sign in sign up
<<
Home , Afrotheria, Cursorial, Digitigrade, Elephant shrew, Hyopsodontidae, Leptictida

several modernplacentalorders (,Artiodactylaand andMiocene,ahighdegreeofcursorialityevolved in oftheextinctionevent(O’Learyetal.,2013).During the modern placentalmammalsemerged withinhundredsofthousands evacuated bythenon-neornitheandinosaurs,firstmembers of the 2013). Fromthisancestor, andgiventhefreedom toradiateintoniches , inwhichtheheelmakescontactwithground(O’Leary et al., mammalian ancestorshowstheancestralconditionofaplantigrade (O’Leary etal.,2013).Thereconstructionofthisvirtualplacental dweller thatlookedsomewhatlikeanopossumwithabushy 1316 Received 12August 2013;Accepted5December2013 ([email protected];*Authors for correspondence [email protected]) 3209, SouthAfrica. KwaZulu-Natal, Private BagX01,Scottsville Life Sciences, University of School of a small(6–245 (MYA) (O’Learyet al.,2013).Theancestralplacentalmammalwas – (K–Pg)boundary65.5 drove thenon-neornitheandinosaurstoextinctionat from asinglelineagethatsurvivedtheasteroidimpactevent The extraordinarydiversityofmodernplacentalmammalsevolved Cursors, KEY WORDS:-,Macroscelidae,Runningspeed, trail . case ofcursorialityinmammalssmallerthan1 In thisstudywereportontheevolutionofmicro-cursoriality, aunique Barry G.Lovegrove*andMetoborO.Mowoe* The evolutionofmicro-cursorialityinmammals RESEARCH ARTICLE © 2014.PublishedbyTheCompanyofBiologistsLtd|JournalExperimentalBiology(2014)217,1316-1325doi:10.1242/jeb.095737 INTRODUCTION ABSTRACT newly evolved pre-adaptive toopen,aridhabitats,andbecamemorederivedinthe toavoidpredators.DuringtheMiocene,micro-cursorialitywas selection forrapidrunningspeedsfacilitatedbylocalknowledge,in cursoriality evolvedfirstinforests,presumablyresponseto earlierthaninothermammaliancrowngroups.Micro- the plesiomorphicplantigradefootofpossum-likeancestral stem macroscelids.Micro-cursorialityinmacroscelidsevolvedfrom Prodiacodon established intheEarlyEoceneearliestmacroscelid extraordinary runningspeedanddigitigradyofelephant-shrewswas though, remainednon-,plantigradeandsmall(<1 in theOligoceneandMiocene.Themajorityofmammalspecies, global coolingandthereplacementofforestswithopenlandscapes in theArtiodactyla,PerissodactylaandCarnivoracoincidentwith typically associatedwithfastunguligradecursors.Cursorialityevolved possess exceptionallyhighmetatarsal:femurratios(1.07)thatare those ofmostmammalssmallerthan1 mammals. South ,whichwecomparedwithpublisheddataforother shrews ( running speedandlimbmorphologydatafortwospeciesofelephant- Elephantulus , butwasprobablyinheritedfromPaleocene,Holarctic Elephantulus

g), insectivorous,tree-climbing(scansorial)- spp., Macroscelidae)fromNamaqualand, maximum runningspeedswerehigherthan and kg.

