phalangeal morphology of the paromomyidae (?primates, plesiadapiformes): the evidence for gliding...

Phalangeal Morphology of the Paromomyidae (?Primates,Plesiadapiformes): The Evidence for Gliding Behavior Reconsidered

MARK W. HAMRICK,* BURT A. ROSENMAN, AND JASON A. BRUSHDepartment of Anthropology, Kent State University, Kent, Ohio 44240

KEY WORDS finger morphology; vertical climbing; verticalclinging; dermoptera

ABSTRACT A comparative morphometric analysis of isolated proximaland intermediate phalanges attributed to the paromomyids Ignacius graybul-lianus and Phenacolemur simonsi was undertaken to test the hypothesis thatthese fossil phalanges exhibit evidence of a dermopteran-like interdigitalpatagium. Linear dimensions were collected for the fossil phalanges and acomparative sample of associated proximal and intermediate phalangesrepresenting extant tree squirrels, tree shrews, dermopterans (colugos),gliding rodents and marsupials, and prosimian primates. Quantitative dataindicate that the proximal and intermediate phalanges of paromomyids aremost similar in their overall shape to those of the dermopteran Cynocephalus.The proximal phalanges of paromomyids and colugos possess well-developedflexor sheath ridges and broad, high shafts, whereas the intermediatephalanges of these taxa are most similar to one another in their trochlearmorphology. Discriminant analysis indicates that all of the paromomyidintermediate phalanges resemble those from colugo toes more so than thosefrom colugo fingers. Moreover, the relative length and midshaft proportions ofboth the proximal and intermediate phalanges of paromomyids closelyresemble those of several squirrels that lack an interdigital patagium. Thefollowing conclusions are drawn from this study: 1) paromomyids share anumber of derived phalangeal features with modern dermopterans that maybe indicative of a phylogenetic relationship between them, 2) existingintermediate phalanges of paromomyids are inconsistent with the ‘‘mittengliding’’ hypothesis because they do not possess the distinctive length andmidshaft proportions characteristic of colugo manual intermediate phalan-ges, and 3) paromomyids share with colugos and the scaly-tailed squirrelAnomalurus several aspects of phalangeal morphology functionally related tofrequent vertical clinging and climbing on large-diameter arboreal supports.Am J Phys Anthropol 109:397–413, 1999. r 1999 Wiley-Liss, Inc.

Paromomyidae is a family of early Ter-tiary mammals traditionally included withinthe primate suborder Plesiadapiformes(Fleagle, 1988; Van Valen, 1995). Previousworkers (e.g., Szalay and Delson, 1979; Sza-lay et al., 1987) proposed that Plesiadapifor-mes is the sister group of Euprimates (butfor alternative views, see Wible and Covert,1987; Kay et al., 1992). Beard (1989, 1990)and Kay et al. (1990), however, recently

suggested that the plesiadapiform familyParomomyidae shares a number of derivedpostcranial and basicranial features withliving dermopterans. Beard (1990, 1993b)

Grant sponsor: National Science Foundation; Grant number:IBN-9603808.

*Correspondence to: Dr. Mark W. Hamrick, Department ofAnthropology, Box 5190, Kent State University, Kent, OH 44242.E-mail: [email protected]

Received 24 August 1998; accepted 20 March 1999.

AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 109:397–413 (1999)

r 1999 WILEY-LISS, INC.

suggested further that the derived postcra-nial traits shared by paromomyids and ex-tant dermopterans are not only indicative ofa close phylogenetic relationship betweenthe two but also indicate that paromomyidswere ‘‘mitten gliders’’ like living colugos.

Modern mammals that glide between arbo-real supports possess a skin membrane, orpatagium, that extends between the fore-and hindlimb and increases the lift and dragof the animal when it is airborne (Thoring-ton and Heaney, 1981; Thorington, 1984).Colugos differ from other gliding mammalssuch as gliding rodents and marsupials inthat they possess an interdigital patagiumbetween their fingers and toes (Pocock, 1926).Colugo fingers exhibit especially long inter-mediate phalanges (Leche, 1886; Shufeldt,1911; Pocock, 1926) and, as Beard (1993b, p.65) noted, ‘‘only the highly derived phalan-geal osteology of these animals bears such adirect functional relationship to their pata-gium and associated gliding habits.’’ Beard(1990, 1993b) argued that the isolated inter-mediate phalanges he attributed to the paro-momyid Phenacolemur simonsi resembledthose from the hand of extant colugos inbeing slender, straight, and significantlylonger than the proximal phalanges. Beard’s(1989, 1990, 1993b) functional analysis ofthese fossil specimens led him to concludethat paromomyids possessed an interdigitalpatagium like that of modern dermopterans,and therefore paromomyids were dermop-teran-like ‘‘mitten gliders.’’

The hypothesis that paromomyids weremitten gliders has been challenged by bothKrause (1991) and Runestad and Ruff (1995).Krause (1991) concluded that the inferenceof mitten gliding in paromomyids is basedupon the assumption that the isolated paro-momyid phalanges were assigned correctlyto the hand and foot. Krause (1991) foundthat this assumption had not been validatedby sufficient evidence. Beard himself stated(1989, p. 178) that ‘‘there is no direct evi-dence that any of the proximal phalangesknown for Phenacolemur represent the ma-nus as opposed to the pes,’’ and (p. 183)‘‘there is no direct evidence that any of theintermediate phalanges known for Phenac-olemur represent the manus as opposed tothe pes.’’ Runestad and Ruff (1995) later

provided preliminary data indicating thatthe isolated phalanges attributed to Paromo-myidae also resembled those of certain non-mitten-gliding mammals such as Microce-bus in their length and relative midshaftdimensions. Moreover, Runestad and Ruff(1995) demonstrated that all modern glidingmammals share distinctive long bone diaphy-seal dimensions not observed in the longbones of paromomyids. Thus, the question ofwhether or not paromomyids were mittengliders is one that remains actively debatedand is still unresolved (Martin, 1993).

Recent studies (e.g., Meldrum andYuerong, 1988; Begun, 1993; Begun et al.,1994; Hamrick et al., 1995; Jungers et al.,1997) demonstrated that integrated func-tional and morphometric analyses can beapplied to reconstruct the positional behav-iors of fossil taxa based on their phalangealmorphology. Furthermore, Hamrick et al.(1995) showed that quantitative methodscan be used successfully to attribute isolatedphalanges of fossil anthropoid primates tothe hand and foot. Our study used a compara-tive functional and morphometric approachin order to resolve the question of whether ornot paromomyid phalanges exhibit evidenceof a dermopteran-like interdigital patagium.The implications of these data for under-standing the locomotor and postural behav-iors of paromomyids, as well as for under-standing archontan phylogeny, are discussed.

MATERIALS AND METHODSSample

The fossil phalanges included here forstudy are from the collections of the UnitedStates National Museum of Natural History(USNM), Smithsonian Institution, Washing-ton, DC. The nine fossils included for analy-sis are complete proximal and intermediatephalanges from the USNM collections attrib-uted by Beard (1989) to either Phenacole-mur simonsi or Ignacius graybullianus(Table 1). Krause (1991) suggested that cer-tain phalanges attributed by Beard (1989) toP. simonsi might actually represent I. gray-bullianus, and vice versa. Our results andconclusions would not differ if the speciesallocations were switched for any of thefossil specimens, as long as the fossils re-mained attributable to the Paromomyidae.

