MOLECULAR PHYLOGENETICS AND EVOLUTION Molecular Phylogenetics and Evolution 27 (2003) 113–120 www.elsevier.com/locate/ympev
Napoleon Bonaparte and the fate of an Amazonian rat: new data on the taxonomy of Mesomys hispidus (Rodentia: Echimyidae)q Ludovic Orlando,a Jean-Francßois Mauffrey,b Jacques Cuisin,c James L. Patton,d Catherine H€ anni,a and Francßois Catzeflisb,* a
CNRS UMR 5534, Centre de G en etique Mol eculaire et Cellulaire, Universit e Claude Bernard, Lyon I, 16, Rue Rapha€ el Dubois, B^ atiment G. Mendel, Villeurbanne Cedex 69622, France b CNRS UMR 5554, Laboratoire Pal eontologie, Case Courrier 064, Universit e Montpellier 2, Montpellier 34095, France c Laboratoire ‘‘Mammif eres et Oiseaux’’, Museum National d ÕHistoire Naturelle, 55, rue Buffon, Paris 75005, France d Museum of Vertebrate Zoology, University of California, 3101 Valley Life Sciences Building, Berkeley, CA 94720, USA Received 17 May 2002; revised 4 September 2002
Abstract The spiny rat Mesomys hispidus is one of many South American rodents that lack adequate taxonomic definition. The few sampled populations of this broadly distributed trans-Amazonian arboreal rat have come from widely separated regions and are typically highly divergent. The holotype was described in 1817 by A.-G. Desmarest, after NapoleonÕs army brought it to Paris following the plunder of Lisbon in 1808; however, the locality of origin has remained unknown. Here we examine the taxonomic status of this species by direct comparison of 50 extant individuals with the holotype at the morphometric and genetic levels, the latter based on 331 bp of the mitochondrial cytochrome b gene retrieved from a small skin fragment of the holotype with ancient DNA technology. Extensive sequence divergence is present among samples of M. hispidus collected from throughout its range, from French Guiana across Amazonia to Bolivia and Peru, with at least seven mitochondrial clades recognized (average divergence of 7.7% Kimura 2-parameter distance). Sequence from the holotype is, however, only weakly divergent from those of recent samples from French Guiana. Moreover, the holotype clusters with greater that 99% posterior probability with samples from this part of Amazonia in a discriminant analysis based on 22 cranial and dental measurements. Thus, we suggest that the holotype was originally obtained in eastern Amazonia north of the Amazon River, most likely in the Brazilian state of Amap a. Despite the high level of sequence diversity and marked morphological differences in size across the range of M. hispidus, we continue to regard this assemblage as a single species until additional samples and analyses suggest otherwise. Ó 2002 Elsevier Science (USA). All rights reserved.
1. Introduction q Abbreviations used: AMNH, American Museum of Natural History, New York; INPA, Instituto Nacional de Pesquisas da Amazonia, Manaus; MNHN, Museum National dÕHistoire Naturelle, Paris; MUSM, Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos, Lima; MV, Michael Valqui, University of Florida; MVZ, Museum of Vertebrate Zoology, Berkeley; USNM, National Museum of Natural History, Washington, DC. Field numbers of: ALG, Alfred L. Gardner; JLP, James L. Patton; JUR, Jay R. Malcolm; LHE, Louise H. Emmons; MDC, Michael D. Carleton; MNFS, Maria Nazareth F. Da Silva; LPC, Leonora P. Costa; V-, Francßois M. Catzeflis; VPT, Victor Pacheco T. * Corresponding author. E-mail addresses:
[email protected] (L. Orlando),
[email protected] (F. Catzeflis).
The taxonomy of many South American mammals is problematic, as few specimens have been collected and few taxa have been revised following initial species descriptions, mostly in the 19th century. Furthermore, the exact geographic origin of specimens, including holotypes, is often unknown, vague, or implausible, which makes revisionary studies even more difficult. Examples are common, such as the type locality ‘‘Asia America’’ specified by Linnaeus (1758) for the murine opossum, Marmosa murina (subsequently restricted to Suriname by Thomas, 1911), or ‘‘Virginia, Louisiana, Mexico, Brazil, and Peru’’ as the type locality for the common
1055-7903/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. doi:10.1016/S1055-7903(02)00372-X
114
L. Orlando et al. / Molecular Phylogenetics and Evolution 27 (2003) 113–120
opossum Didelphis virginiana (Kerr, 1792; later restricted to Virginia by Allen, 1901). The spiny rat Mesomys hispidus (Rodentia: Echimyidae) presents a similar vexing taxonomic problem. This beautiful, arboreal rodent was described by Desmarest in 1817 based on a specimen housed in the Paris National Museum of Natural History (MNHN) that had previously been deposited in the Museu Real dÕAjuda of Portugal in Lisbon. In 1808, the armies of Napoleon plundered Lisbon and removed to Paris many invaluable specimens of mammals collected by the naturalist Alexandre Rodrigues Ferreira (1756–1815) during his travels in Amazonian Brazil. Etienne Geoffroy St. Hilaire and Anselme-Gaetan Desmarest then described many new species of mammals based on these specimens. However, the original notes and catalogs were lost, and key ancillary data that provided information on the site of original collection remained unknown. Lacking such information, Desmarest (1817) gave the type locality for M. hispidus simply as ‘‘Amerique meridionale.’’
