Molecular discrimination of pouched four-eyed opossums from the Mamirauá Reserve in the Brazilian Amazon

Previous cytochrome B (CytB) mtDNA studies have suggested four species for the opossum genus Philander (four-eyed opossums), three (P. mcilhennyi, P. andersoni and P. opossum) from the Amazon and one (P. frenata) from the Brazilian Atlantic forest. During a faunal survey nine specimens of Philander sp. and four of Didelphis marsupialis were collected in the Mamirauá Sustainable Reserve, Amazonas State, Brazil. Preliminary analyses based on morphology and geographical distributions were not conclusive, suggesting that Philander specimens could belong to either P. andersoni or P. opossum. In order to elucidate the relationship of this taxon to the remaining Amazonian taxa, seven Philander and two Didelphis specimens animals were sequenced for the cytB mtDNA gene and compared to other previously studied taxa. The maximum likelihood (ML), neighbor-Joining (NJ) and maximum parsimony (MP) consensus bootstrap trees depicted six groups: Didelphis., P. frenata, P andersoni, P. mcilhennyi, P.o. opossum and Philander sp. and Philander canus in a common assemblage supported by significant bootstrap values, suggesting that the Philander sp. from Mamiraua in fact belongs to the species Philander canus.

According to Nowak (1999) the order Marsupialia presents 15 families, with the Didelphidae being the larger extant New World representative of this order.The Didelphidae encompasses 11 recent genera and 77 species, members of which occur from southeastern Canada through the eastern United States and Mexico and into South America to about 47 degrees South in Argentina.Kirsch (1977) proposed the recognition of two living subfamilies: Didelphinae (Marmosa, Monodelphis, Lestodelphis, Metachirus, Didelphis, Philander, Lutreolina, Chironectes) and Caluromyinae (Caluromys, Caluromysiops, and Glironia).Recently, nuclear gene sequence investigations by Jansa and Voss (2000) have found strong support for the monophyly of both the family Didelphidae (American opossums) and the subfamily Didelphinae.On the other hand, Gardner and Creighton (1989) suggested that the genus Marmosa is not monophyletic but is comprised of a series of separate taxa of unknown phylogenetic relationships, these authors creating the genus Gracilinanus to include some species while also suggesting that Marmosops, Micoureus, and Thylamys should be raised to the status of genera.The lack of monophyly in the Marmosa has been corroborated by all recent molecular analyses (Jansa and Voss 2000;Kirsch and Palma 1995;Patton et al. 1996).In relation to the genus Philander (four-eyed opossums), Tiedemann 1808, which is widely distributed and euritopic in the tropical Amazon forest, the taxonomic status of many of species and subspecies is still highly controversial.According to Voss and Emmons (1996), congeneric species of Philander may be sympatric in several parts of the Amazon forest.Based on morphological data, Gardner (1993) recognized two Philander species, P. andersoni distributed from southern Venezuela to Peru and P. opossum occurring from Mexico to Paraguay and northeastern Argentina.Mitochondrial cytochrome B (CytB) data prompted Patton and Silva (1997) to propose four Philander species, three (P.mclihennyi, P. andersoni and P. opossum) from the Amazon region and one (P.frenata) from the Brazilian Atlantic forest.
In the study described in this paper, we collected a sample of Philander specimens from the Mamirauá Sustainable Developmental Reserve, which is situated at the confluence of the Solimões and Japura rivers between 3°10' S and 64°41' W in the Brazilian state of Amazonas.Preliminary analyses based on morphology and geographical distributions were not conclusive, suggesting that the specimens could belong either to P. andersoni or P. opossum.Considering the large amount of data available on didelphid mitochondrial CytB, we decided to test whether or not this marker could be helpful in solving this question.During a faunal survey at the Mamirauá Reserve we used Tomahawk (17 x 17 x 61 cm) traps to capture nine Philander (two of which were released) and four Didelphis specimens (Table 1) which were subsequently sacrificed.The specimens were measured and their skin and skulls prepared and deposited at the Mastozoology collection of the Emilio Goeldi Paraense Museum, Belém (PA), Brazil.Fresh tissues (blood or liver) were collected and preserved in 95% ethanol for the DNA procedures.Table 1 show details of all the opossum specimens used in this study.