398 M.W. HAMRICK ET AL.

We therefore follow Beard’s (1989) speciesallocations of the fossil specimens for ease ofcommunication. Beard (1989) also includedseveral partial proximal and intermediatephalanges (c.f., P. jepseni) from the Univer-sity of Michigan and the United StatesGeological Survey collections in his analy-sis; however, due to the incomplete nature ofthese specimens they were excluded fromthis study. Phalangeal morphology was alsoexamined in a large comparative sample ofextant gliding and nongliding mammals rep-resenting five orders of therian mammals.The sample includes gliding rodents, marsu-pials, and dermopterans as well as nonglid-ing, arboreal primates, tree shrews, rodents,and marsupials (Table 2). In all, 342 phalan-ges were measured, of which 181 were proxi-mal phalanges and 161 were intermediatephalanges (Table 2).

Measurements

The following nine measurements weretaken on the fossil specimens as well ason associated proximal and intermediatemanual and pedal phalanges of the extanttaxa included in the comparative sample(Fig. 1): 1) maximum phalangeal length (L),from the most proximal and distal points oneach phalanx; 2) midshaft breadth (MSB),from the medial and lateral margins of thephalanx at approximately midshaft; 3) mid-shaft height (MSH), from the dorsal and

ventral borders of the phalanx at approxi-mately midshaft; 4) proximal articularbreadth (PAB), from the most medial andlateral margins of the proximal articularsurface; 5) proximal articular height (PAH),from the most dorsal and ventral margins ofthe proximal articular surface; 6) dorsaltrochlear breadth (DTB), from the most me-dial and lateral margins of the phalangealtrochlea on its dorsal surface; 7) palmar/plantar trochlear breadth (PTB), from themost medial and lateral margins of thephalangeal surface on its ventral surface; 8)trochlear length (TL), from the most proxi-mal and distal points on the trochlear sur-face; and 9) trochlear height (TH), from themost ventral and dorsal points on the troch-lea. These linear dimensions were chosen foranalysis because they capture functionallysignificant aspects of phalangeal shape dis-cussed by previous authors (e.g., Meldrumand Yuerong, 1988; Beard, 1993b; Begun,1993; Hamrick et al., 1995). Linear measure-ments were taken from the extant speci-mens using digital calipers, whereas metricdimensions were taken from the fossil speci-mens using a stereo light microscope with areticle attachment.

Statistical analysis

Statistical analyses were performed onlog-shape ratios derived from the raw data

TABLE 1. Fossil paromomyid phalanges includedfor study

Taxon1Specimennumber Specimen

Phenacolemursimonsi USNM 442248 Proximal phalanx

Phenacolemursimonsi USNM 442249 Proximal phalanx

Phenacolemursimonsi USNM 442250 Intermediate phalanx




Ignacius gray-bullianus USNM 442256 Proximal phalanx

Ignacius gray-bullianus USNM 442253 Intermediate phalanx

Ignacius gray-bullianus USNM 442255 Intermediate phalanx

1 Taxonomic attribution according to Beard (1989).

TABLE 2. Extant sample includedfor comparative analysis

Taxon IndividualsProximal

phalanges1Intermediatephalanges1

Order MarsupialaPetaurus brevi-

ceps 4 13/14 9/14Order Rodentia

Anomalurusbeecrofti 2 3/4 3/3

Glaucomysvolans 4 13/12 13/12

Sciurus niger 4 11/12 9/12Order Scandentia

Tupaia tana 4 8/8 8/8Order Dermop-

teraCynocephalus

sp. 4 14/15 14/12Order Primates

Galago senega-lensis 4 10/10 9/9

Tarsius sp. 4 16/15 14/121 Manual phalanges/pedal phalanges.

399PAROMOMYID PHALANGEAL MORPHOLOGY

following the procedure described by Mosi-mann and James (1979), Falsetti et al.(1993), Jungers et al. (1995), and Hamrick(1998). Specifically, a size variable was cre-ated for each phalanx that was the geomet-ric mean of all measurements taken on thatphalanx. Each linear dimension on thatphalanx was then divided by this size vari-able and logged to create a log-shape vari-able. The statistical analyses were chosen inorder to answer two primary questions. First,what modern taxon possesses phalangesthat are most similar in shape to those of thefossil taxa? Second, can the metric data beused to discriminate manual from pedalphalanges in the extant sample and, if so,then can we use these data to allocate thefossil phalanges to the hand and foot?

The first question was investigated usinga multivariate principal components analy-sis (PCA) to summarize the osteometric data.PCA was the preferred multivariate tech-

nique because the goal of the analysis was toexplore variation in the metric data as wellas examine the distribution of the extantand fossil specimens in multidimensionalmorphospace (Neff and Marcus, 1980; deQueiroz and Good, 1997). The proximal andintermediate phalanges were analyzed sepa-rately, but both the manual and pedal pha-langes were included in each analysis. Vi-sual inspection of the plotted PCA factorscores and examination of the log-shaperatio data allowed us to assess pheneticsimilarities in phalangeal morphology be-tween the fossil and extant species. Thesecond question, i.e., should each fossil pha-lanx be allocated to either the hand or thefoot, was investigated using multivariatediscriminant analysis. Two discriminantanalyses, one including proximal phalangesand the other including intermediate phalan-ges, were run for the extant sample in orderto test the hypothesis that phalanges from

Fig. 1. Intermediate phalanx of a paromomyid in dorsal (a), lateral (b), proximal (c), and distal(d) views, showing linear measurements included for analysis. Measurement abbreviations are explainedin text and in Table 3.


the hand could be discriminated from thoseof the foot. Two discriminant analyses, oneincluding proximal phalanges and the otherincluding intermediate phalanges, were thenrun for phalanges of the mitten-glider Cyno-cephalus only. A separate pair of discrimi-nant analyses was run for Cynocephalusbecause previous authors (Beard, 1990; Kayet al., 1990) suggested that paromomyidsshare a close phylogenetic relationship withmodern dermopterans. Log-shape ratio val-ues for the fossil paromomyid phalangeswere entered into each analysis as un-knowns and then classified as either hand orfoot based on their factor scores. Posteriorprobabilities for group assignment are pro-vided for each fossil specimen. USNM 442251was excluded from the multivariate analysisbecause the damaged trochlea on this speci-men prevented calculation of all measure-ments, and therefore precluded calculationof a log-size variable.