Recently, the taxonomy of the genus Mesomys has undergone significant progress. Two of the five species listed by Woods (1993) have been reassigned to the genus Makalata (Emmons, 1993) with an additional species, M. occultus, described by Patton et al. (2000). By current understanding, therefore, the genus consists of M. hispidus with a trans-Amazonian distribution, M. stimulax from eastern Amazonian Brazil south of the Amazon River, M. leniceps from the Andean foothills of northern Peru, and M. occultus from the central Amazon of Brazil. However, no thorough review of these forms has been attempted, and the ‘‘identification of its constituent species has long been problematic’’ (Voss et al., 2001). This is particularly true for M. hispidus, which displays considerable variation in body size and in molecular characters over its large geographic range (Patton et al., 2000; Voss et al., 2001). Patton et al. (2000) examined variation among 798 bp of the mitochondrial DNA cytochrome b gene for 37 specimens of M. hispidus collected from much of its
Fig. 1. Amazonian origins of the Mesomys specimens. Letters refer to the clades defined by cytochrome b sequence analysis. Stippled areas circumscribe localities of clades A, C, and F. Circles stand for M. hispidus, squares for M. occultus, and diamonds for M. stimulax. Localities are numbered: 1, La Paz; 2, Cuzco; 3, Madre de Dios; 4, Jurua1; 5, Jurua2; 6, Loreto; 7, Jurua3; 8, Jurua4; 9, Urucu; 10, Jau; 11, French Guiana (Sa€ ul, Les Nouragues, Cayenne); 12, Mato Grosso; 13, Santa Cruz; 14, Altamira; 15, Parauapebas. Specimens are detailed in Appendix A.
L. Orlando et al. / Molecular Phylogenetics and Evolution 27 (2003) 113–120
range (Guyana, Venezuela, Brazil, Peru, and Bolivia). These samples comprised five strongly differentiated clades that formed a monophyletic group in relation to both M. stimulax from eastern Brazil and the sympatric M. occultus. Voss et al. (2001) commented that these divergent samples of M. hispidus were divisible into two morphological groups, a small-toothed and generally small-sized animal in eastern Amazonia and the Guianan Region, and a large-toothed and large-bodied animal in western Amazonia. These observations thus raised the possibility that M. hispidus as envisioned by Patton et al. (2000) and other authors might be a composite of two or more species. Thus, a key question remains as to the provenance of the holotype of M. hispidus, as this would fix the name hispidus Desmarest to one or more of the geographic units currently identified and allow other available names to be applied to the remaining units. Tate (1939, p. 179) restricted the type locality of M. hispidus to ‘‘Borba, Rio Madeira, Brazil,’’ which is in the central Amazon within the range of the large-toothed western form. If correct, and if more that one species is definable within what is now called M. hispidus, then this name would appear to apply to the western form. Thus, following the suggestion of Robert S. Voss (in littera, 1999), we decided to clarify the taxonomy of M. hispidus (sensu lato) by using ancient DNA technology to sequence a mitochondrial fragment from the holotype, and to compare it to those samples from Amazonia and the Guianan subregion already available (Patton et al., 2000) or more recently obtained (see Fig. 1). We also compare cranio-dental measurements of the holotype with samples of M. hispidus from throughout Amazonia that span the known size variation in this taxon. By both sets of analyses, we reassess the question of the provenance of the holotype and species limits within this complex of spiny tree rats.
