AMERICAN MUSEUM

Novitautes PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2817, pp. 1-14, figs. 1-4, tables 1-6 June 11, 1985 A Review of the Genus Isothrix (Rodentia, Echimyidae)

JAMES L. PATTON' AND LOUISE H. EMMONS2

ABSTRACT Two species of the echimyid rodent genus Iso- geographic races (bistriata [including villosa De- thrix are recognized within the Amazon and upper ville, negrensis Thomas, molliae Thomas, and bo- Orinoco basins of South America based on color liviensis Petter and Cuenca Aguirre] and orinoci and color pattern, bacular morphology, cranial Thomas), and I. pagurus Wagner. No localities of morphometrics, and karyotype. These represent sympatric contact between these two species are significant departures in the previously recognized known, although their ranges come very close in name combinations: I. bistriata Wagner, with two the general vicinity of Manaus in central Brazil.

INTRODUCTION The genus Isothrix contains arboreal rats transverse plates which persist into advanced ofmoderate body size that are a well-marked age. component of the morphologically and eco- Despite the relative distinctness ofIsothrix logically diverse neotropical rodent family as a generic unit, however, no clear under- Echimyidae. Erected by Wagner in 1845 to standing of its component taxa is currently accommodate three species, it is character- available. Specimens ofthis genus are rare in ized by a pelage lacking spines, a bushy tail, collections, a state which has resulted in the and teeth with complex, irregular re-entrant description ofgeographic units at the specific angles and enamel folds. The combination of or subspecific levels without benefit ofproper soft-haired pelage and bushy tail differen- comparisons to permit a realistic assessment tiates Isothrix from other echimyid rodents ofindividual relative to geographic variation. that have similar occlusal patterns (for ex- The present paper was motivated by the ex- ample, Echimys Cuvier [1809] and AMesomys istence of large series of Isothrix from three Wagner [1845]). Diplomys Thomas (1916a), general areas of the Amazon Basin and ad- also a soft-haired form without spines, differs jacent Venezuela, mostly in the collections of from Isothrix by its rather simple teeth with the American Museum of Natural History.

I Research Associate, Department of Mammalogy, American Museum of Natural History; Curator of Mammals, Museum of Vertebrate Zoology, University of California, Berkeley, California 94720. 2 Research Associate, Division ofMammals, National Museum ofNatural History, Smithsonian Institution, Wash- ington, D.C. 20560.

Copyright © American Museum of Natural History 1985 ISSN 0003-0082 / Price $1.70 2 AMERICAN MUSEUM NOVITATES NO. 2817

These allow us to examine for the first time bistriata negrensis Thomas, 1920 both individual variation and geographic Isothrix villosa villosa (Deville, 1852) trends, and thus to make a better assessment villosa molliae Thomas, 1924 of the taxonomic diversity within the genus. Isothrix pictus (Pictet, 1843) TAXONOMIC HISTORY OF This view has been accepted even though ISOTHRIX Oldfield Thomas suggested that several of Nine names have been proposed as taxa these named forms most likely represented within the genus Isothrix. Wagner (1845), in the same taxon. In 1928, for example, he the original definition ofthe genus, described stated that L v. villosa and L v. molliae from three species: bistriata, with type locality Rio northern Peru were the same entity, and both Guapore, Mato Grosso, Brazil; pachyura, type in 1916 and 1928 he suggested that L villosa locality not identified (but subsequently des- might in fact grade into L bistriata and thus ignated as Cuiaba, Mato Grosso, Brazil by would represent only an upper Amazon sub- Cabrera [1961]); and pagurus, type locality species of the latter. Borba, Rio Madeira, Amazonas, Brazil. In 1848 Wagner renamed pachyurus as crassi- METHODS AND ABBREVIATIONS cauda without reason. A fourth species, vil- losa, was proposed by Deville (1852), based Ninety-six specimens of Isothrix from on Castlenau's specimen from Sarayacu Mis- eastern Peru, eastern Bolivia, southern Ven- sion, Rio Ucayali (not Rio Urubamba as giv- ezuela, and central Brazil were examined in en in Cabrera [1961]), Loreto, Peru. Subse- this study. Eighty-four of these were consid- quently, Oldfield Thomas described three ered adults by the criteria given below. Stan- forms as subspecies ofeither bistriata (orinoci dard external measurements were missing [Thomas, 1899] and negrensis [Thomas, from the labels of most specimens, so anal- 1920]) or villosa (molliae [Thomas, 1924]). yses of individual and geographic variation Finally, the name boliviensis was recently were based on external color and color pat- proposed as a Bolivian subspecies ofbistriata tern and morphometric variables of the cra- by Petter and Cuenca Aguirre (1982). nium. For the cranium, the following 19 mea- The named form pictus (Pictet, 1843), a surements were taken from each skull with soft-haired echimyid from coastal Bahia, dial calipers reading to 0.05 mm: (1) greatest eastern Brazil, has been varyingly placed in length of skull [GSL]; (2) basilar length of the genus Isothrix (for example, Waterhouse, Hensel [BaL]; (3) zygomatic breadth [ZB]; (4) 1848; Ellerman, 1940; Cabrera, 1961), in a mastoid breadth [MB]; (5) least interorbital monotypic genus Nelomys (Pictet, 1843; constriction [IOC]; (6) rostral length [RL]; (7) Goldman, 1916; Thomas, 1916a), or in Echi- nasal length [NL]; (8) rostral width [RB]; (9) mys (Tate, 1935; Moojen, 1952). This dif- rostral depth [RD]; (10) diastema length [D]; ference of opinion results from pictus com- (11) palatal length A [PLA]; (12) palatal length bining simple laminated upper cheek teeth B [PLB]; (13) maxillary tooth row length (characteristic ofDiplomys) with the complex [MTRL]; (14) maxillary breadth [MaxB]; (15) lower teeth of Isothrix and Echimys. We fol- incisive foramen length [IFL]; (16) bullar low Cabrera (1961) and Honaki, Kinman, and length [BuL]; (17) post palatal length [PPL]; Koeppl (1982) in placing pictus within Iso- (18) cranial depth [CD]; and (19) mesopter- thrix, but do not consider it further. ygoid fossa width [MPFW]. Each of these Most recent authors (for example, Honaki, measurements is described and figured by Kinman, and Koeppl, 1982; Petter and Cuen- Patton and Rogers (1983) for the related ge- ca Aguirre, 1982) have followed the taxo- nus Proechimys. nomic arrangement given by Cabrera (1961) Variation in- bacular dimensions was ex- in amined from cleared and stained glandes, recognizing three species, as follows: prepared as in Hooper (1959). Specimens were Isothrix bistriata bistriata Wagner, 1845 aged according to nine categories of tooth bistriata pachyura Wagner, 1845 eruption and wear patterns: (I) dP4 erupted, bistriata pagurus Wagner, 1845 M' erupted but below occlusal level; (II) MI bistriata orinoci Thomas, 1899 to occlusal level; M2 erupting; (III) M3 erupt- 1985 PATTON AND EMMONS: ISOTHRIX 3