RESULTSProximal phalanges

Summary statistics for the proximal pha-lanx log-shape ratio values are shown inTable 3. Results of the principal componentsanalysis performed on the log-shape ratio

values for the proximal phalanges includedfor study are shown in Figure 2a and Table4. The first principal component axis ac-counts for approximately 33% of the vari-ance and separates colugos, the scaly-tailedsquirrel Anomalurus, and the paromomyidsfrom the other taxa in the sample. Factorscores on this axis are most highly corre-lated with log-shape ratio values for mid-shaft dorsoventral height and mediolateraldiameter (Table 4). The paromomyids, colu-gos, and Anomalurus each have low scoreson this axis, and each exhibits relativelybroad and high midshaft dimensions (Table3). Similarities among these taxa in theirproximal phalanx midshaft dimensions ap-pear to reflect the very well-developed, proxi-mally extensive flexor sheath ridges observedon the ventral surface of their proximalphalanges (Fig. 3). The second principalcomponent axis accounts for approximately18% of the variance and separates the pri-mates and two of the fossil specimens, whichhave high scores on this axis, from the othertaxa. This axis is most highly correlated(negatively) with proximodistal length ofthe trochlea, revealing that the paromomy-ids and primates tend to have trochleae thatare quite short proximodistally. The third

TABLE 3. Mean and standard deviation (in parentheses) for proximal phalanx log-shape ratio values1

Taxon L MSB MSH PAB PAH DTB PTB TL TH

Manual phalangesPb 1.45 (.14) 2.39 (.11) 2.45 (.11) .25 (.12) 2.08 (.59) 2.24 (.09) 2.07 (.07) 2.25 (.13) 2.20 (.12)Ab 1.63 (.05) 2.35 (.12) 2.18 (.02) .01 (.08) .02 (.12) 2.42 (.10) 2.26 (.04) 2.08 (.02) 2.36 (.02)Gv 1.64 (.06) 2.64 (.06) 2.55 (.10) .12 (.08) 2.10 (.11) 2.13 (.08) 2.02 (.04) 2.01 (.04) 2.29 (.04)Sn 1.51 (.05) 2.43 (.06) 2.50 (.08) .24 (.07) 2.03 (.08) 2.31 (.14) .01 (.04) 2.17 (.04) 2.31 (.05)Tt 1.43 (.07) 2.44 (.08) 2.56 (.08) .27 (.07) 2.05 (.11) 2.16 (.05) 2.05 (.03) 2.12 (.07) 2.30 (.04)Cs 1.60 (.05) 2.32 (.08) 2.26 (.06) .08 (.08) 2.08 (.06) 2.39 (.08) 2.27 (.03) 2.17 (.07) 2.17 (.09)Gs 1.63 (.06) 2.41 (.05) 2.54 (.10) .18 (.03) 2.02 (.04) 2.33 (.15) 2.05 (.06) 2.16 (.06) 2.27 (.06)Ts 1.89 (.08) 2.48 (.08) 2.49 (.10) .16 (.08) .01 (.09) 2.39 (.17) 2.11 (.06) 2.28 (.08) 2.29 (.09)

Pedal phalangesPb 1.36 (.13) 2.43 (.10) 2.39 (.11) 2.18 (.07) .04 (.11) 2.21 (.08) 2.10 (.05) 2.18 (.09) 2.25 (.05)Ab 1.49 (.12) 2.33 (.04) 2.14 (.13) .03 (.04) 2.01 (.04) 2.31 (.04) 2.23 (.04) 2.19 (.08) 2.28 (.07)Gv 1.64 (.65) 2.61 (.07) 2.61 (.05) .21 (.04) 2.05 (.07) 2.11 (.04) 2.04 (.03) 2.07 (.11) 2.34 (.03)Sn 1.57 (.04) 2.40 (.05) 2.46 (.09) .28 (.07) 2.04 (.09) 2.30 (.15) .02 (.03) 2.21 (.06) 2.44 (.08)Tt 1.48 (.07) 2.43 (.08) 2.47 (.09) .27 (.06) 2.06 (.07) 2.21 (.10) 2.06 (.02) 2.17 (.05) 2.33 (.08)Cs 1.55 (.08) 2.25 (.05) 2.26 (.06) .12 (.05) 2.14 (.11) 2.41 (.08) 2.28 (.14) 2.21 (.18) 2.10 (.16)Gs 1.66 (.09) 2.50 (.10) 2.49 (.05) .10 (.03) .09 (.04) 2.30 (.05) 2.11 (.04) 2.15 (.05) 2.29 (.04)Ts 1.86 (.10) 2.56 (.09) 2.51 (.07) .21 (.05) .07 (.09) 2.27 (.10) 2.06 (.06) 2.32 (.06) 2.40 (.09)

Fossil phalanges442248 1.69 .04 2.07 2.18 2.30 2.58 .01 2.22 2.39442249 1.75 2.04 2.27 2.22 2.27 2.55 .01 2.11 2.27442256 1.65 2.31 2.22 .04 2.24 2.34 2.06 2.12 2.371 Pb, Petaurus breviceps; Ab, Anomalurus beecrofti; Gv, Glaucomys volans; Sn, Sciurus niger; Tt, Tupaia tana; Cv, Cynocephalus sp.;Gs, Galago senegalensis; Ts, Tarsius sp.; L, maximum phalangeal length; MSB, midshaft breadth; MSH, midshaft height; PAB,proximal articular breadth; PAH, proximal articular height; DTB, dorsal trochlear breadth; PTB, palmar/plantar trochlear breadth;TL, trochlear length; TH, trochlear height. Specimen numbers shown in the first column refer to fossil specimens described in Table 1.


principal component axis accounts for 14%of the variance and is mostly highly corre-lated (negatively) with breadth of the proxi-mal articular surface. The primates, glidingsquirrels, Cynocephalus, and the three fossilspecimens have high scores on this axis, andall share proximal articular surfaces thatare relatively narrow mediolaterally.

The discriminant analysis run on theproximal phalanx log-shape ratio data issuccessful in correctly classifying only 60%of the manual and 66% of the pedal phalan-ges in the entire comparative sample(Wilkes-Lambda F 5 2.52, P 5 0.01). Thediscriminant analysis classified all of theproximal phalanges of paromomyids to the

Fig. 2. Bivariate plots offactor scores for the first twoprincipal component axes of(a) principal componentsanalysis of proximal phalanxdimensions, and (b) principalcomponents analysis of inter-mediate phalanx dimensions.Ellipses enclose the range ofvalues for extant taxa. Fossilspecimens are indicated byasterisks, and specimen num-bers are shown on the plot.


hand with posterior probabilities of 0.70–0.72. The discriminant scores are most highlycorrelated with height of the trochlea (Table5). The manual phalanges have low discrim-inant scores and trochleae that are highdorsopalmarly, whereas the pedal phalan-ges have high scores and trochleae that aremore compressed dorsoplantarly. When thediscriminant analysis is performed on thelog-shape ratio values for the proximal pha-langes of Cynocephalus alone, the discrimi-nation is highly significant (Wilkes-LambdaF 5 5.24, P 5 0.001) and correctly assigned

93% (27 of 29) of the colugo proximal phalan-ges to the hand and foot (Fig. 4a). The colugodiscriminant analysis assigned the fossilspecimens USNM 442256 and 442249 to thehand and USNM 442248 to the foot (Fig. 4a).Posterior probabilities are very high (P 51.00) for the assignment of specimens 442256and 442248, but lower (P 5 0.63) for USNM442249. The attribution of USNM 442256and 442249 to the hand agrees with theassignment of Beard (1989). Beard (1989,

Fig. 3. Proximal phalanges of (A) Tupaia, (B) Galago, (C) Ignacius graybullianus (USNM 442256),and (D) Cynocephalus in lateral view. Arrows in C and D indicate the well-developed and proximallyextended flexor sheath ridges of Ignacius and Cynocephalus. Tupaia, Ignacius, and Cynocephalus weremodified and redrawn from Beard (1993b). Not to scale.