115
with a mortar in liquid nitrogen. Decalcification, protein digestion, DNA purification, amplification, cloning, and sequencing were conducted as described in Orlando et al. (2002) with slight modifications (2 ll of ancient extract per PCR tube; annealing temperature 49–50 °C; 2–7 clones sequenced per amplification product). Primers were designed by comparison with Mesomys spp. cytochrome b sequences in order to amplify three short fragments in any species of the genus (first fragment, 166 bp: 22L 50 -CACCCCCTTGCCAAAATTAT and 188H 50 -TCTGTGGTGATGTCGGAA; second fragment, 145 bp: 400L 50 -CCCTGAGGACAAATATC TTT and 545H 50 -TGGGCTAAGAAACGGAAGGT; and third fragment, 166 bp: 479L 50 -TAGTTGAATGA ATCTGAGGAGG and 645H 50 -GTCCTAATTGTAG TCTGAGG; L and H stand for light and heavy, respectively). The second and third fragments overlap through 25 bp. Three independent blanks were carried out for each set of PCR experiment: (i) an extraction blank on mock extractions, (ii) a PCR blank, that remained open during the preparation of PCR mixes, and (iii) an extraction room blank, that remained open when ancient extracts were added into the PCR tubes. All blanks were negative, suggesting no exogenous contamination occurred. To insure the holotype sequence was not due to contamination of the sample during the storage in MNHN at Paris, two tiny pieces of dry skin from the same storage drawers were extracted: MNHN 1903–756 from Proechimys longicaudatus (in the collection since 1902) and MNHN 1962–1023 from Echimys villosus (in the collection since ca. 1850). No amplification was obtained using the three couple of primers on Proechimys and Echimys extracts, although extensive low stringency attempts. 2.2. Molecular analysis of field-caught specimens
2. Materials and methods 2.1. Molecular analysis from museum specimens The holotype of M. hispidus consists of a mounted skin on a wooden board (specimen MNHN 1998–2075) in the Laboratoire Mammiferes and Oiseaux and of a cleaned skull which was until recently located in the nearby building of the Laboratoire dÕAnatomie Comparee (Catalog No. A-7668). As detailed in Voss et al. (2001, p. 153), both items most probably belong to the same animal; consequently, they have been stored together since 1998, under the same catalog number (MNHN 1998–2075). A tiny piece of the dry dorsal skin (near the basis of the tail) was sampled from the holotype by staff at MNHN. The extraction procedure was conducted in Lyon in a devoted lab with ancient DNA facilities. 100 mg of dried skin was reduced to powder
DNA was extracted from 95% ethanol-preserved pieces of liver using phenol/chloroform, proteinase K/ RNAse methods (Sambrook et al., 1989) and Wizard DNA clean-up system (Promega). In Montpellier, the complete cytochrome b of three Mesomys cf. hispidus specimens recently caught in French Guiana (V-924, V-953, and V-1118, respectively, from Arataye, Les Nouragues, and Sa€ ul) were amplified and sequenced using the conserved primers L15 and H6, as reported in Steiner et al. (2000). In Lyon, in rooms independent from the ancient DNA lab, a fragment of 623 bp for an additional six vouchered animals from French Guiana was obtained 10 months after the analysis of ancient specimens (primers 22L and 645H; Appendix A). Both strands were sequenced, either by manual sequencing with the thermo-sequenase radiolabeled terminator cycle sequencing kit (Amersham) or by use of an automatic sequencer (ABI 373 Perkin–Elmer) with Dye
116
L. Orlando et al. / Molecular Phylogenetics and Evolution 27 (2003) 113–120
Terminator Cycle Polymerase Sequencing (Applied Biosystems).
3. Results 3.1. Cytochrome b analysis
2.3. Sequence analysis Cytochrome b sequences from all other Mesomys were retrieved from Patton et al. (1994, 2000) or provided by one of us (J.L.P.). Proechimys cuvieri and P. cayennensis (EMBL AJ251401 and AJ251395) were used as outgroups. Alignments were checked manually. Both nucleotide and amino acid translated sequences were checked for irregularities that could indicate nuclear homologs (Arctander, 1995; Zhang and Hewitt, 1996). Alignment, cytochrome b translation, and estimates of KimuraÕs two parameters (K2P) divergence were performed using MUST (Philippe, 1993) software. Genealogical relationships were reconstructed by maximum likelihood (ML), distance (NJ—neighbor-joining), and parsimony (MP) methods using Paup 4.0b8. For distance analyses, we applied the general time reversible (GTR: Yang, 1994) estimator that corrects for differences in nucleotide frequencies as well as unequal rates of transitions and transversions. Gamma shape parameters (with eight rate categories) were calculated for a reduced data set (24 sequences) under ML using the NJ tree as a starting point. For ML analyses we specified the GTR model of sequence evolution with an estimate of the relative rates of each kind of substitution. Rate variation among sites was described by a Gamma distribution with eight categories. Parsimony trees were generated using the heuristic search option in Paup 4.0b8 with TBR branch swapping and 100 random taxon additions. The robustness of topologies was examined by 1000 bootstrap replicates for both NJ and MP and 100 for ML.
The 798 bp sequences from the 45 individuals of M. hispidus now available form a monophyletic assemblage of six well-differentiated clades in relation to both M.