FIG. 1. Sampled localities of Isothrix specimens examined in this report; numbered localities are as identified in the text. The type localities for all named forms are indicated by asterisks.

ing; (IV) median re-entrant fold confluent with degree of character difference between the primary fold on dP4 to M3; (V) median fold named forms recognized by Cabrera (1961) separate on dP4, open on MI through M3; (VI) and others. Finally, overall geographic trends median fold open on M2 through M3; (VII) and degrees of differentiation between local- median fold open only on M3; (VIII) median ities were examined by the multivariate pro- fold separate from primary fold on all teeth; cedure ofdiscriminant function analysis. The and (IX) entire tooth row worn, all folds iso- statistical routines of SAS (Barr et al., 1982) lated. Age classes V through IX were consid- and SPSS (Nie et al., 1975) were used for all ered to represent adult individuals, although analyses. Samples from each locality were no data are available on individual repro- considered as separate entities in the exam- ductive status in each ofthe various age classes ination of both univariate and multivariate recognized. geographic trends. The degree ofnon-geographic variation due Karyotypes from two individuals, one from both to age and sex was assessed by two-way northern Peru and one from central Brazil, analysis of variance (ANOVA) following the were prepared following the technique ofPat- suggestion ofLeamy (1983). One-way ANO- ton (1967). VAs were utilized to assess the significance Specimens forming the basis of this report of single character variation within a taxon are in the collections of the American Mu- over a geographic area and to examine the seum ofNatural History, New York (AMNH), 4 AMERICAN MUSEUM NOVITATES NO. 2817