TABLE 4. Principal component loadings for the firstthree axes of the PCA performed on proximal phalanx

log-shape ratio values

Measurement1Factor 1(33.2%)

Factor 2(18.3%)

Factor 3(14.0%)

MSH 2.88 .03 2.10MSB 2.77 .01 2.41PTB .72 2.06 2.37TH 2.59 2.42 .01DTB .59 2.55 .06PAB .51 .01 2.70TL .12 2.68 .49L .15 .60 .34PAH .26 .56 .30

1 Measurement abbreviations are explained in Table 3.

TABLE 5. Pearson correlation coefficients (loadings)between discriminant scores and log-shape ratio

variables included in discriminant analyses of manualand pedal proximal phalanges

Measurement1

Canonical loadings

Analysis withall taxa

Analysis withCynocephalus only

L 2.19 .46MSB 2.12 2.51MSH .19 .03PAB .40 2.38PAH .29 .33DTB .22 .14PTB 2.12 .05TL 2.34 .18TH 2.46 2.33



1990) did, however, believe that USNM442248 belonged to the hand of P. simonsi.

The discriminant scores for the analysis ofcolugo proximal phalanges are most highlycorrelated with log-shape ratio values forproximal phalanx length and midshaft diam-eter (Table 5). An index of proximal phalanxmidshaft diameter relative to length, shownin Figure 5a, illustrates that colugo manualphalanges are slightly longer and more slen-der than their pedal phalanges. The rela-tively poor discrimination between manualand pedal proximal phalanges in the first setof discriminant analyses is due to the factthat the proximal phalanges of the foot arerelatively more gracile than those from thehand in Sciurus, Galago, Tarsius, andTupaia, whereas the reverse is true in Cyno-cephalus, Glaucomys, Petaurus, and Anoma-lurus (Fig. 5a). Thus, although relative mid-

shaft diameter can be used to separatemanual from pedal phalanges within extantdermopterans, the same criterion cannot beapplied across taxa. Moreover, it is clearfrom Figure 5a that the proximal phalangesof a mammal possessing an interdigital pata-gium do not differ significantly in theirrelative length and midshaft dimensionsfrom those of a mammal that lacks aninterdigital patagium.

Intermediate phalanges

Summary statistics for the intermediatephalanx log-shape ratio values are shown inTable 6. Results of the principal componentsanalysis performed on the log-shape ratiovalues for the intermediate phalanges in-cluded for study are shown in Figure 2b andTable 7. The first principal component axisaccounts for approximately 35% of the vari-

Fig. 4. Univariate plotsof discriminant scores fordiscriminant analysis of (a)proximal phalanx dimen-sions and (b) intermediatephalanx dimensions in thedermopteran Cynocephalus.Fossil specimens are indi-cated by asterisks, and speci-men numbers are shown onthe plot. Pedal phalangesare represented by opencircles and manual phalan-ges are represented by solidcircles.


ance and separates Cynocephalus, Anomalu-rus, and the paromomyids, which has highscores on this axis, from the other taxa,which has low scores on this axis. This axisis most highly correlated with mediolateralbreadth of the trochlea. Shape ratios indi-cate that Cynocephalus, Anomalurus, and

the fossil taxa have relatively low values forthis dimension and therefore possess troch-leae that are relatively compressed mediolat-erally. The second principal component axisaccounts for approximately 18% of the vari-ance and separates the primates, Cynocepha-lus, and the paromomyids from most of the

Fig. 5. Plots show-ing ratios of midshaftdiameter to phalanxlength for (a) proximalphalanges and (b) in-termediate phalanges.Fossil specimens are in-dicated by open circles,and specimen numbersare shown on the plot.The vertical line in themiddle of each box isthe mean value, andthe horizontal line rep-resents one standarddeviation. Manual pha-langes are indicated byopen vertical bars andpedal phalanges are in-dicated by solid verti-cal bars.


other mammals in the sample, includingAnomalurus (Fig. 2b). Factor scores on thisaxis are most highly correlated with dorso-ventral height of the trochlea and phalanxlength. Examination of the shape ratio dataindicates that these taxa, which all havehigh scores on this axis, possess trochleaethat are reduced dorsally and phalangealshafts that are quite long (Fig. 6). The thirdprincipal component axis accounts for ap-proximately 14% of the variance and is alsohighly correlated (negatively) with phalan-geal length. The primates, gliding squirrels,and Cynocephalus have low scores on thisaxis and relatively elongate phalanges incontrast to the five fossil specimens, which

have high scores on this axis and compar-atively lower values for relative phalanxlength.

The discriminant analysis run on the inter-mediate phalanx log-shape ratio data for theentire sample is successful in correctly clas-sifying only 64% of the manual and 62% ofthe pedal phalanges (Wilkes-Lambda F 51.90, P 5 0.06). The discriminant analysisclassified all of the intermediate phalangesof paromomyids to the hand with posteriorprobabilities of 0.68–0.75. The discriminantscores are most highly correlated with rela-tive phalanx length (Table 8). The manualphalanges have high scores and long shaftswhen compared to the pedal phalanges,which have low scores and relatively shortershafts. When the discriminant analysis isperformed on the log-shape ratio values forthe intermediate phalanges of Cynocephalusalone, the discrimination is highly signifi-cant (Wilkes-Lambda F 5 7.28, P , 0.001)and correctly assigned 100% (24 of 24) of thecolugo proximal phalanges to the hand andfoot (Fig. 4b). The discriminant scores aremost highly correlated with log-shape ratiovalues for intermediate phalanx length andmidshaft diameter (Table 5). An index ofintermediate phalanx midshaft diameterrelative to phalanx length shows that

TABLE 6. Mean and standard deviation (in parentheses) for intermediate phalanx log-shape ratio values1