kg. We obtainednew elephant-shrews with

million yearsago Elephantulus

kg). The also metatarsals and distalmusclereductions(Evans, 1942;Carrano, been studiedmorphologically to datedisplayextremelyelongated ‘ the Latinprefix‘ because thewordisderivedfrom confirms theearlyrecognition ofunusualhindlimbmorphology shrews) (Stanhopeetal.,1998). TheetymologyofMacroscelidea sister familytoAfrosoricida(,goldenmolesand otter are placedinthesuperorderAfrotheria(Springeretal.,1997) asa cursorial animalsrelativetotypicaltaxa.Elephant-shrews us totestthehypothesisthatelephant-shrewsareexceptionalmicro- physiology ofelephant-shrewsorsengis(Macroscelidea)prompted to1.4inthegiraffe (Carrano, 1999). and Janis,1993).MT:F ratiosrange from <0.1insomeplantigrade maximum runningspeedsbutverydifferent MT:F ratios(Garland speed; twomammalswithsimilarbodysizescanhave is nodirectrelationshipbetweenMT:F ratioandmaximumrunning distance locomotion(GarlandandJanis,1993).Nevertheless, there running speeds(SteudelandBeattie,1993)cost-effective long- they arealsoindicativeofmorespecializedlimbadaptationsforfast hindlimb length,stridelengthandrunningspeed(Hildebrand,1974), Although higherMT:F ratiosareoftenassociatedwithincreased cursoriality inmammals(GarlandandJanis,1993;Carrano,1999). the (F),MT:F ratio,isoftenusedasaproxyfor and Janis,1993).Thelengthratiobetweenthemetatarsals(MT) in theheelbeingraisedoff theground(Hildebrand,1974;Garland particular, areelongatedrelativetootherhindlimbbones,resulting andunguligradelimbs,inwhichthemetatarsals, commonly associatedwithcursoriality, namelythederived metatarsals, andmoredistalmuscleinsertionpoints(Carrano,1999). graviportal mammalshavemorerobustlimbelements,shorter muscle insertionpointlocatedclosertothehipjoint,whereas metatarsals, moreslenderlimbelements,shorterfemora,anda locomotor performance.Inshort,cursorialtaxahavelonger analysis providedbiologicallyrealisticindicesofmammalian measures ofmultiplemorphologicaltraitsinaprincipalcomponents ‘cursorial’ and ‘graviportal’ (weight-bearing)locomotionbasedupon (Carrano, 1999)showedthatamorphologicalcontinuumbetween making distinctionsbetweenlocomotorperformance…’.Carrano argued that‘…morphologyshouldremainthefundamentalbasisfor evaluation ofthesedefinitionproblems,Carrano(Carrano,1999) Janis, 1993;SteinandCasinos,1997;Carrano,1999).Inan (Taylor etal.,1970;Garland,1983a;1983b;Garlandand including behaviour, biomechanics,physiologyandmorphology because locomotorperformanceisinfluencedbymultiplevariables, However, moreexplicitdefinitionsofcursorialityremainobscure Lovegrove, 2012b;LovegroveandMowoe,2013). Janis andWilhelm, 1993;Yuanqing etal.,2007;Jardine 2012; grasslands followingtheEoceneThermalMaximum(Janis,1993; Perissodactyla) inresponsetotheemergence ofopenlandscapesand skelis’, meaninghiporthigh.Indeed, elephant-shrewsthathave Several publishedobservationsonthemorphologyand Variations inthedimensionsofthesetraitsarebornelimbs Loosely defined,cursorialmammalsarethosethatrunfast. macro’, meaninglarge, andtheGreekword Macroscelides, whichcomprises

The Journal of Experimental Biology running speed(MRS;km 1979; Rathbun,2009). been describedas‘themicro-cursorialadaptivesyndrome’ (Rathbun, RRS arequestionable becausetwobody-sizerelated variables(log allometric scalingpatternsthathave beenreportedandwhichquantify relative runningspeedasanotable ecologicalconsideration,the (Iriarte-Díaz, 2002).However, althoughweaccepttheconceptof confirms thatRRSdecreasesmore rapidlywithincreasingbodysize for mammalslarger than500 lengths body temperature( (Tenrecidae) andgoldenmoles(Chrysochloridae),displayamean 1999). Moreover, whereastheirclosestrelatives,tenrecs RESEARCH ARTICLE inflection ata function ofbodymass(M and Beattie,1993;Iriarte-Díaz,2002).TheregressionofRRS asa small mammalsdisplayhigherRRSsthanlarge mammals(Steudel display higherMRSsthansmallmammals(Garland,1983a),whereas Dooren, 1999;Iriarte-Díaz,2002).Onaverage,large mammals monogamy, smallprecociallittersandbodysize<1 (exposed sheltersites,mixedherbivoryandinsectivory, social the hindones.Itisveryfast….’ which canbemovedinmanyways.Itsfrontlegsmuchshorterthan ‘By me,called Is egterseergeswindinhetlopen….’ TheEnglishtranslationreads: manieren bewegenkan,sijnvoorpotenveelkorteralsdeagterpoten. oliphantsmuis genaamt,omsignlangesnuitdewelkehijopallerley Africa byRobertJacobGordonon2August1779:‘Doormij, during thefourthjourneyintonorthwesternCapeofSouth Macroscelides proboscideus accompanied adrawingnowthoughttohavebeenthatof description oftheseAfricansmallmammalsinanannotationthat elephant-shrews werealsorecognizedintheveryfirstwritten (Carrano, 1999). principal componentsanalysisofmammallimbdimensions ( unguligrade cursors(Lovegrove,2012b).Last,elephant-shrews ratio fortheirbodysize,comparabletothoseofthefastest 2010). Elephant-shrewsalsodisplayanexceptionallyhighMT:F and arethoughttoenhancemuscleperformance(ClarkePörtner, correlated withtheMT:F ratioinothercursors(Lovegrove,2012b), supplementary materialFig. Afrotherians (Lovegrove,2012a;Lovegrove,2012b)(see apomorphy (derivedcharacteristic)of4.4°Cbetweensister elephant-shrews is37.2°C( Elephantulus Two ecologicallyrelevantmeasures ofrunningspeed,maximum The limbsandotheruniquecharacteristicsofelephant-shrews The extraordinarylarge hindlimbs/quartersandspeedof R maximumrunningspeed(kmh maximumlikelihood MRS ML femur Cretaceous–Paleogeneboundary confidenceinterval M Akaike’s informationcriterion K–Pg F CI AIC R relativerunningspeed(body phylogeneticgeneralized leastsquares ordinaryleastsquares RRS metatarsal:femurratio metatarsal PGLS OLS MT:F MT T List of symbolsand abbreviations List of b b s − 1 ), havebeenusedintheliterature(Van DammeandVan M ) displayedthehighestindexofcursorialityina body temperature(°C) body mass oliphantsmuis b of ~500 T b ) of32.8°C(n h g, i.e.thenegativeslopeof regression − b 1 ) hasanegativeslope,anddisplaysan ) andrelativerunningspeed(RRS;body

