sampled by traplines was not enough to represent the home range of several species, especially the larger ones, the reduced persistence time of the average small mammal individual was nonetheless striking. Average persistence times ranged from 1.2 months for male Mannosa 130 BULLETIN PLORIDA STATE MUSEUM 34(3) incana to a maximum of 3. 8 months for Marmosa cinerea females, with most species remaining in trapIines within the range of 1.7 to 2.7 months. While it is reasonable to assume that some of the disappearances can be attributed to home range shifts (see Nitikman and Mares 1987) and/or juvenile migration and dispersal (Lidicker 1975), predation might also play an important role in increasing turnover: rates among tropical small manimals. Monthly turnover rate for the small cricetid rodent Akodon cursor approached 80% in this study. Several authors have suggested that predators might cause variations in local abundances (August 1983), and sometimes they do take a large percentage of small mammal standing biomass (Hershkovitz 1969; Pearson 1985; Emmons 1987; Sunquist et al. 1987). Predation in traps was observed in this study on several occasions and it is felt that it may play a major role in the structure of these communities (Fonseca 1988). LITERATURE CITED Alho, C. J. R 1981. Small mammal populations of Brazilian Cerrado: The dependence of abundance and diversity on habitat complexity. Rev. Brasil. Biol. 41(1):223-230. . 1982. Brazilian rodents: their habitats and habits. In M. A Mares and H. H. Genoways (eds.). Mammalian Biology in South America. Spec. Publ. Ser. 6, Pymatuning Lab. Ecol., Univ. Pittsburgh Press, Pittsburgh. , and L. A. Pereira. 1985. Population ecology of a Cerrado rodent community in Central Brazil. Rev. Bras. Biol. 45(4):597-607. , , and A. C. Paula. 1986. Patterns of habitat utilization by small mammal populations in cerrado biome of central Brazil. Mammalia 50(4):447460. Alonso, M. T. A. 1977. Vegetacao. Regiao Sudeste. In Geografia do Brasil. Instituto Brasileiro de Geografia e Estatistica, IBGE, Rio de Janeiro, R J. August, P. V. 1983. The role of habitat complexity and heterogeneity in structuring tropical mammal communities. Ecology 64:1495-1513. . 1984. Population ecology of small mammals in the Llanos of Venezuela. In R. E Martin and B. R Chapman (eds.). Contributions in Mammalogy in Honor of Robert L. Packard. Texas Tech. Univ., Lubbock, Texas. Austad, S. N., and M. E. Sunquist. 1986. Sex-ratio manipulation in the common opossum. Nature 324:58-60. Avila-Fires, F. D., and E. Gouvea. 1977. Mamiferos do Parque Nacional do Itatiaia. Bol. Mus. Nac. 291:1-29. Atramentowicz, M. 1982. Influence du milieu sur l'activite locomotrice et la reproduction de Caturomysphilander (L). Revue d'Ecologie 36:376-395. Bigger, M. 1976. Oscillations of tropical insect populations. Nature 259:207-209. Bishop, I. R 1974. An annotated list of Caviomorph rodents collected in North-Eastern Mato Grosso Brazil. Mammalia 38(3):489-502. Borchert, M.,and.R. L. Hansen. 1983. Effects of flooding and wildfire on valley side wet campo rodents in Central Brazil. Rev. Bras. Biol. 43(3):229-240. Botelho, J. R., and P. M. Linardi. 