TABLE 1 Amorin (5, AMNH), Rio Tapajoz, Limoal Non-geographic Variation for Cranial Variables (1, AMNH). in a Combined Sample of Isothrix from Northern Locality 12-BRAZIL: Amazonas; 80 km Peru (Localities 31 to 34)a N (by rd) Manaus (1, USNM). Variableb Age Sex Age x Sex Residual Locality 21-BOLIVIA: Beni; Rio Itenez at Costa Marques (1, AMNH). GSL nsc (24.6) ns (7.9) ns (0.0) (67.5) Locality 31-PERU: Loreto; Pto. Indiana, BaL ns (22.6) ns(ll.5) ns (0.0) (65.9) Rio Amazonas (25, AMNH). ZB *d (15.9) * (13.5) ns (11.0) (59.6) MB ns (0.0) * (7.2) * (33.6) (59.2) Locality 32-PERU: Loreto; Orosa, Rio IOC * (28.3) ns (4.2) ns (0.0) (67.5) Amazonas (5, AMNH). RL * (28.0) ns (1.2) ns (0.0) (70.8) Locality 33 -PERU: Loreto; Sarayacu, Rio NL ns (16.7) ns (4.1) ns (0.0) (82.2) Ucayali (14, AMNH). RW ns (11.8) ns (6.0) ns (0.0) (82.2) Locality 34-PERU: Loreto; Boca Rio RD ns (1.2) ns (0.0) ns (0.0) (98.8) Curaray (1, AMNH). D ns (15.0) ns (9.6) ns (0.0) (75.4) Locality 35-PERU: Loreto; Rio Itaya near PLA ns (12.2) ns (3.9) ns (0.0) (83.9) Iquitos (1, AMNH). PLB ns (6.5) ns (0.0) ns (0.0) (93.5) Locality 36 -PERU: Ucayali; Lagarto, Rio MTRL ns (0.0) ns (0.0) ns (8.9) (91.1) Boca Rio Urubam- MaxB ns (0.0) ns (0.0) ns (22.0) (78.0) Alto Ucayali (5, AMNH), IFL ns (10.9) ns(l3.7) ns (0.0) (75.4) ba (1, AMNH). BuL ns (0.3) ns (14.6) ns (0.0) (85.1) Locality 37-PERU: Amazonas; La Poza, PPL ns (5.1) ns (15.1) ns (0.0) (79.8) Rio Santiago (3, MVZ). CD ns (8.0) ns (9.9) ns (0.0) (82.1) Locality 38-PERU: Pasco; Nevati Mis- MPFW ns (7.5) * (25.1) ns (0.0) (67.4) sion (1, AMNH). Mean (11.29) (7.76) (3.97) (76.98) Locality 41-VENEZUELA: Amazonas; Boca del Rio Ocampo, Rio Orinoco (4, a The significance level due to age, sex, and the inter- AMNH). action between age and sex is given for each variable Locality 42-VENEZUELA: Amazonas; based on a two-way ANOVA; the proportion ofvariation attributed to each component as well as the residual Rio Casiquiare, Orilla Izquierda El Merey (2, variation is given in parentheses. AMNH), Casiquiare Canal, Capibara (1, b See text for full description of each variable. USNM). c Non-significant. Locality 43-VENEZUELA: Amazonas; d p < 0.05. Rio Orinoco, Cerro Duida (6, AMNH). Locality 44-VENEZUELA: Amazonas; Esmeralda, Mt. Duida, 335 m (10, AMNH), the Museum of Vertebrate Zoology, Univer- 68 km SE Esmeralda, Boca Mauaca (4, sity of California, Berkeley (MVZ), and the USNM). National Museum ofNatural History, Smith- sonian Institution, Washington, D.C. RESULTS OF ANALYSES (USNM). In addition, three holotypes in the British Museum (Nat. Hist.), London, were UNIVARIATE CRANIAL ANALYSES: Non-geo- examined: orinoci Thomas (BM[NH] graphic variation was examined for samples 99.9.11.45), negrensis Thomas (BM[NH] from the combined region of northern Peru 20.7.1.20), and molliae Thomas (BM[NH] (localities 31-34) for which the largest sam- 24.2.22.25). ples were available (21 males, 19 females). Figure 1 provides the mapped localities for Two-way ANOVA assessed the level of sig- material examined in this report; a two-digit nificance of age, sex, and age-sex interaction locality code is employed to signify both for each individual cranial variable, and the country and specific locality. All type local- VARCOMP procedure of SAS apportioned ities are indicated. Localities and their des- the total variation ofeach character into age, ignations are as follows: sex, interaction, and residual components. As Locality 11 -BRAZIL: Para; Rio Tapajoz, is evident in table 1, there is only minimally Igarape Brabo (2, AMNH), Rio Tapajoz, Ina- significant expressed variation due to speci- jatuba (8, AMNH), Rio Tapajoz, Igarape men age, sex, or the interaction of these two 1985 PATTON AND EMMONS: ISOTHRIX 5

variables; only three variables show either TABLE 2 age or sexual variation and only one variable Univariate Levels of Geographic Variation within shows a significant interaction, all at the 0.05 Samples of Isothrix, Based on One-way Analysis level. The low level of non-geographic vari- of Variance ation is clearly evident from the variance Between component estimations. Less than 12 percent Peru- ofthe total variance averaged across all char- Between Bolivia, acters is attributable to age, less than 8 per- Peru- Vene- cent to sexual and less Within Within Bolivia zuela, and dimorphism, than 4 Peru- Vene- and Vene- Central percent to their interaction. Fully 77 percent Variablea Bolivia zuela zuela Brazil of the variation is residual, or due to other ***e factors. Therefore, there is statistical justifi- GSL nsb ns ns cation for combining all adult individuals BaL ns * ns ZB *c ns within localities in subsequent * analyses. MB ns ns Univariate geographic trends were as- IOc ns ns sessed by one-way ANOVAs (1) within the RL ns ns samples from Peru-Bolivia and Venezuela, NL ns *n respectively; (2) between the samples from RW ns ns ns these two geographic regions; and (3) between RD ns *

D ns ns the available samples from all areas. The ns ** PLA ns ns rationale for this "hierarchical" procedure is ns as follows: The first level permits a test ofthe PLB ns ns MTRL ns ns amount ofvariation within limited geograph- ns** MaxB ns ns ic regions and, ns thereby, examines the degree IFL ns of * regional homogeneity in the available sam- BuL ns ns ples. This is desirable because of the small PPL ns ns sample sizes. The second analysis directly as- CD ns ns **

sesses the degree of differentiation between MPFW ns ns ns samples traditionally allocated to the sepa- a See text for full description of variables. rate species, I. villosa and L bistriata, but for b ns = non-significant. which Thomas (1916b, 1928) suggested pos- c * = p < 0.05. sible conspecificity. Finally, the third analysis d** = p < 0.01. includes the sample from central Brazil (lo- e *** = p < 0.001. cality 1 1), specimens ofwhich are clearly dis- tinct in pelage characters in comparison with villosa and bistriata (see below). Table 2 summarizes the level of character is somewhat greater relative to that within differentiation in each of the above three each geographic region. Eight of the 19 vari- analyses. As is evident, character differentia- ables show significant differences between tion is minimal to non-existent within the these two regions, five at the 0.05 level, two combined localities of Peru and Bolivia (lo- at the 0.01 level, and one at the 0.001 level calities 21 and 31 to 38) or Venezuela (lo- (table 2). Of the characters which do show calities 41-44). Only three of 19 and five of significant geographic variation, only two of 19 variables, respectively, show significant these exhibit any apparent geographic pat- geographic variation within each region, and tern. The Venezuelan samples consistently then only at p < 0.05. Thus, the combined have both the longest nasals (character NL) geographic regions are virtually homoge- and shortest palates (character PLB) relative neous within themselves, with little or no to Peruvian samples, with a sizeable gap be- geographic character variation exhibited. tween them (table 3). In the comparison of Peruvian-Bolivian Finally, when the comparison incorporates samples (mostly assignable on present all available samples from the three geo- grounds to L villosa) and Venezuelan L bi- graphic regions of Peru-Bolivia, Venezuela, striata, the level of character differentiation and central Brazil, all characters show highly 6 AMERICAN MUSEUM NOVITATES NO. 2817