g issteeper, whichtheauthorssuggest n (Rookmaker, 1989).Itwasrecorded

(elephant mouse),forthelongsnout =8 ),indicatingaprofound S1). Highbodytemperaturesare lengthss =8 species),themean −1 ) −1 )

kg), have T b of possibly asearlythePaleocene. environments withsomesuretyduringtheEarlyEocene,but display amicro-cursorialcapacitywhichevolvedinforest data wereavailable.We testedthehypothesisthatelephant-shrews appropriate mammalmodelsforwhichrunningspeedandMT:F kirkii body MRS, maximumabsoluterunningspeed;MT:F, metatarsal:femur. plotted asfunctionoflog Fig. edwardii dataset of135mammalshadMT:F ratioshigherthanthoseof 1).Excludingthegiraffe, onlyfivemammalsinthecombined (Table The averageMT:F ratioofthetwospecies Body mass(g) open-habitat trend ofdecreased bodysizeandincreasedMT:F ratiosthatoccurredin unguligrade datasets,respectively. Theredarrowindicatestheevolutionary lines arethePGLSregression fortheseparateunguligradeandnon- regression lineforthecombineddataset, whereastheblackandblue edwardii morphology oftwospeciesrockelephant-, using aphylogeneticallyinformedapproach. Díaz, 2002).Consequently, we resortedheretoanalysesofMRSonly, maximal runningspeed(km Namaqualand, SouthAfrica.We recordedtheMT:F ratioand Elephantulus rupestris Table MRS range(km Metatarsal:femur ratios RESULTS Femur length(mm) eaasllnt m)26.68±1.15( Metatarsal length(mm) MRS (km MT:F ratio

Metatarsal:femur ratio In thisstudywemeasuredtherunningspeedsandlimb

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 .The metatarsal:fermurratios(MT:F) of135mammal species 1. .0 .10.1 0.01 0.001 , andthespringbok lengths 0 1. Bonedimensionsandmaximumrunningspeedsof h Elephantulus and − 1 (Smith 1839)and ) Elephantulus The JournalofExperimentalBiology(2014)doi:10.1242/jeb.095737

Lipotyphla Tenrecidae Rodentia Perissodactyla Carnivora Artiodactyla s E. rupestris

− h 1 − 1 and 14.4–28.8 ) M spp. b species duringMiocene aridification. 10 and E.edwardii ) wereregressedagainsteachother(Iriarte- body mass. (three 60.10±5.02 ( E. rupestris 23.6±4.8 (n 26.54±0.69 ( 1.067±0.041 ( Fg 1).No Antidorcas marsupialis)(Fig. Body mass(kg)

h

01010 10,000 1000 100 10 1 E. rupestris − Gazella 1 ) andcomparedthesedatawith The redlineisthePGLS =10) Other Elephantulus PGLS non-unguligrademammals PGLS unguligrademammals PGLS alldata Elephantulus n n n 1)49.90±4.22( =10) 5 26.13±0.38( =5) 5 24.54±1.26( =5) n 5 1.075±0.042( =5) spp., thedikMadoqua Elephantulus (Smith 1831),from Giraffe species (thisstudy) species 10.8–21.6 E. edwardii 19.4±2.2 ( species Elephantulus was 1.07 n =4) 1317 n n n =4) =5) =5) n =5) E.