The DNA was extracted according to Sambrook et al. (1989) and about 440 base pairs of the CytB region of the mitochondrial DNA (mtDNA) was sequenced for seven Philander and two Didelphis specimens using the MVZ05 (5'CGAAGCTTGATATGAAAAA CCATCGTTG3') and MVZ14 (5'GGTCTTCATCTYHGGYTTACAAGAC3') primers (Smith and Patton, 1991).The polymerase chain reaction (PCR) was carried out in a final volume of 50 µL containing 10 ng of genomic DNA, 10 mM Tris-HCl (pH 8.85), 25 mM KCl, 5 mM (NH 4 ) 2 SO 4 , 0.2 mM dNTP, 50 pM of each primer and 1 unit of Taq DNApolymerase (Qiagen, USA).Amplification was performed in a MJ Research thermocycler with a cycling profile of 94 °C for 3 min followed by 30 cycles of 94 °C for 1 min, 40 °C for 45 s, 72 °C for 1 min and an additional extension period of 72 °C for 5 min in the last cycle.The PCR products were purified using ExoSAP-IT (USB, USA) and then sequenced using a dye-terminator cycle sequencing kit in a 377ABI automatic sequencer, according to protocols supplied by the manufacturers (Applied Biosystems, Foster, CA, USA).(Thompson et al., 1994) and subsequently slightly modified using the BIOEDIT sequence editor (Hall, 1999).
A divergence matrix was constructed using evolutionary model parameters selected by the Modeltest program version 3.06 (Posada and Crandall, 1998) and phylogenetic analysis was carried out using maximum parsimony (MP) and neighbor joining (NJ) analysis and the PAUP* program version 4b10 (Swofford, 2003).The Treefinder program (Jobb et al., 2004) was used in the maximum likelihood (ML) analyses including bootstraps.Tree topology confidence was assessed by bootstrapping 1000 pseudo-replicates for the MP and NJ analyses and 500 pseudo-replicates for ML analyses.
The CytB sequences were tested for saturation by plotting TN93 distances against transition and transversion values using the DAMBE program (Xia and Xie, 2001) and no indication of saturation was observed even at the highest divergencies.The evolutionary model that best fitted the data set was TN93 (Tamura and Nei, 1993) with rates following a gamma distribution.The parameters of the evolutionary model chosen by the ModelTest used in the PAUP* program were as follows: Base = 0.3205, 0.2335 and 0.1409; Nst = 6; Rmat = 1.0000, 6.5252, 1.0000, 1.0000 and 19.0258;Rates = gamma; Shape = 0.2352; and Pinvar = 0. Distances were estimated using both the PAUP4b10 and MEGA 2.1 (Kumar et al., 2001) programs, which generate similar values except that Mega 2.1 has the option of comparing average genetic divergencies between groups (Table 2) as well as group standard errors.
We found that the TN93 genetic distances (d) varied from 0.04 to 0.36, with our Philander sp.specimens from Mamirauá clearly being most closely related to P. canus (d = 0.004 ± 0.02) followed by P. mcilhennyi (d = 0.047 ± 0.02).The next grouping most strongly related to our Philander sp specimens was that of P. andersoni and P. opossum whose distance from the Philander sp (d = 0.07 to 0.08) were quite similar.Our data shows that P. frenata and Didelphis are undoubtedly the most divergent taxa, being this degree of divergence compatible to differences between distinct genera and similar results having been described previously (Patton and Silva, 1997).
Our ML, NJ and MP data suggests that the Philander sp.specimens from Mamirauá belong to the species Philander o. canus.However, the relationships of P. o. canus, P. andersoni and P. mclilhennyi, P. o. opossum, P. o. fuscogriseus, were not resolved.Indeed, the bootstrap values supporting P. o. canus, P. mcilhennyi and P. o. opossum (ML = 64%, NJ = 63 and MP = 0%) also suggests an unsolved phylogenetic relationship, as previously observed by Patton and Costa (2003).Clearly, the hypothesis of P. frenata from the Brazilian Atlantic forest being a different taxonomic entity (Patton and Costa, 2003) was also confirmed by our work.Indeed, our genetic divergence estimates between P. frenata and other Philander species shown in Table 2 strongly suggest that they may belong to different genus instead of different species of the same genus.Our genetic divergence estimates using the evolutionary model that best fitted the data showed divergence values of P. frenata in relation to the other opossums of the same scale of those from Didelphis in relation to the other opossums.Considering our results, the mitochondrial cytochrome B sequences proved to be useful as a complementary tool for taxonomic inference.As pointed out by Patton and Costa (2003)  supials will require a concentrated effort of multiple individuals working in collaboration, and we believe that the work reported in our present paper was an initial step in this direction.

Figure 1 -
Figure 1 -Consensus tree of 433 base pairs of the mitochondrial Cytochrome B (CytB) gene with bootstrap values on the right corresponding to maximum likelihood (ML), Neighbor-Joining (NJ) and Maximum parsimony (MP) analyses, respectively.

Table 1 -
Specimens and cytochrome B (CytB) sequences used in our study.

Table 2 -
Tamura and Nei (1993) rainforest mar-Genetic distances estimated according toTamura and Nei (1993)modeled by a gamma distribution with alpha parameter equal to 0.23 (lower diagonal).Standard errors estimated using bootstrap (upper diagonal).