1980. Alguns ectoparasitos de roedores silvestres do municipio de Caratinga, Minas Gerais, Brasil. I. Relacoes pulga/hospedeiro. Rev. Bras. Ent. 24(2):127-130. Cabrera, A. 1951 Catalogo de los mamiferos de America del Sun I. Rev. Mus. Cienc. Nat. "Bernardino Rivadavia' 4:1-309. FONSECA & KIERULFF: BRAZILIAN ATLANI'IC FOREST SMALL MAMMALS 131 1961. Catalogo de los mamiferos de America del Sun II. Rev. Mus. Cien. Nat. "Bernardino Rivadavia," 4:310-732. Carvalho, C. T. 1965. Bionomia de pequenos mamiferos da Boraceia. Rev. Biol. Trop. 13(2):239-257. CETEC 1981. Programa de pes4uisas ecologicas no Parque Estadual do Rio Doce. Sistema Operacional de Ciencia e Tecnologia-SOCT-Fundacao Centro Tecnologico de Minas Gerais-CHIEC, Belo Horizonte, M. G., Brazil. Relatorio Final, Segundo Volume. Charles-Dominique, P. 1983. Ecology and social adaptations in didelphid marsupials: comparison with eutherians of similar ecology. In J. F. Eisenberg and D. G. Kleiman (eds.). Advances in the Study of Mammalian Behavior. Amer. Soc. Mamm., Spec. Publ. 7. Coimbra-Filho, A F., and & A. Mittermeier. 1977. Conservation of the Brazilian lion tamarins (Leontopithecus rosalia). In Prince Rainier of Monaco and G. H. B6urne (eds.). Primate Conservation. Academic Press, New York. Crespo, J. A. 1982. Ecologia de la comunidad de mamiferos del parque nacional Iguazu, Missiones. Rev. Mus. Arg. "Bernardino Rivadavia," 3:(2):45-162. Dalby, P. L. 1975. Biology of Pampa Rodents. Michigan State Univ. Ser., Vol. 5, No. 3. Davis, D. E. 1946. The annual cycle of plants, mosquitoes, birds and mammals in two Brazilian forests. Ecol. Monogr. 15:244-295. 1947. Notes on the life histories of some Brazilian mammals. Bol. Mus. Nac. 76:1-8. Delany, M. J. 1986. Ecology of small rodents in Africa. Mamm. Rev. 16(1):1-41. Dietz, J. M. 1983. Notes on the natural history of some small mammals in Central Brazil. J. Mamm. 64(3):521-523. Eisenberg, J. F. 1981. The Mammalian Radiations. An Analysis of Trends in Evolution, Adaptation, and Behavior. Univ. Chicago Press, Chicago. , and R. W. Thorington, Jr. 1973. A preliminary analysis of a Neotropical mammal fauna. Biotropica 5(3):150-161. Elton, C. 1975. Conservation and the low productivity of invertebrates inside neotropical rainforest. Biol. Cons. 7.3-15. Emmons, L. H. 1982. Ecology of Proechimys (Rodentia, Echymyidae) in South-Eastern Peru. Trop. Ecol. 23(2):280-290. 1984. Geographical variation in densities and diversities of non-flying mammals in Amazonia. Biotropica 16(3):210-222. . 1987. Comparative feeding ecology of felids in a neotropical rainforest. Behav. Ecol. Sociobiol. 20:271-283. Fleming, T. H. 1972. Aspects of the population dynamics of three species of opossums in the Panama Canal zone. J. Mamm. 53:619-623. . 1973. The reproductive cycles of three species of opossums and other mammals in the Panama Canal Zone. J. Mamm. 54:439-455. 1975. The role of small mammals in tropical ecosystems. In F. B. Golley, K Petrusewicz, and L. Ryszkowski (eds.). Small Mammals: Their productivity and Populatin Dynamics. Cambridge Univ. Press, Cambridge. Fonseca, G. A. B. 1985a. The vanishing Brazilian Atlantic forest. Biol. Conserv. 34(1):17-34. . 1985b. Observations on the ecology of the muriqui (Brachytetes arachnoides E. Geoffroyi 1806): Implications for its conservation. Prim. Cons. 5:48-52. . 1988. Patterns of small mammal species diversity in the Brazilian Atlantic Forest. Ph.D. Dissertation, Univ. Florida, Gainesville, FL. , L. F. B. Mello, and G. Herrmann. 1987. Inventariamento de mamiferos e estudo dos padroes de diversidade de especies de pequenos mamiferos na EPDA de Peti-Minas Gerais. Report to CEMIG, Belo Horizonte, M. G. , and K R. Redford. 