The Journal of Experimental Biology Non-unguligrade mammals( Unguligrade mammals( phylogenetic signal[ 1318 RESEARCH ARTICLE mammal smallerthan1 All mammals( Statistics evolutionary model(Table CI=0.997] closeto,butnotquiteequalaBrownianmotion lower 95%confidenceinterval(CI)=0.901,and fit tothedata(Fig. PGLS withmaximumlikelihood(ML)estimationshowedthe best PGLS regressionsrendertheOLSregressionmeaningless. The Akaike’s informationcriterion)of >200 betweentheOLSand measure ofeachmodelrelativetothebestmodel,whereAICis the OLSregressionpluslarge valuesofdeltaAIC( non-unguligrade regressions(seebelow).Thusthepositiveslopeof consistent withthenegativeslopesforseparateunguligradeand Fig. indicating anincreaseinMT:F ratiowithbodysize(notshownin However, whereastheslopeofOLSregressionwaspositive,i.e. Table Brownian motionmodelofevolution ( compared withtheOLSregression),whichbothindicated a unguligrade mammalwerethetwoPGLSregressions( 3).Thebestfittingregressionmodelsforthe56speciesof (Table residual MT:F ratiointheunguligrade andnon-unguligradedatasets the MT:F ratiosof >0.7(Fig. showed anMT:F ratioof log relationships. relevance topreviousstudiesthatfoundnosignificantallometric squares (PGLS)modelsfittedtothedatabecausetheybear least squares(OLS)andthephylogeneticgeneralized mammals (Fig. species aresomarkedlyhigherthanthoseofothersimilar-sized mammals isquiteunnecessarybecausethedataforthesetwo regression (− ( CI=0.819; Table phylogenetic signal( provided thebestfittodata (Fig. 79 non-unguligrademammals, thePGLSwithML estimation − R Intercept P R Intercept P Slope Slope R Intercept P Slope AIC Pagel’s lambda AIC Pagel’s lambda AIC Pagel’s lambda 0.124) wassteeperthanthat ofthenon-unguligradePGLS There wasasignificantphylogeneticsignalforboth The OLSandPGLSregressionsofMT:F ratioasafunctionof 10 2 2 2 1), thoseofthephylogeneticregressionswerenegative, M 2. Resultsofvariousregressionmodelsfittedtodatasetsmetatarsal:femurratioasafunctionlog b of thecompletedataset( n 0.031; Table =135)

1). Nevertheless,wereporttheresultsofordinary

2). Theslopeoftheunguligrade PGLSregression E. edwardii n

λ 1, redline)andalsoshowedasignificant =56) λ =0.964, significantlydifferent toboth =0.589, lower95%CI=0.279, upper95%

kg, apartfromE.edwardii n

=79) 2, Fig.

2). and 1). n 15 eesgiiat(al 2). =135) weresignificant(Table E. rupestris

1). A statisticalcomparisonof 1) andconfirmedasignificant λ =1; Table 0.410 0.029 1.252 <0.001 0 − 0 − 0 0.479 <0.001 0.073 OLS − 0.061 − 0.327 − 0.138 0.020 0.237 157.8 26.2 16.2 with thoseofother

2, Fig. λ and E.rupestris, =1, upper95% ∆ 1). Forthe AIC=54.9 ∆ M AIC; a 0.338 0.024 0.919 <0.001 − − 1 1 1 0.448 <0.01 − − − 0.064 0.193 0.064 PGLS Brownian − b λ 0.033 0.124 146.7 81.1 216.6 0.042 and =0, recorded was28.8 mammals (excluding inflection at20 143 mammalspecies. Fig. The meanMRSsofE.rupestris regression oflog (Blomberg etal.,2003) significant phylogeneticsignalsasdetectedbyBlomberg etal.’s Maximum running speedmodel Maximum running ( mammal inthesmalldatasetwasCapehuntingdog 2).Theinflectionoccurredatabodymassof20 (Fig. ( mammals. 19.4 whereas thosewiththelowestresidualsweregroundhog, species withthehigheststudentizedresidualswerelagomorphs, Purvis, 1997)orCook’s distanceD were nooutliers,thatis,studentizedresiduals>3(Jonesand ( lagomorphs, 12marsupials,10carnivoresandoneartiodactyal Lycaon pictus Perognathus longimembris)to6000 Madoqua kirkii Piecewise regressionidentifiedasignificantinflectioninthe In thesignificantOLSregressionofsmallmammaldatathere