2.4. Morphological analyses We took three external measurements (TAL, tail length; EAR, ear height; and HBL, head and body length) from newly collected adult specimens from French Guiana. Twenty-two cranial and external variables for 40 adult individuals from along the Rio Jurua (localities Jurua1 to Jurua4 in Fig. 1) have been previously used in Patton et al. (2000, table 58, p. 201)); they were measured on the preserved skulls of our 11 French Guianan specimens (see Table 2 for legend). Thus, the data set includes 51 individuals originated from 3 geographic regions (Rio Jurua clade A: 21, Rio Jurua clade C: 19, and French Guiana: 11). We used one-way ANOVA on all measurements to compare the three geographic samples. As the holotype presents intermediate values for most measurements available, we conducted a Linear Discriminant Analysis to address its morphological affinities with each of the three geographic samples.
Fig. 2. Phylogenetic tree derived from a ML analysis of 51 Mesomys spp. cytochrome b sequences and rooted with two species of Proechimys. Nodal values are average amounts of divergence (KimuraÕs two parameters estimator). The robustness values of the ancestral segments numbered 1–7 (supported by at least 75% in at least one reconstruction procedure) are presented in Table 1. Empirical nucleotidic frequencies are 29.7% A, 26.9% C, and 12.7% G. Estimated substitution rates are 1.95 (AC), 11.96 (AG), 2.78 (AT), 0.90 (CG), 25.62 (CT), and 1.00 (GT). Alpha-parameter (8 categories Gamma rate) is 0.3211.
L. Orlando et al. / Molecular Phylogenetics and Evolution 27 (2003) 113–120 Table 1 Robustness values for the ancestral segments labeled 1–7 on the phylogenetic tree of Fig. 2 Clade
Segment
ME
MP
ML
Clade A (western Amazonia) Clade C (Rio Jurua, Rio Urucu) (French Guiana, holotype) Clade F Clades C and F M. stimulax and M. hispidus Genus Mesomys
1 2
99 99
89 99
89 97
3 4 5 6 7
100 87 66 99 100
99 88 77 93 100
100 77 72 100
In the data set, 183 out of 248 variable sites are parsimony-informative against 134 out of 205 among the 46 Mesomys cf. hispidus sequences. Abbreviations: ME, minimum evolution (distance with GTR estimator); MP, maximum parsimony; ML, maximum likelihood.
stimulax and M. occultus, regardless of analytical method used in tree-building (Fig. 2, mapped in Fig. 1). These include three broadly distributed units (clade A, from western Amazonian Brazil and adjacent Peru; clade C, from west-central Amazonian Brazil; and clade F, which ranges from French Guiana to the southern Amazon of central Brazil and eastern Bolivia) as well as three geographically limited haplotypes (clade B, from northeastern Bolivia; clade D, from southern Venezuela and adjacent Brazil; and clade E, from Guyana). A stretch of 126 (positions 42–168 in the alignment of the complete cytochrome b gene) and 205 nucleotides (positions 420–625) were successfully retrieved from the holotype of M. hispidus. This sequence is different from all other specimens of M. hispidus obtained to date, except that of V-924 (from Arataye, French Guiana), the samples of which were never stored together and were extracted by different groups in separate laboratories. Hence, there is no likelihood that the holotype sequence results from contamination from V-924. The holotype sequence is nested within a terminal clade with three specimens from French Guiana. The nucleotide uniformity of these specimens is noteworthy, as their complete (1140 bp) cytochrome b sequences differ by only 0.3% substitutions, on average. Only two substitutions in 623 bp were found in an additional six individuals from four localities (data not presented). The holotype and specimens from French Guiana are related to a clade of three specimens from the Brazilian state of Mato Grosso and the Bolivian state of Santa Cruz, which we have labeled as clade F (Fig. 2 and Table 1; bootstrap values 77–88%). Although support is weak, this clade appears to be the sister of the central Amazonian clade C (Fig. 2; see also Patton et al., 2000). The average K2P value for the node connecting clades C and F is about 4.8%. A larger divergence (7.2%) exists between members of the western Amazonian clade A and all remaining haplotypes of M. hispidus.