1984. The mammals of IBGE's ecological reserve and an analysis of the role of gallery forests in increasing diversity. Rev. Bras. Biol., 44:517-523. Foster, R. B. 1982. The seasonal rhythm of fruitfall on Barro Colorado Island. In E. G. Leigh, Jr., A. S. Rand, and D. M. Windsor (eds.). The Ecology of a Tropical Forest-Seasonal Rhythms and Long-Term Changes. Smithsonian Inst. Press, Washington, D. C. Handley, C. 0., Jr. 1976. Mammals of the Smithsonian Venezuela project. Brigham Young Univ. Sci. Bull. 20:1-91. 132 BULLErIN FLORIDA SI'ATE MUSEUM 34(3) Hershkovitz, P. 1969. The evolution of mammals on southern continents. VI. The recent mammals of the neotropical region: A zoogeographical and ecological review. Quart. Rev. Biol. 44:1-70. Hueck, K 1972. As Florestas da America do Sul. Editora Poligono, Sao Paulo, Brazil. Hunsaker, D. 1977. Ecology of New World marsupials. In D. Hunsaker (ed.). The Biology of Marsupials. Academic Press, New York. Janzen, D. H. 1973. Sweep samples of tropical foliage insects: Effects of seasons, vegetation types, elevation, time of day and insularity. Ecology 54:687-708. Janzen, D. H.,and T. 1. Schoener. 1968. Differences in insect abundance and diversity between wetter and drier sites during a tropical dry season. Ecology 49:96-110. Lacher, T. H., Jr., M. A. Mares, and C. J. R. Alho. in press. The structure of a small mammal community in a central Brazilian savanna. In K R. Redford (ed.). Mammals of the Americas: Essays in Honor of Ralph M. Wetzel. Univ. Chicago Press, Chicago. Laemmert, H. W., L. C. Ferreira, and X M. Taylor. 1946. Investigation of Vertebrate Hosts and Arthropod Vectors. Amen J. Trop. Med. 26(suppl.):23-69. Lee, A. K, and A. Cockburn. 1985. Evolutionary Ecology of Marsupials. Cambridge Univ. Press, London. Lidicker, W. Z., Jr. 1975. The role of dispersal in the demography of small mammals. In F. B. Golley, K Petrusewicz, and L. Ryszkowski (eds.). Small Mammals: Their Productivity and Population Dynamics. Cambridge Univ. Press, Cambridge. Lovejoy, T. E., J. M. Rankin, R. 0. Bierregaard, K S. Brown, L. H. Emmons, and M. E. Van der Voort. 1984. Ecosystem decay of Amazon forest fragments. In M. H. Nitecki (ed.). Extinctions. Univ. Chicago Press, Chicago. , R. O. Bierregaard, A. B. Rylands; J. R. Malcom, C. E. Quintela, L. H. Harper, K S. Brown, A. H. Powell, G. V. N. Powell, H. O. R. Schubart, and M. B. Hays. 1986. Edge and other effects of isolation on Amazon Forest fragments.In'M. E. Soule (ed.). Conservation Biology. Sinauer Press, Sunderland, Massachussets. Malcolm, J. 1987. Small mammal abundances in isolated and non-isolated primary forest reserves near Manaus, Brasil. In Resumos do Congresso Brasileiro de 76010gia, Juiz de Fora, Minas Gerais, Brasil. Mares, M. A., and Genoways, H. H. 1982. Introduction. In M. A. Mares and H. H. Genoways (eds.). Mammalian Biology in South America. Spec. Publ. Pymatuning Lab. Ecol., Univ. Pittsburgh, Pennsylvania. Mello-Leitao, C. 1946. As zonas de fauna da America Tropical. Rev. Bras. Geogr. 8:71-118. Melo, D. A. 1977. Observacoes preliminares sobre a ecologia de algumas especies de roedores do Cerrado, municipio de Formosa, Goias, Brasil. Rev. Bras. Pesq. Med. Biol. 10(1):3944. Miles, M. A., A. A. de Souza, and M. M. Povoa. 1981. Mammal tracking and nest location in Brazilian forest with an improved spool-and-line device. J. Zool. London 195:331-347. Mittermcier, K A, A. F. Coimbra-Filho, I. D. Constable, A. B. Rylands, and C. Valle. 1982. Conservation of primates in the Atlantic forests of Eastern Brazil. Int. Zoo. Yearbook 1982. Moojen, J. 1952. Os Roedores do Brasil. Institute Nacional do Livro, Rio de Janeiro, Brazil. Muller, P. 1973. The Dispersal Centers of Terrestrial Vertebrates in the Neotropical Realm. W. Junk Publishers, The Hague, The Netherlands. Nitikman, L. Z, and M. A. Mares. 