.Maximum runningspeed(MRS)asafunctionofbodymass 2. km log10 Maximum running speed –1 (km h ) h − 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 1 The JournalofExperimentalBiology(2014)doi:10.1242/jeb.095737 , respectively(Table

kg separatingthe allometries oftheMRSsmalland large ). The80species<20 .0 .1011 0.1 0.01 0.001 spp. Elephantulus 0.357 <0.01 0.919 <0.001 − − 0.589 1 0.964 0.449 <0.01 PGLS ML − − − 0.089 0.193 0.057 − ). Bothlog 10 0.031 0.124 177.5 81.1 221.6 0.040 MRS asafunctionoflog km The verticaldashedlineindicatesa significant h Elephantulus − K 1 by afemale estimate (Table 10 M Body mass(kg) 20 kg b and residuallog

1). Thefastestindividualrun and

kg comprised49rodents,eight -values >0.5.Fourofthefive spp.) ranginginM 10 E. ruprestris. E. edwardii kg (Loxodontaafricana body mass(kg)

3). 10

10 M b of 143species 10

were 23.6and kg. Thelargest 100 1000 MRS showed b from 9 g )

The Journal of Experimental Biology log Non-PGLS MANOVA Variable Marmota monax,thestripedskunk, c b a Trait most obviousobservationswerethat(1)theMRSofseven the 3).Inthisgraphicalrepresentation,the (Blomberg etal.,2012)(Fig. an OLStothephylogeneticallyindependentlinearcontrasts intervals toaBrownianmotionPGLS,whichisequivalentfitting smaller than20 comparison oftheMRSelephant-shrewswiththosemammals ML estimatedbranchlengthtransformationsthatwouldallowa 95% confidenceandpredictionintervalstoaPGLSregressionwith phylogenetic signal.To ourknowledge,thereisnowayoffitting RESEARCH ARTICLE log log log log log maximum runningspeed(MRS)andmetatarsal:femur(MT:F) ratio Table mammals smallerthan20 Table PGLS MANOVA western pygmypossum,Cercatetus concinnus and <1,upper95%CI=0.988;Table Pagel’s lambda( which thebranchlengthtransformationswereestimatedwith AIC, andhencethebestfitofmodels,wasPGLSmodelin (star phylogeny).However, theevolutionarymodelwithlowest fit tothedatathanamodelthatassumednophylogeneticstructure Brownian motionevolutionarymodelprovidedaconsiderablybetter than thatoftheOLSregression( the smallmammaldatawassignificantanditsAICvaluelower showed asignificantphylogeneticsignal(Table kirkii saltatorial), butexcludedthesolitaryunguligradedatumfor locomotor modes(,digitigrade,lagomorph-likeand comprised 80mammalssmallerthan20 against whichtheelephant-shrewMRSswereinitiallycompared influence oflocomotormodeonMRS(Table phylogenetic effects, thePGLSMANOVA showed nosignificant mammal datasetbecauseofsamplesize.Oncecorrectedfor unguligrade datum( 4).We omittedthesolitary effect oflocomotormodeonMRS(Table locomotor modes.A non-PGLSMANOVA confirmedasignificant suggested thattheremaybedifferences inMRSbetweenthefour 76 non-unguligradespecies. 56 unguligradespecies. 80 mammalspecies<20 Locomotor mode Residuals Locomotor mode log Residuals log We resortedinsteadto fittingthe95%confidenceandprediction The phylogeneticPGLSBrownianmotionregressionmodelof The presenceofsomanylagomorphswithhighMRSresiduals 10 10 10 10 10 10 M M M 10 10 residual MT:F ratio residual MT:F ratio residual MRS . BothbodymassandresidualMT:F ratiointhisdataset b b b 3. Resultsofrandomizationtestsusedtodetectphylogeneticsignal( 4. ResultsofMANOVA analysestestingtheinfluenceofbody mass andlocomotormodeonthemaximumrunningspeedsof M M for MT:F ratio for MT:F ratio for MRS b b a

kg. a λ =0.905, significantly>0,lower95%CI=0.699, c b c b M. kirkii

kg.