Taxon L MSB MSH PAB PAH DTB PTB TL TH

Manual phalangesPb 1.06 (.23) 2.36 (.09) 2.38 (.06) .21 (.05) .01 (.05) 2.08 (.04) 2.06 (.06) 2.18 (.13) 2.18 (.09)Ab 1.72 (.05) 2.75 (.08) 2.40 (.04) .12 (.03) .12 (.08) 2.46 (.05) 2.28 (.01) 2.08 (.03) .03 (.02)Gv 1.53 (.09) 2.61 (.03) 2.62 (.05) .18 (.04) 2.04 (.15) 2.12 (.06) 2.13 (.06) .01 (.09) 2.18 (.03)Sn 1.35 (.13) 2.46 (.04) 2.53 (.03) .21 (.07) .02 (.09) 2.20 (.13) 2.09 (.04) 2.18 (.06) 2.11 (.03)Tt 1.14 (.12) 2.31 (.10) 2.42 (.17) .19 (.07) 2.05 (.09) 2.08 (.08) 2.01 (.05) 2.17 (.11) 2.28 (.11)Cs 2.03 (.08) 2.63 (.11) 2.42 (.07) 2.05 (.08) .09 (.06) 2.33 (.10) 2.23 (.05) 2.33 (.14) 2.10 (.06)Gs 1.45 (.11) 2.26 (.08) 2.55 (.12) .20 (.03) .09 (.07) 2.09 (.18) 2.01 (.04) 2.32 (.08) 2.51 (.05)Ts 1.73 (.20) 2.39 (.12) 2.51 (.07) .22 (.05) 2.01 (.08) 2.15 (.13) 2.06 (.09) 2.33 (.17) 2.48 (.15)

Pedal phalangesPb 1.05 (.13) 2.47 (.14) 2.45 (.07) .14 (.05) .05 (.07) 2.06 (.09) 2.06 (.08) 2.12 (.13) 2.07 (.09)Ab 1.38 (.08) 2.50 (.09) 2.37 (.04) .13 (.01) .06 (.02) 2.34 (.10) 2.37 (.03) .02 (.04) .01 (.02)Gv 1.42 (.02) 2.56 (.06) 2.59 (.06) .21 (.03) .02 (.04) 2.15 (.04) 2.15 (.03) .01 (.07) 2.21 (.02)Sn 1.34 (.05) 2.41 (.05) 2.43 (.05) .28 (.06) .01 (.07) 2.23 (.10) 2.11 (.04) 2.23 (.04) 2.21 (.16)Tt 1.15 (.07) 2.27 (.07) 2.43 (.09) .24 (.04) 2.07 (.05) 2.11 (.06) 2.04 (.03) 2.19 (.07) 2.27 (.04)Cs 1.68 (.13) 2.51 (.08) 2.38 (.09) .03 (.09) .16 (.06) 2.34 (.07) 2.26 (.06) 2.27 (.09) 2.09 (.09)Gs 1.43 (.14) 2.29 (.12) 2.54 (.08) .20 (.03) .06 (.10) 2.01 (.08) .01 (.04) 2.34 (.06) 2.51 (.10)Ts 1.52 (.23) 2.44 (.10) 2.48 (.10) .16 (.13) 2.02 (.08) 2.06 (.14) .05 (.09) 2.28 (.17) 2.43 (.14)

Fossil phalanges442250 1.82 2.62 2.33 .13 2.03 2.55 2.19 2.19 2.03442252 1.75 2.48 2.32 .12 2.03 2.67 2.13 2.23 2.01442253 1.82 2.50 2.27 .19 2.12 2.54 2.27 2.31 .01442254 1.90 2.43 2.31 .04 2.01 2.68 2.23 2.23 2.04442255 1.79 2.51 2.38 .02 2.03 2.56 2.06 2.22 2.03

1 Abbreviations are explained in Table 3. Specimen numbers shown in the first column refer to fossil specimens described in Table 1.

TABLE 7. Principal component loadings for the firstthree axes of the PCA performed on intermediate

phalanx log-shape ratio values

Measurement1Factor 1(35.62%)

Factor 2(18.31%)

Factor 3(14.23%)

PTB 2.84 .03 2.11DTB 2.77 2.19 2.22MSB 2.65 .43 .45TH .65 2.45 .36PAB 2.64 2.09 .08L .57 .46 2.54TL .09 2.89 .03MSH .34 .36 .77PAH .40 .23 2.26



manual intermediate phalanges of Cyno-cephalus have low values for this index,whereas pedal intermediate phalangeshave higher values (Fig. 5b). Thus, themanual intermediate phalanges of colugosare relatively longer and have narrowershafts than their pedal intermediate phalan-ges.

The discriminant analysis performed oncolugo phalanges assigns all five of the fossil

specimens to the foot with very high poste-rior probabilities (P . 0.98) for each assign-ment (Fig. 4b). Note that this classificationis precisely the opposite of that obtainedwhen the discriminant analysis is run for alltaxa in the comparative sample. USNM442251, excluded from the multivariateanalysis because of its damaged trochlea,has a relatively high midshaft diameter/length ratio like the other intermediate pha-langes. We therefore suggest that USNM442251 be classified in the same manner asthe other specimens in each analysis (Fig.5b). The attributions of USNM 442250,442251, 442252, and 442255 to the footagree with the assignments of Beard (1989).Beard (1989, 1990) did, however, believethat USNM 442253 belonged to the hand ofI. graybullianus and USNM 442254 be-longed to the hand of P. simonsi.

The two discriminant analyses, one runfor the entire sample and one run only forCynocephalus, classified the paromomyidphalanges differently in each analysis. Theplot of intermediate phalanx midshaft

TABLE 8. Pearson correlation coefficients (loadings)between discriminant scores and log-shape ratio

variables included in discriminant analyses of manualand pedal intermediate phalanges

Measurement1

Canonical loadings

Analysis withall taxa

Analysis withCynocephalus only

L .84 .87MSB 2.26 2.61MSH 2.29 2.32PAB 2.26 2.57PAH 2.28 .02DTB 2.31 .40PTB 2.10 2.08TL 2.21 2.32TH 2.15 2.25


Fig. 6. Intermediate phalanges of (A) Tupaia, (B) Galago, (C) Ignacius graybullianus (USNM 442253),and (D) Cynocephalus in lateral view. Arrowhead in A indicates the dorsally expanded articular surface onthe trochlea of Tupaia. Tupaia, Ignacius, and Cynocephalus were modified and redrawn from Beard(1993b). Not to scale.


breadth relative to length shown in Figure5b provides some insight into why the twodiscriminant analyses differ in their attribu-tion of the fossil specimens. Colugos possessmanual intermediate phalanges that exhibitthe lowest values for this index; however,colugo pedal intermediate phalanges alsohave quite narrow midshafts for their length.The discriminant analysis run on intermedi-ate phalanges for the entire sample misclas-sified 10 of the 26 colugo intermediate pha-langes. All of these misclassified colugophalanges are pedal phalanges that wereassigned to the hand by the discriminantfunction. Thus, the pedal intermediate pha-langes of colugos are quite similar in theirrelative length and midshaft dimensions tothe manual intermediate phalanges of non-‘‘mitten-gliding’’ mammals such as Tarsiusand Glaucomys (Fig. 5b). This explains whythe paromomyid intermediate phalangeswere classified differently in the two discrim-inant analyses: the intermediate phalangesof paromomyids are similar in their relativelength and midshaft dimensions to thosefrom the hand of vertical clingers (e.g., Tar-sius and Glaucomys) as well as those fromthe foot of Cynocephalus.