1987. Ecology of small mammals in a gallery forest of Central Brazil. Ann. Carnegie Mus. 56:75-95. Nowak, A M., and J. L. Paradiso. 1983. Walker's Mammals of the World. Johns Hopkins Univ. Press, Baltimore, Maryland. O'Connell, M. A 1979. Ecology of didelphid marsupials from Northern Venezuela. In J.F. Eisenberg (ed.). Vertebrate Ecology in the Northern Neotropics. Smithsonian Inst. Press, Washington, D. C. Pearson, 0. P. 1985. Predation. InK H. Tamarin (ed.). Biology of New World Micron£s. Amen Soc. Mamm., Spec. Publ. 8. Phillips, C. J., and J. K Jones. 1965. Notes on reproduction and development in the four-eyed opossum, Philander opossum, in Nicaragua. J. Mamm. 50(2):345-348. Pine, R W. 1973. Mammals (exclusive of bats) of Belem, Para, Brasil. Acta Amaz. 3:42-79. Rails, K 1976. Mammals in which females are larger than males. Quart. Rev. Biol. 51:245-276. FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 133 Redford, K H. 1984. Mammalian myrmecophagy- Foraging, feeding and«food preference. Ph. D. Thesis, Harvard Univ., Cambridge, Massachusetts. . 1987. Ants and termites as food: Patterns of mammalian myrmecophagy. In H. H. Genoways (ed.). Current Mammalogy, Volume 1. Plenum Press, New York. Smythe, N. 1982. The seasonal abundance of night-flying insects in a neotropical forest. In E. G. I,eigh, A. S. Rand, and D. M. Windsor (eds.). The Ecology of a Tropical F6rest: Seasonal Rhythms and Long-Term Changes. Smithsonian Inst. Press, Washington, D. C. , E. Glanz, and E. G. Leigh, Jr. 1982. Population regulation in some terrestrial frugivores. In E. G. Leigh, A. S. Rand, and D. M. Windsor (eds.). The Ecology of a Tropical Forest: Seasonal Rhythms and Long-Term Changes. Smithsonian Inst. Press, Washington, D. C. Sokal, R. R.,and Rohlf, F. J. 1981. Biometry. W. H. Freeman Co., New York. Streilein, K R. 1982. Behavior, ecology and distribution of the South American marsupials. In M. A. Mares and H. H. Genoways (eds.). Mammalian Biology in South America. Spec. Publ. Ser., Pymatuning Lab. Ecol., Univ. Pittsburgh Press, Pittsburgh, Pennsylvania. Sunquist, M. E., S. N. Austad, and F. Sunquist. 1987. Movement patterns and home range in the common opossum (Didephis marsupialis). J. Mamm. 68(1):173-176. Terborgh, J. W., J. W. Fitzpatrick, and L. Emmons. 1984. Annotated checklist of bird and mammal species of Cocha Cashu biological station, Manu National Park. Fieldiana 21:1- 29. Tyndale-Biscoe, C. H., and R. B. Mackenzie. 1976. Reproduction in Didelphis marsupialts and D. albiventris in Colombia. J. Mamm. 57(2):249-265. Veiga-Borgeaud, T. 1982. Donnees ecologiques sur Olyzomys nigripes (Desmarest, 1819) (Rongeurs ; Cricetides) dans le foyer naturel de peste de Bamcao dos Mendes (Etat de Rio de Janeiro, Bresil). Mammalia 46(3):335-359. Vieira, C. 0. C. 1955. Lista remissiva dos mamiferos do Brasil. Arq. Zool. S. P. 8:341-474. Walter, H. 1971. Ecology of Tropical and Subtropical Vegetation. Oliver and Boyd Press, Edinburgh, Scotland. Wolda, H. 1978. Seasonal fluctuations in rainfall, food and abundance of tropical insects. J. Anim. Ecol. 47:369-381. 134 BULLEI'IN FLORIDA STATE MUSEUM 34(3) APPENDIX I -- A-L. A. CALUROMYS PHILAADER Biometric data for three males and three females of Caluromys phUander. Weight is given in gms, and other body measurements in mm. MALE FEMALE Mean St.Dev. Mean St.Dev. Weight 171.3 57.6 1803 403 Body length 210.7 30.0 206.8 11.4 Tail length 301.0 25.9 298.8 20.7 Hind Foot 36.3 2.5 36.3 2.6 Ear length 313 13 303 1.9 B. DIDELPHIS MARSUPIALIS Biometric data for adult male and female Didelphis marsupiatis. Weight is given in gms, and other body measurements in mm. N= number of individuals, St. Dev. = Standard Deviation. P refers to p-value associated with analysis of variance for sexual dimorphism (n.s.