TABLE 3 TABLE 4 Minimally Non-significant Geographic Subsets of Standardized Canonical Discriminant Functions Isothrix Population Samples from Peru-Bolivia for Geographic Samples of Isothrix and Venezuela for Those Two Variables that Show Geographic Patternsa Variablea DFI DFII DFIII -1.477 -2.516 Char- Char- GSL -0.285 acter Mean Locality acter Mean Locality BaL -0.636 2.085 1.487 ZB 0.156 0.057 -0.720 NL 17.50 43 I PLB 8.72 32 I MB 0.457 -0.084 0.163 17.20 41 II 8.52 31 III IOC -0.087 0.381 -0.590 16.98 44 II 8.25 37 IIII RL 2.111 -1.183 -3.790 16.95 42 II 8.18 33 IIII NL -3.684 4.393 1.427 16.35 33 III 8.14 36 IIII RW -0.031 -0.006 -0.225 16.32 31 III 7.54 43 IIII RD 0.604 -0.782 1.775 16.22 32 III 7.36 44 IIII D 1.219 -0.244 0.167 15.53 36 III 7.05 42 II PLA 1.537 -1.565 1.113 14.60 37 I 6.93 41 I PLB -0.011 -4.494 0.213 a Based on Duncan's multiple range test with a = 0.05. MTRL 0.764 -1.092 0.219 MaxB 0.003 -0.208 -0.494 IFL -0.056 -0.203 -0.561 BuL 0.419 -0.254 1.380 significant differentiation (18 of 19 at p < PPL 0.414 0.181 -0.899 0.001; table 2), with central Brazilian speci- CD 0.436 0.319 1.724 mens smaller in 16 of the 19 measurements. MPFW -1.215 0.520 0.450 Clearly, therefore, on a univariate level, there Eigenvalue 5.984 2.414 1.093 is virtually no differentiation among locali- % variance ties within each ofthe three regional samples; explained 49.04 20.11 8.95 only relatively slight differentiation is present a See text for full description of variables. in the comparison between Peru-Bolivia and Venezuela samples of L villosa and L bi- striata, respectively, while the central Bra- zilian animals are sharply differentiated from to most ofthe differentiation along DFI while all others. NL and palatal length (PLB) contribute to MULTIVARIATE CRANIAL ANALYSIS: A step- discrimination along the second axis (table wise discriminant function (DF) analysis was 4). Thus, this multivariate view is consistent performed to determine the degree of multi- with the results of the univariate analyses. variate separation among samples within and Figure 2 presents a plot ofthe discriminant between each geographic region. Analyses scores for individuals from each separate lo- were performed considering each locality as cality. Three groups of localities are visually a separate a priori group and with localities apparent, with the sample from the Rio Ta- combined into the three major geographic pajoz in central Brazil (locality 11) most dif- units as a priori groups. The results of these ferentiated. The Venezuelan (localities 41- two analyses did not differ, and only the first 44) and Peruvian-Bolivian (localities 21 and will be presented here. 31-38) samples are less differentiated, al- Eighteen discriminant functions were ex- though they are non-overlapping along the tracted, but only the first three functions had first two axes. Individual samples within eigenvalues greater than 1.0. These three either ofthese two regions are virtually iden- functions account for slightly more than 78 tical. This visual perspective is supported by percent ofthe total variance in the data, with a multivariate analysis of variance compar- 49 percent attributable to DFI alone (table ing the Mahalanobis distances between group 4). Most of the total variance in the samples centroids (table 5). Only four of 31 two-way relates to both absolute and proportional size locality comparisons within the Peruvian- differences involving the rostrum and palate: Bolivian or Venezuelan geographic units show nasal and rostral length measures contribute significant F-values, whereas 20 of 32 com- 1985 PATTON AND EMMONS: ISOTHRIX 7

4.

3.

2-

I -

Peru - 9: 0. Bolivia

-1.

* s. n \ 11 ' * central N, s1%. -3- III., Brazil '. / -4.

5U %O 9 . . - -7 -6 -5 -4 -3 -2 -I 1 2 4 5 I FIG. 2. Plot ofthe first two discriminant functions for geographic samples ofIsothrix: circles represent localities from Peru and Bolivia; triangles those from Venezuela; and squares those from central Brazil. Centroids for each locality are indicated by their respective numerical designations (see text). The placement in discriminant space of the holotypes for three named forms is illustrated.