117
3.2. Morphological analysis Table 2 provides the values for three external and 22 cranio-dental variables of three samples of M. hispidus: those from clade A, clade C, and the French Guianan specimens of clade F. These samples encompass both the molecular and morphological diversity now recognized. Clades A and C differ by more than 7% in sequence, are partially sympatric in the middle Rio Jurua in western Brazil, and, while separable morphometrically by discriminant analysis, both are of the large-toothed western form of the species (Patton et al., 2000). clade F includes small-toothed animals from the eastern part of the sampled range in French Guiana. All animals examined are adults with fully erupted molariform teeth (age class 9 and 10; Patton and Rogers, 1983). Animals from French Guiana are smaller overall in relation to their Amazonian counterparts, with most measurements significantly different (one-way ANOVA significance levels in Table 2). Some individual variables (zygomatic breadth, ZYB; diastema, DIL; and maxillary toothlength, MTL) are almost completely discriminatory. The holotype is an unsexed adult and the skull has fully erupted and slightly worn dentition (age class 9). Some of its measurements are similar to those of the French Guiana samples (e.g., ROL, NAL, MTL, and OCB), but others are closer to those of specimens of clades A and C from western Amazonia (GSL, CIL, BAL, ORL, DIL, and MAB: Table 2). Thus, while the holotype has a small tooth row, in overall univariate characters the skull appears somewhat intermediate between typical large-toothed Amazonian and smalltoothed Guianan M. hispidus. Linear Discriminant Analysis, however, shows a clear morphological affinity of the holotype to the specimens from French Guiana (Fig. 3). The distributions of individuals from Amazonian and Guianan areas are completely non-overlapping on DF-1, with the second axis only partially separating clades or individuals within each geographic region. Moreover, the holotype both clearly clusters with French Guiana individuals and is assigned to that group with a posterior probability of 0.99. Based on this restricted geographic sampling, therefore, we suggest that the holotype came from a source area close to French Guiana, not from the central Amazon as argued by Tate (1939).
4. Discussion The 331-bp sequence available clearly shows that the holotype belongs to the clade otherwise comprised of specimens from French Guiana (Fig. 2). We are thus confident that animals from this region can safely be identified as M. hispidus (Desmarest, 1817). The question, then, remains as to the species status of the other
118
Table 2 Selected external and cranial measurements of M. hispidus from the Rio Jurua Basin (clade A, upriver; clade C, downriver), from French Guiana, and from the holotype MNHN-1998–2075 Clade A
184:2 6:9 13:4 1:0 178:2 12:3 44:36 1:63 39:84 1:36 33:48 1:49 23:37 0:73 11:22 0:71 14:54 0:64 12:66 0:66 7:74 0:50 8:69 0:53 12:53 0:43 9:92 0:54 7:28 0:24 4:05 0:42 15:11 0:66 5:42 0:35 19:72 0:65 9:75 0:40 8:59 0:34 19:70 0:64 10:05 0:55 3:55 0:32 14:36 0:46
(174–197) (11–15) (155–196) (41.61–46.93) (37.27–42.43) (31.05–36.40) (22.29–24.65) (10.00–12.51) (13.20–15.58) (11.33–14.11) (7.00–8.83) (7.84–9.82) (11.52–13.43) (9.07–10.93) (6.86–7.83) (3.42–4.85) (13.92–16.46) (4.92–6.06) (18.70–20.72) (9.04–10.59) (7.96–9.39) (18.57–20.89) (8.61–10.76) (2.88–4.24) (13.3–15.21)
N N N N N N N N N N N N N N N N N N N N N N N N N
¼ 13*** ¼ 21*** ¼ 13*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21*** ¼ 21 ns ¼ 21***
Rio Jurua
Clade C
192:4 10:4 13:4 0:9 186:1 15:3 44:18 1:49 39:63 1:57 33:49 1:50 23:37 0:78 10:95 0:61 14:17 0:67 12:35 0:76 7:64 0:40 8:74 0:40 12:54 0:53 9:85 0:51 7:1 0:30 4:15 0:32 14:96 0:69 5:25 0:37 19:67 0:89 9:97 0:36 8:56 0:31 19:46 0:82 10:04 0:71 3:62 0:21 14:54 0:64
(175–203) (12–15) (163–202) (41.59–47.32) (36.69–43.30) (31.12–36.81) (22.23–25.02) (9.90–11.98) (12.93–15.71) (11.26–14.32) (6.87–8.67) (8.10–9.63) (11.51–13.71) (9.09–10.81) (6.62–7.70) (3.42–4.77) (13.99–16.35) (4.62–5.77) (17.98–21.30) (9.34–10.63) (8.15–9.50) (18.51–21.05) (9.04–11.55) (3.11–3.99) (13.36–15.82)
N N N N N N N N N N N N N N N N N N N N N N N N N
¼ 7*** ¼ 17*** ¼ 7*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19*** ¼ 19 ns
French
Guiana