=p> 0.05;*=p< 0.05; „=p< 0.01;;... =P<0.001). MALE PEMALE Mean St.Dev. N Mean St.Dev. N P Weight 1,024.0 361.4 63 946.0 157.0 34 ** Body length 354.0 40.4 63 354.0 29.7 34 n.s. Tail length 3483 30.9 62 349.6 27.4 31 n.s. Hind Foot 58.1 3.9 62 56.1 4.3 34 ** Ear length 51.2 4.2 62 50.0 3.6 34 n.s. FONSECA & KIERULFF: BRAZILIAN ATLANTIC FORESr SMALL MAMMALS 135 Litter sizes and sex ratios of Didephis marsupialis pouch young. Number of Number of Total Sex Ratio ID Number Female Young Male Young Young M:F 0076 9 0088 9 0055 7 0072 9 0087 9 0126* 3 0156 7 0162* 4 0067 7 0192 7 0193 - 10 - 0209 3 6 9 2.0:1.0 0072 - 9 0081 11 0220 8 0251 8 0257 8 0258 - - 8 - 0274 3 5 8 1.7:1.0 0209 ** 2 7 9 33:1.0 0278 - - 5 - 0281 3 4 7 1.3:1.0 0227 5 5 10 1.0:1.0 0480 5 5 10 1.0:1.0 14/10 2 7 9 33:1.0 42P 2 5 7 23:1.0 11/12 4 4 8 1.0:1.0 Means 8.6 1.7:1.0 *Females were in poor condition, with multiple injuries. Therefore, their litter sizes were not used in calculating means. ** Second consecutive litter. 136 BULLEI'IN FLORIDA STATE MUSEUM 34(3) J100 80 - 60 - 40 - 0 IZZ-20 - i fit-fl) « lf= 'j I Jun Jul AugSepOctNovDecJan Feblvlar AprMayJun Jul AugSepOct EEZ] Number of Captures I Rainy Season Limits Figure 12. Number of captures of Didetphis marsupialis according to monthly trapping period. Straight lines enclose boundaries of rainy season (September to February). 14 - 12 - 10 -//8- 6- 4- 2- Ill I r _1=~_ p ~ Jun Jul Aug Sep Oct Nov Dec Jan FebMar Apr May Jun Jul Aug Sep Oct EE] Breeding Females ~ Number of Juveniles -*- Mean Aainfall (mm) Figure 13. Number of breeding females and juvenites of Dide*his marsupialis, according to monthly trapping period. Data points in line represent mean daily rainfall (in mm). FONSECA & KIERULFF: BRAZILIAN ATLANTIC FORESr SMALL MAMMALS 137 C. MARMOSA AGILE Biometric data for four males and one female Marmosa agilis. Weight is given in gms, and other body measurements in mm. MALE FEMALE Mean St. Dev. N Mean St. Dev. N Weight 29.5 11.3 4 25.0 - 1 Body length 109.3 13.2 4 99.0 - 1 Tail length 155.3 7.1 4 151.0 - 1 Hind Foot 18.3 13 4 17.0 - 1 Ear length 203 0.6 4 21.0 - 1 D. MARMOSA CINEREA Biometric data for adult male and female Marmosa cinerea. Weight is given in gms, and other body measurements in mm. P refers to p-value associated with analysis of variance for sexual dimorphism (n.s. =p >0.05; * =p <0.05; ** =p <0.01; ... =p < 0.001) MALE FEMALE Mean St. Dev. N Mean St. Dev. N P Weight 117.3 29.4 24 99.7 16.7 11 n.s. Body length 176.1 20.0 24 177.1 20.5 11 n.s. Tail length 258.9 26.0 24 261.7 17.9 11 n.s. Hind Foot 28.9 3.9 24 28.0 3.1 11 n.s. Ear length 30.4 2.8 24 29.7 2.1 11 n.s. 138 BULLEMN FLORIDA STATE MUSEUM 34(3) 20- ~ O 15 - 10 - 00 0 0 5- .1. O 11 ~ .i,-' 1 111= , 1,-,1 1 i - -]> Jun Jul AugSepOct NovDe(Jan FebMar AprMayJun Jul AugSepOct F551 Number of Captut.es MIM Rainy S eason Limits Figure 14. Number of captures of Mannosa cinerea according to monthly trapping period. Straight lines enclose boundaries of rainy season (September to February). Differential Trapping Success (in %) 1 60- 50 - 40 - 30 - 20 Males Females AN Terrestrial Traps F71 Arboreal Traps Figure 15. Differential trapping success (in percentages) of Mannosa cinerea in arboreal and terrestrial traps. FONSECA & KIERULPF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 139 E. MARMOSA INCANA Biometric data for adult male and female Marmosa incana. Weight is given in gms, and other body measurements in mm. P refers to p-value associated with analysis of variance for sexual dimorphism (n.s. =p > 0.05;* =p < 0.05; ** =p < 0.01; *** =p < 0.001). MALE FEMALE Mean St. Dev. N Mean St. Dev. N P Weight 82.3 23.6 62 68.4 23.1 23 ** Body length 1503 18.0 62 141.0 153 23 *** Tail length 203.9 18.2 62 191.2 18.2 23 *** Hind Foot 23.4 2.0 61 213 3.9 23 ** Ear length 28.1 2.4 61 263 3.0 23 *** 40 - 30- (- £-120 - 10-,577 , Jun Jul AugSepOct NovDec JanFebMar AprMay Jun Jul AugSepOct EE3 Number of Captures I Rainy Season Limits Figure 16. Number of captures of Marmosa incana according to monthly trapping period. Straight lines enclose boundaries of rainy season (September to February). 140 BULLETIN FLORIDA STATE MUSEUM 34(3) 10 e 8- - 0 6- 4- 2 - CD '' '' , -r /./ Jun Jul Aug SepOct Nov Ded Jan Feb Mar Apr May Jun Jul Aug Sep Oct EZZI Number of Juveniles -*= Mean Rainfall (mm) Figure 17. Number of juveniles of Mannom incana according to monthly trapping period. Data points in line represent mean daily rainfall (in mm). 10 F 8 2 fl=ft_,_4__ -li - ~0 & 1 1-{ Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct &W Breeding Males EES Breeding Females -*- Mean Rainfall (mm) Figure 18. Number of breeding mates and females of Marinosa incana according to monthly trapping period. Data points in line represent mean daily rainfall (in mm). FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 141 Differential Trapping Success (in %) 1 j 50- 40 - 30 - 20 Males Females I Terrestrial Traps E~| Arboreal Traps Figure 19. Differential trapping success (in percentages) of Mannosa incana in arboreal and terrestrial traps. F. METACHIRUSNUDICAUDATUS Biometric data for adult male and female Metachirus nudicaudams. Weight is given in gms, and other body measurements in mm. P refers to p-value associated with analysis of variance for sexual dimorphism (n.s. =p> 0.05; * =p < 0.05; " =p<0.01; -I =p<0.001) MALE FEMALE Mean St. Dev. N Mean St. Dev. N P Weight 352.7 94.0 51 280.8 57.1 33 *** Body length 252.0 28.7 51 232.1 25.4 32 ** Tail length 324.1 303 51 315.9 34.0 31 n.s. Hind Foot 44.9 3.1 51 42.3 3.3 32 * Ear length 37.4 3.9 50 36.5 2.9 32 n.s. 142 BULLEI'IN FLORIDA SI'ATE MUSEUM 34(3) Litter sizes and sex ratios of Metachirus nudicaudans pouch young. Number of Number of Total Sex Ratio ID Number Female Young Male Young Young M:F 271 - - 9 - 322 - - 5 - 338 - - 8 . 346 3 2 5 0.7:1.0 338* 6 3 9 03:1.0 Means 7.2 0.6:1.0 * Second consecutive litter. 20- ~ 15 - 0 C= C==2 10 - CD CD 5 - C I Z iii, i 1 0 Jun Jul AugSepOct NovDec JanFebMar AprMay Jun Jul AugSepOct E33 Number of Captures ~ Rainy Season Limits Figure 20. Number of captures of Memchirus nudicaudams according to monthly trapping period. Straight lines enclose boundaries of rainy season (September to February). FONSECA & KIERULFF: BRAZILIAN ATLANI'IC FORESr SMALL MAMMALS 143 10- 8- 6-1' - 1 1 Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug SepOct [~ Breeding Males EED Breeding Females -*- Mean Rainfall (mm) Figure 21. Number of breeding males and females of Metachims nudicaudatus according to trapping period. Data points in line represent mean daily rainfall (in min). G. PH1LANDER OPOSSUM Biometric data for seven males and three females of gray four-eyed opossum, Philander opossum. Weight is given in gms, and other body measurements in mm. MALE FEMALE Mean St. Dev. Mean St. Dev. Weight 394.9 98.7 295.0 