DISCUSSIONFunctional morphology of paromomyid

phalanges

The manual and pedal proximal phalan-ges of paromomyids exhibit a number offeatures related to powerful flexion of thefingers and toes. Foremost among thesefeatures are the well-developed flexor sheathridges. The flexor sheath ridges on the proxi-mal phalanges of Ignacius and Phenacole-mur are pronounced and flare both ventrallyand laterally, creating a concavity betweenthe ridges on the ventral phalangeal sur-face. Qualitative and quantitative data pre-sented here indicate that the morphology ofthis region in paromomyids resembles thatobserved in proximal phalanges of the colugoCynocephalus and the scaly-tailed glidingsquirrel Anomalurus. Anomalurus frequentlyclings to vertical tree trunks, and the scaleson the ventral surface of its tail stabilize theanimal during clinging postures by interlock-ing with the trunk surface (Nowak, 1991).Bristles on the ventral surface of the tail in

woodpeckers serve a similar function (Rich-ardson, 1942). Cynocephalus is also knownto frequently cling to large-diameter verticalsupports, and both Cynocephalus andAnomalurus climb up vertical trunks afterlanding at the end of a glide (Wharton, 1950;Lekagul and McNeely, 1977; Nowak, 1991).Colugos and scaly-tailed flying squirrels (andpossibly paromomyids; see Szalay and Lu-cas, 1993) also share terminal phalangesthat are very deep proximally and distallyand compressed mediolaterally (Fig. 7). Thismorphology again appears related to theirhabit of frequent clinging and climbing onvertical tree trunks, using interlocking be-tween their well-developed claws and thearboreal substrate (Feduccia, 1993). Addi-tional aspects of paromomyid proximal pha-

Fig. 7. Third digits in lateral view of (top to bottom)Anomalurus beecrofti (manual), Cynocephalus volans(pedal), Phenacolemur simonsi (USNM 442248 and442250, and USGS 17847), Galago crassicaudatus(pedal), and Tupaia glis (pedal). Note similarities inmorphology of the proximal, intermediate, and distalphalanges among Anomalurus, Cynocephalus, and Phe-nacolemur. Not to scale.


lanx morphology related to frequent andforceful flexion at the proximal interphalan-geal joints have been discussed in detail byBeard (1993b), and include strong dorsoven-tral curvature of the shaft and ventral exten-sion of the articular surface on the trochlea.These data together suggest that paromomy-ids were capable of frequent and forcefulflexion of the proximal interphalangeal jointsof the hand and foot, typical of animals thatfrequently use vertical postures and climb-ing in an arboreal environment.

The intermediate phalanges of paromomy-ids, like the proximal phalanges, also ex-hibit features related to the use of verticalarboreal supports. The first of these featuresis the ventrally extended and dorsally re-stricted articular surface on the trochlea.The ventrally extended trochlear articularsurface provides a large area of articularcontact between the intermediate and distalphalanges when the distal interphalangealjoint is flexed. In contrast, arboreal mam-mals such as tree shrews, which often ex-tend their terminal phalanges at the distalinterphalangeal joints (Jenkins, 1974), pos-sess dorsally expanded articular surfaces ontheir trochleae (Fig. 6). The intermediatephalanges of colugos and Anomalurus alsoresemble those of paromomyids in havingtrochleae that are compressed mediolater-ally. Mediolateral compression of the troch-lea on the intermediate phalanx is corre-lated with the mediolateral compression ofthe distal phalanx in these taxa.

The robusticity indices for the intermedi-ate phalanges (Fig. 5b) also illustrate thatparomomyids share with tarsiers, glidingsquirrels, and colugos intermediate phalan-ges that have quite narrow midshafts rela-tive to their length. These comparative datasuggest further that taxa that frequentlyuse vertical arboreal supports in their loco-motor and postural behaviors share rela-tively elongate and gracile intermediate pha-langes, particularly on their fingers (Fig. 7).A similar pattern is observed in the pedalphalanges of tree-trunk-climbing birds (Rich-ardson, 1942; Bock and Miller, 1959; Hilde-brand, 1995; Clark et al., 1998). Bock andMiller (1959) suggested that the more elon-gate intermediate and subungual phalanges

of trunk-climbing birds enabled them tospread their toes far apart when clinging tovertical trunks. In the case of mammals,however, there is no reason why increasingthe length of the intermediate phalangeswould increase the span between the toesany more than would increasing the lengthof the proximal phalanges. An alternativeexplanation is that the long intermediatephalanges of vertical clingers increase thelength of the distal portion of the digits, sothat the hand and foot can effectively sub-tend a greater central angle on large-diameter, vertical, cylindrical, arboreal sup-ports (Cartmill, 1985). Evidence from thephalanges of paromomyids therefore ac-cords well with evidence from other regionsof the paromomyid postcranial skeletonwhich suggests that vertical climbing andclinging were frequent locomotor and pos-tural behaviors practiced by these animals(e.g., Beard, 1991).

Comparative functional analysis of paro-momyid phalanges suggests that these mam-mals possessed capabilities for powerful digi-tal flexion using well-developed claws.Extant clawed, arboreal mammals that pos-sess similar capabilities for forceful fingerand toe flexion utilize postural and locomo-tor behaviors such as vertical clinging andclimbing in an arboreal environment. Evi-dence from the cheiridium as well as fromthe long bones suggests that paromomyidsprobably practiced each of these behaviorsto at least some degree. Quantitative datapresented here reveal that the intermediatephalanges of paromomyids do not possessthe distinctive length and midshaft propor-tions that characterize those from the fin-gers of ‘‘mitten gliders.’’ Existing phalangesof paromomyids, as well as previously de-scribed paromomyid long bones (Runestadand Ruff, 1995), therefore provide no conclu-sive evidence that paromomyids possessed acolugo-like patagium.

Significance of phalangeal morphology forunderstanding archontan phylogeny

Primate superordinal relationships havebeen widely debated in recent years (e.g.,Pettigrew, 1986; Szalay et al., 1987; Wibleand Covert, 1987; Adkins and Honeycutt,1991; Bailey et al., 1992; Kay et al., 1992;


Novacek, 1992; MacPhee, 1993, and refer-ences therein). Postcranial evidence has fig-ured prominently in these debates, usuallyin support of a phylogenetic relationshipbetween primates and plesiadapiforms (e.g.,Szalay et al., 1987; Beard, 1993a). Szalayand Lucas (1993) identified two derived char-acters from the cheiridium that support aphylogenetic relationship between dermop-terans and bats: 1) ungual phalanges thatare compressed mediolaterally and deep bothproximally and distally (see also Yalden,1985; Feduccia, 1993); and 2) elongation ofthe fourth and fifth pedal rays.