TABLE 5 Summary of Differentiation between Geographic Samples of IsothriXa Number of Number of Pair-wise Significant Geographic Region Being Compared Comparisons D2 + st dev Comparisons Within Peru-Bolivia 28 1.328 ± 0.463 3 @ p < 0.05 1 @p < 0.01 Within Venezuela 5 0.386 ± 0.157 0 Between Peru-Bolivia and Venezuela 32 3.012 ± 0.836 6 @ p < 0.05 6@p < 0.01 8 @ p < 0.001 Between Peru-Bolivia and central Brazil 8 6.385 ± 1.443 2 @ p < 0.01 6 @ p < 0.001 Between Venezuela and central Brazil 4 4.660 ± 0.870 4 @ p < 0.001 Between Peru-Bolivia-Venezuela and 12 5.810 ± 1.662 2 @ p < 0.01 central Brazil 10 @ p < 0.001 a Based on a multiple analysis of variance of sample pair-wise Mahalanobis distances (D2), with df5,1I. 8 AMERICAN MUSEUM NOVITATES NO. 2817

A

Perul -II E A Venezuela E 11- * BruziI -c CP G_ A C A C 7. 0 .0

5- UI I I 48 50 52 54 56 58 GSL (mm) FIG. 3. Representative bacula for Isothrix specimens from central Brazil (A = AMNH 94885; B = AMNH 95652; locality 11), Venezuela (C = AMNH 78114; locality 41), and Peru (D = AMNH 74072; locality 32). Bar = 5 mm. The inset graph illustrates the proportional difference in bacular length in relation to overall body size,I as indexed by greatest skull length (GSL), among the three geographic samples.

parisons between these two units are signif- in all specimens available, regardless oftaxon icantly different. All 12 pair-wise distances or locality: an elongate bony structure, with between the Peruvian-Bolivian-Venezuelan a rather bulbous proximal base tapering to a samples and that from central Brazil were narrow distal tip which, in specimens from found to be highly significantly different. Peru and Venezuela but not those from cen- Again, these multivariate results mirror the tral Brazil, has a moderate median notch (see patterns of character differentiation within fig. 3). A simple, shallow, triangular carti- and between geographic samples based on the laginous tip is present. In lateral view the univariate analyses. baculum is straight. The tip is rather ovoid BACULAR VARIATION: Cleared and stained in cross-section; a slight ventral groove de- glandes are available from a total of 12 spec- velops at midshaft and extends to near the imens representing the three geographic areas base. This general bacular type is character- herein considered: Peru (N = 5), Venezuela istic ofmost echimyid rodents, notably Echi- (N = 5), and central Brazil (N = 2). The bac- mys, Diplomys, Thrinacodus, and Dactylo- ulum of Isothrix has not been previously fig- mys, as well as several species of Proechimys ured, to our knowledge (see Didier, 1962, for (see Didier, 1962; Patton and Gardner, 1972; a synopsis of bacular variation in echimyid also, Patton, unpubl. data). rodents). General bacular form is the same Despite an overall similarity in bacular 1985 PATTON AND EMMONS: ISOTHRIX 9

TABLE 6 Measurements (in Millimeters) of Bacula in Geographic Samples of Isothrix (Mean plus or minus one standard deviation are listed for each measurement.) Geographic Region Age N Bacular Length Width at Tip Width at Base Central Brazil 8 2 5.83 ± 0.54 0.86 ± 0.14 1.14 ± 0.01 Peru 6 2 7.28 ± 0.76 1.28 ± 0.01 1.67 ± 0.12 7 1 7.95 1.04 1.51 8 2 9.56 ± 0.60 1.51 ± 0.10 2.26 ± 0.06 Venezuela 7 1 7.76 1.45 2.00 8 4 8.37 ± 0.67 1.43 ± 0.16 2.05 ± 0.30

shape, the baculum ofthe two specimens from The specimens from central Brazil, how- central Brazil is proportionally shorter than ever, show a radical departure in both color that of Peruvian or Venezuelan specimens, and pattern from those of Bolivia, Peru, or even when allowing for individual age dif- Venezuela. First, and most striking, these ferences (see fig. 3 and table 6). It is not pos- specimens lack even a hint of black supraor- sible to differentiate the bacula of Peruvian bital stripes, they have a reddish snout with and Venezuelan specimens on either absolute that color extending to above and below the or relative size (fig. 3). eyes, and, in adults, the tail grades uniformly COLOR AND COLOR PATTERN: Only two from gray-brown at the base to yellow-brown geographic entities are evident on the basis toward the tip. Tail color ofjuveniles varies of sharply distinguishable color patterns. All from uniformly gray-brown to dark brown. specimens from Venezuela, Peru, and Bolivia The tail thus not only lacks any black hairs, are characterized by a grizzled, yellowish gray but shows the opposite color pattern, being head with a distinct pair of black or dark lighter at the tip than the base in adults. The brown supraorbital stripes extending onto the dorsal color ranges from grizzled yellow or nape ofthe neck, and a pale creamy patch on gray-brown from the forehead to midback the forehead between these stripes. The dor- with the distal sides and rump distinctly rus- sum is rather uniformly grizzled yellow- set. The venter is as in Peruvian-Venezuelan brown, often with a dark middorsal patch. specimens, but generally paler. The venter is uniformly pale yellow to buff, KARYOTYPES: Chromosome characters are with basal gray showing in many specimens. available from only two individuals, one Pe- The tail is varyingly colored, almost always ruvian specimen referable to I. villosa and pale auburn to yellow at the base but dark- one specimen of the central Brazilian form. ening to black at the tip. In only one specimen The karyotype of the latter is illustrated in does the distal portion of the tail lack black; figure 4. The diploid number is 22, the num- in all others the black portion encompasses ber of autosomal arms 38. There are eight at least the terminal one-fourth of the tail, autosomal pairs of meta and submetacen- usually one-half or more (72 of 76 speci- trics, the fourth largest ofwhich shows a dis- mens). It is not possible to separate speci- tinct secondary constriction on the long arm, mens from Peru or Bolivia from those from and one autosomal pair each of subtelocen- Venezuela on the basis ofany consistent color trics and acrocentrics. The X-chromosome is feature. Thomas (1920) distinguished ne- a medium-sized submetacentric, and the Y grensis primarily on the basis of black head is a small acrocentric. Differential staining for stripes that fuse on the nape of the neck to constitutive heterochromatin (C-bands; see form a broad black band. Such fusion, how- inset, fig. 4) reveals only two pairs of auto- ever, was seen in 14 percent ofPeruvian spec- somes with slight centromeric heterochro- imens (8 of 57 examined) and in 46 percent matin, whereas eight pairs exhibit telomeric ofthose from Venezuela (12 of25 examined). heterochromatin on one (four pairs) or two 10 AMERICAN MUSEUM NOVITATES NO. 2817