160:0 13:2 11:9 1:3 161:0 14:3 39:10 1:31 34:27 1:51 28:56 1:32 20:79 0:73 9:81 0:66 11:91 0:83 10:92 0:74 6:95 0:51 7:99 0:50 11:09 0:43 8:00 0:65 6:22 0:21 3:27 0:43 12:44 0:77 4:77 0:18 18:34 1:98 8:77 0:46 7:74 0:33 18:05 0:67 9:07 0:63 3:38 0:60 13:13 0:35
(141–178) (10–15) (140–182) (37.10–41.01) (31.19–36.20) (26.70–30.80) (19.95–21.90) (8.44–10.64) (10.14–13.22) (9.54–12.42) (6.11–7.86) (7.27–8.96) (10.35–11.66) (6.83–8.81) (5.91–6.67) (2.60–3.99) (11.31–13.40) (4.44–5.01) (13.21–20.60) (8.38–9.66) (7.34–8.41) (17.06–19.04) (8.38–10.08) (2.50–4.08) (12.49–13.58)
Holotype N N N N N N N N N N N N N N N N N N N N N N N N N
¼5 ¼ 10 ¼ 10 ¼ 10 ¼ 11 ¼ 10 ¼ 11 ¼ 11 ¼ 11 ¼ 11 ¼ 11 ¼ 11 ¼ 11 ¼ 11 ¼ 11 ¼ 11 ¼ 11 ¼ 11 ¼ 10 ¼ 11 ¼ 10 ¼ 10 ¼9 ¼ 11 ¼ 11
41.68 38.09 32.47
12.55 11.0 7.28 8.45 12.06 9.22 6.45 3.55 13.55 4.74
7.89 19.22
Only animals of age classes 9 and 10 (Patton and Rogers, 1983). GSL, greatest length of skull; CIL, condylo-incisive length; BAL, basilar length of Hensel; ZYB, zygomatic breadth; IOC, interorbital constriction; ROL, rostral length; NAL, nasal length; ROW, rostral width, taken outside margins of nasolacrymal capsule; ROD, rostral depth; ORL, orbital length; DIL, diastemal length; MTL, maxillary toothrow length; IFL, incisive foramen length; PLA and PLB, palatal length A (and B) of Patton and Rogers (1983); PPL, post-palatal length; BUL, bullar length; OCB, occipital condyle breadth; MAB, mastoid breadth; MFL, mesopterygoid fossa length; MFW, mesopterygoid fossa width; CRD, cranial depth. Average SD, range, and sample size. Significance of differences between the French Guianan and the Rio Jurua samples: One-way ANOVA: ns; p > 0:05; **, p < 0:01; ***, p < 0:001.
L. Orlando et al. / Molecular Phylogenetics and Evolution 27 (2003) 113–120
TAL EAR HBL GSL CIL BAL ZYB LOC ROL NAL ROW ROD ORL DIL MTL IFL PLA PLB PPL BUL OCB MAB MFL MFW CRD
Rio Jurua
L. Orlando et al. / Molecular Phylogenetics and Evolution 27 (2003) 113–120
Fig. 3. Plot on the two discriminant axes (DA-1, associate eigen value k ¼ 9:32 for 95% of total variability and DA-2, associate eigen value k ¼ 0:39 for 5% of total variability) for the holotype and M. hispidus individuals sampled in Amazonia and French Guiana. The 15 variables available for the holotype (GSL, CIL, BAL, ROL, NAL, ROW, ROD, ORL, DIL, MTL, IFL, PLA, PLB, OCB, MAB) have been integrated in this analysis. Phylogenetic affinities based on cytochrome b sequences are summarized above.
molecular clades that we have uncovered (Figs. 1 and 2). The average genetic divergence (K2P for 798 bp of cytochrome b) is 4.8% (range 3.6–5.1%) between the western Amazonian clade C and French Guiana, a value comparable to what is measured among geographic samples of other rodent species (e.g., Oryzomys macconnelli, Makalata macrura) in Amazonia (Patton et al., 2000). Despite size differences (see below), we thus regard all Mesomys from clade C (Rio Juru a, Rio Urucu, Rio Jau in western and central Brazil), as well as other individuals from clade F (from Mato Grosso, Brazil, and Santa Cruz, Bolivia), to be M. hispidus, following Patton et al. (2000). The species status of the remaining clades (A, B, D, and E), however, is more difficult. These are even further differentiated (7.2–7.7%), values nearly as great as that separating M. hispidus from M. stimulax (8.2%; Fig. 2). However, all specimens from clades A and C so far examined (Patton et al., 2000) have the same karyotype (2N ¼ 60, FN ¼ 116), in sharp contrast with that of the sympatric M. occultus (2n ¼ 42, FN ¼ 54). Thus, despite: (i) the degree of sequence divergence, (ii) the sharp contrast between members of clades A and C in the mid Rio Juru a in morphology as well as in sequence divergence, and (iii) the size differentiation across the entire sampled range, we provisionally regard all seven clades (A through F in Figs. 1 and 2) as members of the single species, M. hispidus. We caution, however, that
119