Our data suggest that bats and dermopter-ans share two additional derived characterstates of the cheiridium. First, chiropterans,including the fossil bat Icaronycteris, anddermopterans exhibit the derived conditionsof having pedal intermediate phalanges thatare elongate relative to their pedal proximalphalanges (Fig. 8; the same condition in

Anomalurus is interpreted as a conver-gence). The primitive eutherian condition,exhibited by tree shrews, squirrels, carni-vores, insectivores, and primates, is to havepedal intermediate phalanges that are sig-nificantly shorter in length than the proxi-mal phalanges. Thewissen and Babcock(1992) showed that bats and colugos are alsoderived in having intermediate phalangeson their fingers that are long relative totheir proximal phalanges. Second, the inter-mediate phalanges of Cynocephalus, Ptero-pus, and paromomyids are distinctive inhaving proximal articular surfaces that arehigh dorsoventrally and compressed medio-laterally (Fig. 9). The primitive condition forArchonta is represented by tree shrews andprimates in which the intermediate phalanxproximal articular surface is broad mediolat-erally but more compressed dorsoplantarly.Colugos and bats also share a ratchet-liketendon-locking mechanism between the ten-

Fig. 8. Plot showing ratios of intermediate phalanxlength to proximal phalanx length for toes II–V invarious archontans. Primates include Cheirogaleus me-dius (n 5 1), Saimiri sciureus (n 5 1), Loris tardigradus(n 5 1), Galago senegalensis (n 5 2), and Tarsius sp. (n 52). Scandentians include Tupaia tana (n 5 4) andPtilocercus lowii (n 5 1). Microchiropterans includeMyotis lucifugus (n 5 2), Eptesicus fuscus (n 5 2), andPteronotus parnelli (n 5 2). Megachiropterans include

Pteropus hypomelanus (n 5 3) and Rousettus amplexicau-datus (n 5 4), and dermopterans include Cynocephalusvolans and C. variegatus (n 5 3). Note that the bats andcolugos both possess intermediate phalanges that arelong relative to the length of their proximal phalangeswhen compared to the tree shrews and primates. Valuesfor the fossil bat Icaronycteris index are from Jepsen(1966).


don of m. flexor digitorum longus and thefibrous flexor sheath that is derived foreutherian mammals (Bennett, 1993; Quinn,1993; Simmons and Quinn, 1994). The major-ity of postcranial characters used previouslyto support a chiropteran-dermopteran cladewere forelimb features related either di-rectly or indirectly to the shared presence ofa patagium (e.g., Wible and Novacek, 1988;Thewissen and Babcock, 1991; Stafford andThorington, 1998). Our study indicates thatbats, dermopterans, and paromomyids re-semble one another in derived, phylogeneti-cally significant aspects of phalangeal mor-phology that are not necessarily related tothe presence of a patagial membrane.

CONCLUSIONS

Comparative morphometric analysis ofproximal and intermediate phalanges attrib-uted to the paromomyid plesidapiforms Ig-nacius graybullianus and Phenacolemur si-monsi reveals that these fossil phalangesare most similar in their overall shape tothose of the dermopteran Cynocephalus. Theproximal phalanges of paromomyids re-semble those of Cynocephalus, as well asthose of the scaly-tailed gliding squirrelAnomalurus, in having well-developed flexorsheath ridges, whereas the fossil intermedi-ate phalanges resemble those of Cynocepha-lus and Anomalurus in having distal articu-lar surfaces that are expanded ventrally,reduced dorsally, and compressed mediolat-erally. Discriminant analysis and robustic-ity indices suggest that the intermediatephalanges of paromomyids are most similarin their relative shape to those from the toes

of Cynocephalus and to those from the fin-gers of arboreal mammals that cling tovertical supports, such as Tarsius, Glauco-mys, and Anomalurus. The hypothesis thatparomomyids were ‘‘mitten gliders’’ is there-fore not supported, since none of the existingparomomyid intermediate phalanges pos-sess the distinctive length and midshaftproportions characteristic of colugo manualintermediate phalanges. Comparative func-tional analysis suggests that paromomyidswere capable of forceful finger and toe flex-ion using clawed digits related to arborealpositional behaviors such as vertical trunk-clinging and -climbing. Finally, comparativeanalysis also demonstrates that bats, der-mopterans, and possibly paromomyids shareproportions and toe pedal intermediate pha-lanx articular surface morphologies that arederived among archontan mammals.

ACKNOWLEDGMENTS

We are grateful to Dr. K.C. Beard forpermitting us to study the paromomyid pha-langes in his care. We also thank Mr. L.Jellema, Ms. Carol Camillo, and Drs. B.Latimer and T. Matson, Cleveland Museumof Natural History, Ms. Linda Gordon andDr. R. Thorington, National Museum ofNatural History, Smithsonian Institution,and Dr. J.G.M. Thewissen, NortheasternOhio Universities College of Medicine, forallowing us access to skeletal collections ofextant mammals in their care. Drs. J.G.M.Thewissen, D. Begun, and B. Richmondprovided helpful comments that improvedthe quality of the manuscript. Funding forthis research was provided in part by National

Fig. 9. Pedal intermediate phalanges of (A) Tupaia, (B) Galago, (C) Ignacius graybullianus (USNM442253), (D) Cynocephalus volans, and (E) Pteropus hypomelanus in proximal view. Note the mediolater-ally compressed and dorsoplantarly expanded proximal articular surfaces of the colugo, bat, andparomomyid. Tupaia, Ignacius, and Cynocephalus were modified and redrawn from Beard (1993b). Not toscale.


Science Foundation grant IBN-9603808 toM.W.H.

LITERATURE CITED

Adkins RM, Honeycutt RL. 1991. Molecular phylogenyof the superorder Archonta. Proc Natl Acad Sci USA88:10317–10321.

Bailey WJ, Slightom JL, Goodman M. 1992. Rejection ofthe ‘‘flying primate’’ hypothesis by phylogenetic evi-dence from the e-globin gene. Science 256:86–89.

Beard KC. 1989. Postcranial anatomy, locomotor adapta-tions, and paleoecology of early Cenozoic Plesiadapi-dae, Paromomyidae, and Micromomyidae (Eutheria,Dermoptera). Ph.D. thesis. Baltimore: Johns HopkinsUniversity.

Beard KC. 1990. Gliding behavior and palaeoecology ofthe alleged primate family Paromomyidae (Mamma-lia, Dermoptera). Nature 345:340–341.

Beard KC. 1991. Vertical postures and climbing in themorphotype of Primatomorpha: implications for loco-motor evolution in primate history. In: Senut B,Coppens Y, editors. Origines de la bipedie chez leshominides. Paris: CNRS. p 79–87.

Beard KC. 1993a. Phylogenetic systematics of the Prima-tomorpha, with special reference to Dermoptera. In:Szalay FS, Novacek MJ, McKenna MC, editors. Mam-mal phylogeny, placentals. New York: Springer-Verlag. p 129–150.

Beard KC. 1993b. Origin and evolution of gliding inearly Cenozoic Dermoptera (Mammalia, Primatomor-pha). In: MacPhee RDE, editor. Primates and theirrelatives in phylogenetic perspective. New York: Ple-num Press. p 63–90.

Begun DR. 1993. New catarrhine phalanges from Ruda-banya (Northeastern Hungary) and the problem ofparallelism and convergence in hominoid postcranialmorphology. J Hum Evol 24:373–402.

Begun DR, Teaford MF, Walker A. 1994. Comparativeand functional anatomy of Proconsul phalanges fromthe Kaswanga Primate Site, Rusinga Island, Kenya. JHum Evol 26:89–165.