a

x Y

FIG. 4. The karyotype of Isothrixpagurus from 80 km N Manaus, Amazonas, Brazil (2n = 22, FN = 38; USNM 555639). Inset: a mitotic metaphase cell differentially stained for constitutive heterochro- matin.

arms (four pairs). The short arm of the examined, however, only 15 specimens, rep- X-chromosome appears totally heterochro- resenting eight named forms. Ofthese, L pic- matic. tus (Pictet) is rather easily distinguished by The chromosome preparations from the virtue ofthe simple and complete transverse single Peruvian specimen are quite poor. plates ofthe upper teeth, an observation made Nevertheless, diploid number counts range by all previous authors (for example, Water- from 54 to 58 and most ofthe chromosomes house, 1848; Thomas, 1916a, 1916b; Eller- appear biarmed. These two karyotypes are man, 1940). However, despite the discussion quite distinct, equivalent to the differences ofdetails differentiating the teeth ofthe named seen among the most divergent species of forms applied to the species L villosa and L Proechimys (Patton and Gardner, 1972; Reig, bistriata, Petter and Cuenca Aguirre were un- 1981), or even among other genera of echi- able to illustrate any dental features which myid rodents (for example, Yonenaga, 1975; can serve to separate consistently these taxa. Souza and Yonenaga-Yassuda, 1982). Indeed, in the present analysis, the complete COUNTERFOLD PATTERN OF THE CHEEK range of re-entrant fold number, details of TEETH: In the most recent taxonomic study position, and lateral versus lingual initial clo- ofIsothrix, Petter and Cuenca Aguirre (1982) sure with wear can be observed among in- focused solely on the number and pattern of dividuals within single localities. We, there- re-entrant folds on both upper and lower fore, find no way to differentiate regional units cheek teeth for taxonomic conclusions. They on the basis of counterfold pattern, with the 1985 PATTON AND EMMONS: ISOTHRIX 11 exception of I. pictus. Certainly, the central (Deville), despite Thomas's (1916b, 1928) Brazilian specimens, so distinctive in other suggestion that these two taxa are likely con- characters, cannot be separated from those specific. In color pattern of the head, body, of Peru or Venezuela, nor can those from and tail, and in tooth and bacular characters, Peru or Venezuela be differentiated from each samples applicable to these two named forms other. are indistinguishable, although they are sep- MORPHOLOGICAL AND BEHAVIORAL NOTE: arable on the basis ofa multivariate analysis The single specimen collected by one of us ofcranial dimensions. Considering the over- north of Manaus in central Brazil was ob- all high degree ofsimilarity among these sam- served for a day in captivity. A striking fea- ples and their extreme level ofdifference from ture of its behavior when at rest was a ten- central Brazilian populations in the genus, it dency to coil the tail tip in a tight downward seems best to consider as members of the spiral ofone and a halfloops. When climbing, single species, L bistriata Wagner, all Vene- it sometimes wrapped its tail longitudinally zuelan, Peruvian, Bolivian, and Brazilian around the branch. These behaviors indicate specimens ofIsothrix with black stripes above a semiprehensile tail, a feature hitherto un- the eyes. The geographic homogeneity within described in echimyid rodents, and not pres- the regional samples from Peru and Bolivia ent in Mesomys or Dactylomys, two other on the one hand and Venezuela on the other, arboreal genera of this family that we have in combination with the degree ofcranial dis- observed in the field. tinctness of specimens from these two areas, argues for recognition oftwo subspecific units within L bistriata. Since the type locality of TAXONOMIC CONCLUSIONS bistriata is quite close to the Bolivian border, The present analyses provide totally con- the Bolivian and Peruvian samples examined cordant results that clearly distinguish central here are best referred to the nominant sub- Brazilian Isothrix from those elsewhere in the species, L b. bistriata Wagner. Thus, the Amazon region of Venezuela, Peru, and Bo- names villosa (Deville) and molliae Thomas, livia. This differentiation is trenchant, being with type localities in Peru, become junior based on cranial morphometric and bacular synonyms. Clearly, the holotype of molliae characters, color and color pattern, and falls within the morphological range exhib- karyotype with available specimens. There ited by these geographic samples (see fig. 2). seems little doubt but that these respective Although we have not examined the type of samples represent distinct species. The ques- boliviensis Petter and Cuenca Aguirre (1982), tion remains, therefore, regarding the proper we consider it likewise to be a synonym. The allocation of names to these two taxa. only name available for the Venezuelan sam- Although Wagner (1845) did not designate ples is orinoci Thomas, so these specimens a type species for his genus Isothrix, Gold- are referred to the subspecies L b. orinoci man (1916) selected L bistriata, the first of Thomas. The named form negrensis Thomas Wagner's species, as the type for the genus. (1920), with type locality at Acajatuba on the The type locality for this taxon is Rio Gua- lower Rio Negro, near Manaus, Amazonas, pore, Mato Grosso, Brazil. Wagner's original Brazil, remains somewhat problematical. This description clearly ascribes to the holotype form is clearly referable to bistriata on the the color and pattern characteristic of all basis ofcolor pattern and cranial features (see Venezuelan, Peruvian, and Bolivian speci- Thomas, 1920 and fig. 2). The DF analysis mens examined here. Most notably, the type places the type, and only examined specimen, is characterized by a pair of black stripes on within the bistriata samples from Peru and the head and a black tail with a yellow base Bolivia rather than with Venezuelan orinoci. (Wagner, 1845, p. 146). It is, therefore, quite We, therefore, subsume negrensis within the clear that the name bistriata should apply to subspecies bistriata until adequate samples animals with the black head stripes. Tradi- become available for analysis. With better tionally, however, there have been two rec- and geographically intermediate samples, the ognized forms having this general color pat- current distinction between bistriata and ori- tern on the head: bistriata and Peruvian villosa noci may in fact disappear into a more 12 AMERICAN MUSEUM NOVITATES NO. 2817