39.7 Body length 282.0 51.3 285.0 48.2 Tail length 3253 21.6 295.0 18.0 Hind Foot 45.3 2.6 40.3 4.0 Ear length 32.7 3.0 29.3 3.1 144 BULLErIN FLORIDA SI'ATE MUSEUM 34(3) 10- - 8- 6- CD 4- 0 2- 7~?t 1 I 1 /// 0 1 .I I11/1I Jun Jul AugSepOctNovDec.Jan FebMar AprMay Jun Jul AugSepOct F51 Number of Captures j Rainy Season Limits Figure 22. Number of captures of Philander opossum according to monthly trapping period. Straight line encloses boundaries of rainy season (September to February). H. AKODON CURSOR Biometric data for adult Akodon cursor. Weight is given in gms, and other body measurements in mm. P refers to p-value associated with analysis of variance for sexual dimorphism (n.s. =p > 0.05; *=P<0.05; **=P<0.01; ***=P<0.001). MALE FEMALE Mean St. Dev. N Mean St. Dev. N P Weight 48.1 8.1 14 455 6.8 8 n.s. Body length 107.8 10.1 13 103.0 14.3 8 n.s. Tail length 98.8 9.9 12 94.9 10.4 8 n.s. Hind Foot 26.9 1.3 13 25.9 1.3 8 n.s. Ear length 18.1 1.6 14 18.8 0.9 8 n.s. FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 145 Stomach contents of fourAkodon cursor. PERCENTAGE OF ID Number Insects Leafy Material Seeds Fruits CEM1/10* 33 33 0 33 47P** 100 0 0 0 101P* 100 *** 0 0 0 106P* 96*** 0 4 0 * Forest traplines. ** Grasslands. *** Coleoptera and ants. Nes - ~ 14 - 12 - 10 - 8- 6- 4- 2- 1 1 /\ \-T 0«~~ ~w A ,/ /r=/r-/r=A \,/r-10 ' ' I 'l /' .i 'I 'i Jun Jul AugSepOct NovDec Jan FebMar AprMay Jun Jul AugSepOct FEl Number of Captures VA Rainy Season Limits Figure 23. Number of captures of Akodon cursor according to monthly trapping period. Straight line encloses boundaries of rainy season (September to February). 146 BULLEnN FLORIDA SI'ATE MUSEUM 34(3) 10 8 6 4 2 - 1%7 Jun Jul Aug SepOct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oet AMW Breeding Males -0- Mean Rainfall (mm) Figure 24. Number of males of Akodon cursor with descended testes according to monthly trapping period. Data points in line represent mean daily rainfall (in mm). L NECTOMYS SQUAMIPES Biometric data for five males and one female Nectomys squamipes. Weight is given in gms, and other body measurements in mm. MALE FEMALE Mean St. Dev. N Mean St. Dev. N Weight 197.0 67.0 5 165 - 1 Body length 188.4 19.9 5 190 - 1 Tail length 207.8 24.9 5 210 - 1 Hind Foot 50.0 2.8 5 50 - 1 Ear length 22.0 2.2 5 22 - 1 FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 147 3. ORYZOMYS NIGRIPES Stomach contents of Oozomys nignpes. PERCENTAGE OF ID Number Insects Fruit Leafy Material Seeds 105P 100 0 0 0 56P 20 0 0 80 CEM3 6 90 4 0 CEM2 30 12 58 0 Biometric data for two males and three females of Ory.omys nigripes Weight is given in gms, and other body measurements in mm. MALE FEMALE Mean St. Dev. N Mean St. Dev. N Weight 19.5 0.7 2 18.3 2.9 3 Body length 87.5 0.7 2 83.0 5.3 3 Tail length 126.0 4.2 2 116.8 6.1 3 Hind Foot 23.5 0.7 2 24.0 1.0 3 Ear length 16.0 0.0 2 17.3 0.6 3 148 BULLETIN FLORIDA SrATE MUSEUM 34(3) K ORYZOMYS TRINITATUS Biometric data for adult male and female Oryzomys trinitatis. Weight is given in gms, and other body measurements in mm. P refers to p-value associated with analysis of variance for sexual dimorphism (n.s. =p > 0.05; * =p < 0.05; " =p< 0.01; *** =p<0.001). MALE FEMALE Mean St. Dev. N Mean St. Dev. N P Weight 63.2 8.2 35 70.0 10.9 28 ** Body length 129.1 8.8 35 136.4 8.1 28 *** Tail length 152.3 10.2 35 159.4 18.2 27 n.s. Hind Foot 30.4 4.1 35 29.1 2.0 28 n.s. Ear length 18.6 3.0 35 19.0 1.8 28 n.s. Differential Trapping Success (in %) 70J 60- 1 1 50 - 40 - 30 - 20 Males Females ~ Terrestrial Traps EEZI Arboreal Traps Figure 25. Differential trapping success of male and female Oryzomys trinitatis in arboreal and terrestrial traps. FONSECA & KIERULFF: BRAZILIAN ATLANTIC FORESr SMALL MAMMALS 149 1& 1 14 - 12- 0 0 0 10- 8- 0 .