Bennett MB. 1993. Structural modifications involved inthe fore- and hind-limb grip of some flying foxes(Chiroptera: Pteropodidae). J Zool 229:237–248.

Bock WJ, Miller WD. 1959. The scansorial foot of thewoodpeckers with comments on the evolution of perch-ing and climbing feet in birds. Am Mus Novit 1931:1–45.

Cartmill M. 1985. Climbing. In: Hildebrand M, BrambleD, Liem K, Wake D, editors. Functional vertebratemorphology. Cambridge: Harvard University Press. p73–88.

Clark J, Hopson J, Hernandez R, Fastovsky D, Montel-lano M. 1998. Foot posture in a primitive pterosaur.Nature 391:886–889.

de Queiroz K, Good DA. 1997. Phenetic clustering inbiology: a critique. Q Rev Biol 72:3–31.

Falsetti A, Jungers WL, Cole T. 1993. Morphometrics ofthe callitrichid forelimb: a case study in size andshape. Int J Primatol 14:551–572.

Feduccia A. 1993. Evidence from claw geometry indicat-ing arboreal habits of Archaeopteryx. Science 259:790–793.

Fleagle JG. 1988. Primate adaptation and evolution.New York: Academic Press. 486 p.

Hamrick MW. 1998. Functional and adaptive signifi-cance of primate pads and claws: evidence from NewWorld anthropoids. Am J Phys Anthropol 106:113–127.

Hamrick MW, Meldrum DJ, Simons EL. 1995. Anthro-poid phalanges from the Oligocene of Egypt. J HumEvol 28:121–145.

Hildebrand M. 1995. Analysis of vertebrate structure.New York: John Wiley and Sons, Inc. 657 p.

Jenkins FA Jr. 1974. Tree-shrew locomotion and theorigins of primate arborealism. In: Jenkins FA Jr,editor. Primate locomotion. New York:Academic Press.p 85–115.

Jepsen GL. 1966. Early Eocene bat from Wyoming.Science 154:1333–1339.

Jungers WL, Falsetti A, Wall C. 1995. Shape, relativesize, and size-adjustments in morphometrics. YrbkPhys Anthropol 38:137–161.

Jungers WL, Godfrey LR, Simons EL, Chatrath PS.1997. Phalangeal curvature and positional behaviorin extinct sloth lemurs (Primates, Palaeopropitheci-dae). Proc Natl Acad Sci USA 94:11998–12001.

Kay RF, Thorington RW Jr, Houde P. 1990. Eoceneplesiadapiform shows affinities with flying lemurs notprimates. Nature 345:342–344.

Kay RF, Thewissen JGM, Yoder AD. 1992. Cranialanatomy of Ignacius graybullianus and the affinitiesof the Plesiadapiformes. Am J Phys Anthropol 89:477–498.

Krause DW. 1991. Were paromomyids gliders? Maybe,maybe not. J Hum Evol 21:177–188.

Leche W. 1886. Die Saugethiergattung Galeopithecus:eine morphologische Untersuchung. Kungl SvenskaVetenskap Akad 21:1–92.

Lekagul B, McNeely JA. 1977. Mammals of Thailand.Bangkok: Sahakarnbhat. 758 p.

MacPhee RDE. 1993. Primates and their relatives inphylogenetic perspective. New York: Plenum Press.383 p.

Martin RD. 1993. Primate origins: plugging the gaps.Nature 363:223–234.

Meldrum DJ, Yuerong P. 1988. Manual proximal pha-lanx of Laccopithecus robustus from the latest Miocenesite of Lufeng. J Hum Evol 17:719–731.

Mosimann JE, James FC. 1979. New statistical methodsfor allometry with applications to Florida red-wingedblackbirds. Evolution 23:444–459.

Neff N, Marcus L. 1980. A survey of multivariatemethods for systematics. New York: American Mu-seum of Natural History. 243 p.

Novacek MJ. 1992. Mammal phylogeny: shaking thetree. Nature 356:121–125.

Nowak RM. 1991. Walker’s mammals of the world.Baltimore: Johns Hopkins University Press.

Pettigrew JD. 1986. Flying primates? Megabats havethe advanced pathway from eye to midbrain. Science231:1304–1306.

Pocock RI. 1926. The external characters of the flyinglemur (Galeopterus temminckii). Proc Zool Soc Lond1926:429–444.

Quinn TH. 1993. A tendon locking mechanism (TLM) inthe foot of Cynocephalus variegatus (Dermoptera). AmZool 33:133.

Richardson F. 1942. Adaptive modifications for tree-trunk foraging in birds. Univ Calif Pub Zool 46:317–368.

Runestad JA, Ruff CB. 1995. Structural adaptations forgliding in mammals with implications for locomotorbehavior in paromomyids. Am J Phys Anthropol 98:101–119.

Shufeldt RW 1911. The skeleton in the flying lemurs,Galeopteridae. Phil J Sci 6:185–214.

Simmons NB, Quinn TH. 1994. Evolution of the digitaltendon locking mechanism in bats and dermopterans:a phylogenetic perspective. J Mamm Evol 2:231–254.


Stafford B, Thorington RW Jr. 1998. Carpal develop-ment and morphology in archontan mammals. J Mor-phol 235:135–155.

Szalay FS, Delson E. 1979. Evolutionary history of theprimates. New York: Academic Press. 580 p.

Szalay FS, Lucas SG. 1993. Cranioskeletal morphologyof archontans, and diagnoses of Chiroptera, Volitan-tia, and Archonta. In: MacPhee RDE, editor. Primatesand their relatives in phylogenetic perspective. NewYork: Plenum Press. p 187–226.

Szalay FS, Rosenberger AL, Dagosto M. 1987. Diagnosisand differentiation of the order Primates. Yrbk PhysAnthropol 30:75–105.

Thewissen JGM, Babcock SK. 1991. Distinctive cranialand cervical innervation of wing muscles: new evi-dence for bat monophyly. Science 251:931–936.

Thewissen JGM, Babcock SK. 1992. The origin of flightin bats. BioSci 42:340–345.

Thorington RW Jr. 1984. Flying squirrels are monophy-letic. Science 225:1048–1050.

Thorington RW Jr, Heaney LR. 1981. Body proportionsand gliding adaptations of flying squirrels (Petauristi-nae). J Mamm 62:101–114.

Van Valen L. 1995. The origin of the plesiadapid pri-mates and the nature of Purgatorius. Evol Monogr15:1–79.

Wharton CH. 1950. Note on the life history of the flyinglemur. J Mamm 31:264–273.

Wible JR, Covert HH. 1987. Primates: cladistic diagno-sis and relationships. J Hum Evol 16:1–22.

Wible JR, Novacek MJ. 1988. Cranial evidence for themonophyletic origin of bats. Am Mus Novit 2911:1–19.

Yalden DW. 1985. Forelimb function in Archaeopteryx.In: Hecht M, Ostrom J, Viohl G, Wellnhofer P, editors.The beginnings of birds. Eichstatt: Bronner & Daent-ler KG. p 91–97.


phalangeal morphology of the paromomyidae (?primates, plesiadapiformes): the evidence for gliding...

Documents