smoothly distributed clinal relationship. Cer- Isothrix pagurus Wagner tainly, the nearly intermediate position ofthe holotype of negrensis in discriminant space Isothrix pagurus Wagner, 1845. relative to samples of bistriata and orinoci RANGE: Amazon Basin of central Brazil, suggests such a possibility (fig. 2). from Rio Madeira east to the Rio Tapajoz The specimens from the Rio Tapajoz in and north to the lower Rio Negro (localities central Brazil examined here seem best re- 11 and 12; see map, fig. 1). Type locality: ferable to the species L pagurus Wagner. Al- Borba, lower Rio Madeira, Amazonas, Bra- though Wagner's original description of this zil. form (Wagner, 1845, p. 146) is lacking in detail, he does make reference to the fulvous Isothrix pictus (Pictet) snout and rust-colored hind parts, and makes no reference to black eye stripes. Unfortu- Nelomys pictus Pictet, 1843. nately, Wagner's specimen ofpagurus lacked RANGE: Coastal forests of eastern Brazil. a tail. The type locality of pagurus is given Type locality: Bahia, Brazil (probably col- as Borba, near the mouth ofthe Rio Madeira, lected in the region north ofthe municipality not far to the south of Manaus, where the of Ilheus [Fazenda Almada]; see Moojen, single karyotyped specimen is from, or to the 1952). west of the Rio Tapajoz (see fig. 1). ACKNOWLEDGMENTS GENUS ISOTHRIX WAGNER We thank Drs. Guy G. Musser, American Isothrix Wagner, 1845. Museum of Natural History, J. E. Hill, Brit- Nelomys Pictet, 1843 (part, not Cuvier, 1837). ish Museum (Nat. Hist.), and Charles 0. Lasiuromys Deville, 1852. Handley, Jr., National Museum of Natural GENOTYPE: Isothrix bistriata Wagner (des- History, Smithsonian Institution, for per- ignation by Goldman, 1916). mission to examine specimens in their care. Special appreciation is extended to Musser Isothrix bistriata bistriata Wagner for reviewing the manuscript, to Dr. Philip Isothrix bistriata Wagner, 1845. Myers for measuring the type specimens in Isothrix pachyura Wagner, 1845. the British Museum, and to Ms. Silva Raker Lasiuromys villosa (Deville, 1852). for aid in data analysis. Patton's fieldwork in Isothrix bistriata negrensis Thomas, 1920. Peru was supported by the National Science Isothrix villosa molliae Thomas, 1924. Foundation (BNS 76-17485, awarded jointly Isothrix bistriata boliviensis Petter and Cuenca with 0. Brent Berlin) and the Direccion Gen- Aguirre, 1982. eral Forestal y de Fauna of the Ministerio de RANGE: Western and southwestern Ama- Agricultura, Lima. Emmon's fieldwork in zon Basin from Mato Grosso in Brazil north- Brazil was supported by the Instituto Na- west through eastern Bolivia, eastern Peru, cional de Pesquizas da Amazonia (Manaus) and western Brazil (localities 21 and 31 and the World Wildlife Fund-US. through 38; see map, fig. 1). Type locality: Rio Guapore, Mato Grosso, Brazil. LITERATURE CITED Isothrix bistriata orinoci Thomas Barr, A. M., J. H. Goodnight, J. P. Sall, W. H. Blair, and D. M. Chilko Loncheres (Isothrix) bistriatus orinoci Thomas, 1982. SAS user's guide, 1982 ed. SAS Insti- 1899. tute, Raleigh, North Carolina, 921 pp. RANGE: Cabrera, Angel Upper Rio Orinoco, Rio Casi- 1961. Catalogo de los mamiferos de America quiare, and upper Rio Negro of Venezuela del Sur. Rev. Mus. Argentino Cien. Nat. and adjacent Colombia (localities 41 through "Bernardino Rivadavia," vol. 4, no. 2, 44; see map, fig. 1). Type locality: Maipures, i-xxii + pp. 310-732. upper Rio Orinoco, Intendencia Vichada, Cuvier, F. Colombia. 1809. Extr. des premiers memoires sur les 1985 PATTON AND EMMONS: ISOTHRIX 13