-I 6- 4-2- D i ~ ~2~. Z Oilil' 111 Jun Jul AugSepOctNovDec Jan FebMar Apr May Jun Jul AugSepOct 533 Number of Captures ~ Rainy Season Limits Figure 26. Number of captures of O,yzomys trinitatis according to monthly trapping period. Straight line encloses boundaries of rainy season (September to February). 10 - 8- 6- Illf _E ~_~__b _j Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug SepOct &~ Breeding Males EEJBreeding Females -*- Mean Rainfall (mm) Figure 27. Number of breeding male and female Ogzomys trinitatis according to monthly trapping period. Data points in line represent mean daily rainfall (in mm). 150 BULLETIN FLORIDA STATE MUSEUM 34(3) L PROECHIMYS SETOSUS Biometric data for adult male and female Proechimys setosus. Weight is given in gms, and other body measurements in mm. P refers to p-value associated with analysis of variance for sexual dimorphism (n.s. =p > 0.05;* =p < 0.05; ** =p < 0.01; *** =P < 0.001). MALE FEMALE Mean St. Dev. N Mean St. Dev. N P Weight 270.7 36.2 43 259.6 43.4 41 n.s. Body length 200.2 14.2 42 196.5 18.9 40 n.s. Tail length 212.3 13.9 39 208.3 15.4 39 n.s. Hind Foot 51.2 2.1 43 50.7 4.6 40 n.s. Ear length 29.2 2.1 43 30.6 2.8 40 n.s. Analysis of seven stomachs of Proechimys setosus. PERCENTAGE OF ID Number Fruit Insects Seeds 39P 81 0 19 93P 64 36 0 95P 100 0 0 9OP 0 100* 0 91P 100 0 0 92P 100 0 0 79P 47 0 53 * Termites and Coleoptera larvae. FONSECA & KIERULFF: BRAZILIAN ATLANTIC FOREST SMALL MAMMALS 151 40- ~ 30 - .-I 20 - 0 10- 0 , Jun Jul AugSepOctNovDe(JanFebMar AprMayJun Jul AugSepOct REI Number of Captures //A Rainy Season Li m i ts Figure 28. Number of captures of Proechimys setosus according to monthly trapping period. Straight line encloses boundaries of rainy season (September to February). 10 8 6 4 2 co, 0 Jun Jul AugSepOct NovDec Jan Feb Mar Apr May Jun Jul Aug Sep Oct EZE] Number of Juveniles - Mean Rainfall (rnrn) Figure 29. Number of Proechimys setosus juveniles according to monthly trapping period. Data points in line represent mean daily rainfall (in mm). 152 BULLETIN FLORIDA STATE MUSEUM 34(3) 16 - 14 - 12- 10 - 8 - 6 ""F Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct 6~ Breeding Males EE] Breeding Females -*- Mean Rainfall (mm) Figure 30. Number of breeding male and female Proechimys setosus according to monthly trapping period. Data points in line represent mean daily rainfall (in mm). SMALL MAMMAL INVENTORIES IN AN EASTERN BRAZILIAN PARK Jody R. Stallings 154 BELLE['IN FLORIDA SI'ATE MUSEUM 34(4) ' S I '9* - - - .,2 , ~. ~ L, ~eiG*«,: tmv~.*FiORaw,f.ki 1 7 2 .- *:fsyl.1 t*j,5#i~~.:f -k:-fi '~ ~ . tit'.4272 -*E #,1,15Jr Plate 1 A. Didelphid marsupials (Didelphis marsupialis [ top] and Metachints nudicaudatus [bottom]) captured in Rio Doce State Forestry Park, Minas Gerais, Brazil (photographs by author). SMALL MAMMAL INVENTORIES IN AN EASTERN BRAZILIAN PARK Jody R. Stallings STALLINGS: BRAZILIAN SMALL MAMMAL INVENTORIES 155 Plate 1 B. Didelphid marsupials (Marmosa cinerea [ top] and Caturoniys philander [bottom]) captured in Rio Doce State Forestry Park, Minas Gerais, Brasil (photographs by author). 156 BULLEI'IN FLORIDA STATE MUSEUM 34(4) Plate 2 A. Cricetid rodents (Oryzongs subftavus [ top] and Abrawayaomys mschii [bottomD captured in Rio Doce State Forestry Park, Minas Gerais, Brazil (photographs by author).