this is a working hypothesis that should be subjected to further tests by additional data derived from more extended field collections and laboratory analyses, including broader geographic sampling, analyses of other genes, chromosomal banding analyses, and more thorough morphological analyses. The morphological results from cranio-dental characters clearly indicate that M. hispidus, as recognized here, is polytypic, with small animals in the Guianan Region (represented by our French Guiana samples; maxillary toothrow length [MTL] of 6:2 0:2 mm, range 5.9–6.7) and larger ones in western Amazonia (clade A: MTL of 7:3 0:2 mm, range 6.9–7.8) and central Amazonia (clade C: MTL of 7:1 0:3 mm, range 6.6–7.7). While size decreases from southwest to northeast, ‘‘large size’’ itself is apparently paraphyletic, as central Amazonian and French Guianan samples are sister-taxa in the mitochondrial trees (Fig. 2; bootstrap values of 66–77%). Of considerable interest is specimen LHE 968, a small-bodied individual from Guyana (West Kanuku Mountains; MTL 6.6 mm: Voss et al., 2001, table 44) that is the sister to a clade comprising the large (central Amazonian) and small (French Guianan) animals. Thus, size is a phyletically variable character and might even reflect the direct influence of nutrition and/or climate on individuals. Finally, we return to our original question: from where did the holotype of M. hispidus originate? This specimen (MNHN 1998–2075) has a MTL of 6.45 mm, a length that falls within the range of specimens from French Guiana but below that of specimens from central and western Amazonia (clades C and A, respectively). Thus, the holotype, most probably collected by the Portuguese naturalist Alexandre Rodrigues Ferreira during his 1783–1792 stay in the Brazilian Amazon (Hershkovitz, 1987), likely came from the state of Amapa, which is sandwiched between the mouth of the Amazon River and French Guiana. This is a region that he explored and whose mammalian faunal similarities lie with French Guiana (e.g., Serra do Navio, Voss et al., 2001). This hypothesis is fully supported by both the discriminant analysis of morphological characters and cytochrome b sequences, the latter which show the holotype to be deeply nested within, and closely similar (K2P distance of 0.0–0.3%) to the crown-group of specimens from French Guiana. Sequence from the holotype is more than an order of magnitude more divergent (>3.9%) in relation to individuals from all other sampled areas. We thus formally restrict the type locality of M. hispidus (Desmarest, 1817) to the state of Amapa, Brazil.
Acknowledgments Field work in French Guiana was funded by SILVOLAB & ECOFOR (PCER-XII ‘‘Rongeurs
120
L. Orlando et al. / Molecular Phylogenetics and Evolution 27 (2003) 113–120
arboricoles en for^et neotropicale: partage de lÕespace et des ressources’’). Laboratory work was funded by Mission pour la Creation du Parc (DIREN-Cayenne), MENRT (ACI), and CNRS (APN). The technical help of Elisabeth Eustaquio was much appreciated. The continuous help of Rob Voss encouraged us throughout this work. We are indebted to Vincent Laudet for critical comments on the manuscript. Appendix A. Morphology List of specimens analyzed: (1) Adults (age classes 9 and 10) from the Rio Jurua Basin: Clade 1 (upriver, labeled A in Fig. 125 in Patton et al., 2000): JLP-15367; JLP-15424; JLP-15431; JLP15445; JLP-15501; JLP-15624; JLP-15678; JLP-15708; JLP-15725; JLP-15853; MNFS-436; MNFS-517; MNFS-564; MNFS-569; MNFS-592; MNFS-593; MNFS-612; MNFS-745; MNFS-1519; MNFS-1531; MNFS-1615. Clade 2 (downriver, labeled C in Fig. 125 in Patton et al., 2000): JLP-16048; JLP-16066; JUR-321; JUR332; JUR-334; JUR-335; JUR-337; JUR-369; JUR-370; JUR-414; JUR-453; JUR-457; JUR-461; JUR-462; JUR-500; JUR-533; JUR-540; MNFS-32; MNFS-200. (2) Adults from French Guiana: Arataye: V-924; Cayenne: V-1245, V-1295, V-1577; Nouragues: V-953, V-1047, V-1048, V-1049; Paracou: AMNH-266596; Petit-Saut: MNHN-1995-215; Sa€ ul: V1118. (3) Holotype housed at Paris: MNHN-1998–2075. Appendix B. Molecular (cytochrome b sequences) M. hispidus: Bolivia: La Paz (LHE 748); Pando (USNM 579619); Santa Cruz (LHE-1554, MCT539 ¼ LHE-1189); Brazil: Acre: Rio Jurua (MNFS1230, INPA-2956, INPA-2957, INPA-2959, INPA-2961, INPA-2963); Amazonas: Rio Jurua (MVZ-190641, MVZ-190644, MVZ-190645, MVZ-190647, JUR-453, INPA-2953, MNFS-432, INPA-2973, INPA-2966, INPA-2974, INPA-2969, INPA-2975, MNFS-568, INPA-2976); Rio Jau (JLP-16735, MNFS-2016, MNFS2055, MNFS-2028, MNFS-2087); Rio Urucu (MNFS32, MNFS-188, MNFS-119, MNFS-189, MNFS-149, MNFS-200, MNFS-157); Pico da Neblina (INPA-2535); Mato Grosso (LPC-574); French Guiana: Arataye (V924); Cayenne (V-1245, V-1295, V-1577); Nouragues (V953, V-1047, V-1048, V-1049); Sa€ ul (V-1118); Guyana: West Kanuku Mountains (LHE-968); Peru´: Cusco (LHE-1455); Loreto (MV-970002, MUSM-13306); Madre de Dios (VPT-727); not recorded (MUSM12135, MUSM-12136); Venezuela: Amazonas, Neblina base camp (ALG-14162).