dents de mammiferes considerees Zool., Louisiana St. Univ., no. 44, 30 comme caracteres generiques. Nouv. pp., figs. 1-9. Bull. Scien. Soc. Philom., Paris, vol. 24, Patton, J. L., and Mary Anne Rogers pp. 393-395. 1983. Systematic implications of non-geo- 1837. Rapport sur un memoire de M. Jourdan graphic variation in the spiny rat genus sur quelques mammiferes nouveaux. Proechimys (Echimyidae). Z. Sauge- Acad. Scien. Nat., Paris, vol. 8, pp. 367- tierk., vol. 48, no. 6, pp. 363-370, figs. 374. 1-2. Deville, M. E. Petter, F., and H. Cuenca Aguirre 1852. Notice sur le genere Dactylomys, et es- 1982. Un Isothrix nouveau de Bolivie (Ron- tablissement d'un genre nouveau, le geurs, Echimyides). La denture des Iso- Lasiuromys, provenant de l'expedition thrix. Mammal., vol. 46, no.2, pp. 191- de M. de Castlenau. Rev. et Mag. Zool., 203, figs. 1-6. ser. 2, vol. 4, pp. 553-561, pls. 15-16. Pictet, F.-J. Didier, Robert 1843. Seconde notice sur les animaux nou- 1962. Note sur l'os penien de quelques ron- veau or peu connus du Musee de Ge- geurs de l'Amerique du Sud. Mammal., neve. Mem. Soc. Phys., Geneve, vol. 10, vol. 26, no. 3, pp. 408-430, figs. 1-24. pp. 201-214, pls. 1-5. Ellerman, J. R. Reig, 0. A. 1940. The families and genera of living ro- 1981. Modelos de especiacion cromosomica dents. London, British Museum (Nat. en las casiraguas (genero Proechimys) de Hist.), vol. 1, i-xxvi + 689 pp., figs. 1- Venezuela. In 0. A. Reig (ed.), Ecology 189. and genetics of animal speciation, Goldman, E. A. Equinoccio, Edit. Univ. Caracas, Simon 1916. Notes on the genera Isothrix Wagner Bolivar, pp. 149-190, figs. 1-5. and Phyllomys Lund. Proc. Biol. Soc. Souza, M. J., and Y. Yonenaga-Yassuda Wash., vol. 29, pp. 125-128. 1982. Chromosomal variability of sex chro- Honaki, James H., Kenneth E. Kinman, and James mosomes and NOR's in Trichomys W. Koeppl apereoides (Rodentia, Echimyidae). Cy- 1982. Mammal species of the world. Assoc. togenet. Cell Genet., vol. 33, no. 3, pp. Syst. Coll., Lawrence, Kansas, 694 pp. 197-203, figs. 1-5. Hooper, Emmet T. Tate, G. H. H. 1959. The glans penis in five genera ofcricetid 1935. The taxonomy of the genera of Neo- rodents. Occas. Papers Mus. Zool., Univ. tropical hystricoid rodents. Bull. Amer. Michigan, no. 613, 11 pp., pls. 1-5. Mus. Nat. Hist., vol. 68, pp. 295-447. Leamy, Larry Thomas, 0. 1983. Variance partitioning and effects of sex 1899. Descriptions of new rodents from the and age on morphometric traits in ran- Orinoco and Ecuador. Ann. Mag. Nat. dombred mice. Jour. Mammal., vol. 64, Hist., ser. 7, vol. 4, pp. 378-383. no. 1, pp. 55-61. 1916a. On the generic names Rattus and Phyl- Moojen, Joao lomys. Ibid., ser. 8, vol. 18, p. 240. 1952. Os reodores do Brasil. Biblio. Cien. 1916b. Some notes on Echimyinae. Ibid., ser. Brasileira, ser. A, II, Rio de Janeiro, 214 8, vol. 18, pp. 294-301. PP. 1920. On mammals from the lower Amazons Nie, N. H., C. H. Hull, J. G. Jenkins, K. Stein- in the Goeldi Museum, Para. Ibid., ser. brenner, and D. H. Bent 9, vol. 6, pp. 266-283. 1975. Statistical package for the social sci- 1924. On a collection of mammals made by ences, 2nd ed. New York, McGraw Hill Mr. Latham Rutter in the Peruvian An- Book Co., 675 pp. des. Ibid., ser. 9, vol. 8, pp. 530-538. Patton, James L. 1928. The Godman-Thomas expedition to 1967. Chromosome studies of certain pocket Peru. VIII. On mammals obtained by mice, genus Perognathus (Rodentia: Mr. Hendee at Pebas and Iquitos, upper Heteromyidae). Jour. Mammal., vol. 48, Amazons. Ibid., ser. 10, vol. 34, pp. 285- no. 1, pp. 27-37, figs. 1-7. 295. Patton, James L., and Alfred L. Gardner Wagner, A. 1972. Notes on the systematics ofProechimys 1845. Diagnosen einiger neuen Arten von Na- (Rodentia: Echimyidae), with emphasis gem unde Handfluglern. Arch. fur Na- on Peruvian forms. Occas. Papers, Mus. turgeschichte, vol. 1, pp. 145-147. 14 AMERICAN MUSEUM NOVITATES NO. 2817

1948. Beitrage zur Kenntniss der Saugthiere vol. 2. London, Hippolyte Bailliere, Amerika's. Abhandl. Akad. Wiss., Publisher, 500 pp., pls. 1-21. Munchen, Bd. 5, Abth. 2, p. 291, pls. Yonenaga, Yatio 1-2. 1975. Karyotypes and chromosome polymor- Waterhouse, G. R. phism in Brazilian rodents. Caryologia, 1848. A natural history of the mammalia, vol. 28, no. 3, pp. 269-286, figs. 1-18.