M. occultus: Brazil: Amazonas (INPA-2890, MNFS201). M. stimulax: Brazil: Para (CJ-7; LHE-572, MDC550).
References Allen, J.A., 1901. A preliminary study of the North American opossums of the genus Didelphis. Bull. Am. Mus. Nat. Hist. 14, 149–195. Arctander, P., 1995. Comparison of a mitochondrial gene and a corresponding nuclear pseudogene. Proc. R. Soc., Lond. 262, 13– 19. Emmons, L.H., 1993. On the identity of Echimys didelphoides Desmarest, 1817 (Mammalia: Rodentia: Echimyidae). Proc. Biol. Soc. Wash. 106, 1–4. Hershkovitz, P., 1987. A history of the recent mammalogy of the Neotropical region from 1492 to 1850. Fieldiana Zool. 39, 1–11. Kerr, R., 1792. The animal kingdom, 1-193. J. Murray, London, 644 pp. Linnaeus, C., 1758. Systema naturae per regna tria naturae, secundum classes, ordines, genera, species, cum characteribus, differentiis, synonymis, locis. 10th ed., tomus 1. Holmiae (Stockholm): Laurentii Salvii. Orlando, L., Bonjean, D., Bocherens, H., Thenot, A., Argant, A., Otte, M., H€anni, C., 2002. Ancient DNA and the population genetics of cave bears (Ursus spelaeus) through space and time. Mol. Biol. Evol. in press. Patton, J.L., Rogers, M.A., 1983. Systematic implications of nongeographic variation in the spiny rat genus Proechimys (Echimyidae). Z. S€augetierk. 48, 363–370. Patton, J.L., da Silva, M.N.F., Malcolm, J.R., 1994. Gene genealogy and differentiation among arboreal spiny rats (Rodentia: Echimyidae) of the Amazon Basin: a test of the riverine barrier hypothesis. Evolution 48, 1314–1323. Patton, J.L., da Silva, M.N.F., Malcolm, J.R., 2000. Mammals of the Rio Jurua and the evolutionary and ecological diversification of Amazonia. Bull. Am. Mus. Nat. Hist. 244, 1–306. Philippe, H., 1993. MUST, a computer package for Management Utilities for Sequences and Trees. Nucleic Acids Res. 21, 5264– 5272. Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Steiner, C., Sourrouille, P., Catzeflis, F., 2000. Molecular characterization and mitochondrial sequence variation in two sympatric species of Proechimys (Rodentia: Echimyidae) in French Guiana. Biochem. Syst. Ecol. 28, 963–973. Tate, G.H.H., 1939. Mammals of the Guiana region. Bull. Am. Mus. Nat. Hist. 76, 151–229. Thomas, O., 1911. The mammals of the tenth edition of Linnaeus; an attempt to fix the types of the genera and the exact bases and localities of the species. Proc. Zool. Soc. Lond. 1910, 120–158. Voss, R.S., Lunde, D.P., Simmons, N.B., 2001. The mammals of Paracou, French Guiana: a Neotropical lowland rainforest fauna. Part 2: nonvolant species. Bull. Am. Mus. Nat. Hist. 263, 1–236. Woods, C.A., 1993. Suborder Hystricognathi. In: Wilson, D.E., Reeder, D.M. (Eds.), Mammal Species of the World. Smithsonian Institution Press, Washington, DC, pp. 771–806. Yang, Z., 1994. Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: approximate methods. J. Mol. Evol. 39, 306–314. Zhang, D.-X., Hewitt, G.M., 1996. Nuclear integrations: challenges for mitochondrial DNA markers. Trends Ecol. Evol. 11, 247–251.