kg ) fromtheMANOVAs ofthesmall NK 90370.002 0.142 0.003 0.044 0.397 <0.001 0.580 0.008 0.772 0.482 79 0.360 79 1.328 56 56 80 80 2 75 3 d.f. 76 1 1 ∆ I=88;Tbe5).Thusa AIC=18.88; Table

5), confirmingasignificant Mephitis mephitis

kg andincludedall . 0.007 1.289 0.613 SS 0.388 0.144 4.002

4). Thusthemodel

3). Variance of , andthe M. 0.003 0.017 0.204 0.005 0.144 4.002 MS K of ourphylogeneticallycorrectedregressionsshowedsignificant expanded, moretaxonomicallydiversedataset( regressions (completedataset,,carnivores).In our there wasnosignificantphylogeneticsignalforresiduallog smaller than0.5 intervals (Fig. lay abovetheregressionline,butnot95%confidence mammals andCarnivora.Thedatafor a comparisonoftheMRSselephant-shrewswithdigitigrade overlap. Thus,asstatedearlier, alackofbody-sizeoverlapobviates rodents, sciuridsandmarsupials);threesquirrelsaprimatedid body sizeoverlapwiththeplantigrademammals(non-sciurid (2) thedigitigrademammalsandlagomorphsshowedamarginal eight lagomorphslayabovetheupper95%confidenceinterval,and relationship betweenMT:F ratioand log carnivores, andfoundnosignificantphylogeneticallycorrected ratio allometryin49mammalspecies,30ungulatesand 19 Garland andJanis(GarlandJanis,1993)analyzedtheMT:F allometry ofMT:F ratios,theresultsofouranalysesarenoteworthy. Although itwasnotaspecificobjectiveofthisstudytoevaluatethe carolinensis. were themarsupial species withabsoluteMRSsslightlyhigherthanthatof 4).Thetwolarger MRS thanotherplantigradesmallmammals(Fig. confidence andpredictionintervals,indicatingasignificantlyhigher regression, thedatumfor edwardii the PGLSmodelsshowedsignificance.Whendatafor the OLSofMRSthesemammalswassignificant,butnone ( as Pagel’s ML lambda,whichwasnotsignificantlydifferent to0 using Blomberg etal.’s (Blomberg etal.,2003) MT:F ratios DISCUSSION P =1), butsignificantlydifferent to1( There were52speciesintheplantigradedatabasethat 0.004 0.571 0.015 0.133 0.001 0.041 Mean random K ) (Blombergetal.,2003)inthedataforbodymass( and The JournalofExperimentalBiology(2014)doi:10.1242/jeb.095737 0.66 11.99 28.17 234.74 FP

3). KZ E. rupestrisweremappedbackontotheOLS

kg (44rodentsandeightmarsupials).Interestingly, Dasyuroides byrnei .7 <0.01 <0.001 <0.001 <0.001 2.875 <0.001 5.688 <0.001 4.589 4.055 4.041 8.036 E. rupestris 0.521 <0.001 <0.001 <0.001 Randomization testprobability( E. edwardii P <0.001). Notsurprisingly, lay aboveboththe95% and thesquirrel 10 M n b K =135 species),all in anyoftheir estimate aswell and E. ruprestris E. rupestris M b Sciurus ), 10 1319 MRS P E. )

The Journal of Experimental Biology Plantigrade mammals<500 E. edwardii closer tothehipjointevolvedinsmallestunguligrademammals elements,shorterfemora,andmuscleinsertionpointslocated (Carrano, 1999).Intermsoffitness,longermetatarsals,moreslender morphological traitsalongthecursorial–graviportalcontinuum we reportherecanprobablybeattributedtothecontinuumof Janis, 1993;JanisandWilhelm, 1993). intypicallarge-bodied herbivorouscursors(Garlandand measured bytheMT:F ratioacceleratedduringtheOligoceneand Mowoe, 2013).However, theratesofevolutioncursorialityas mammals, thatis,thosesmallerthan1 mammals. Thuscursorialityneverevolvedinthemajorityof than 1 non-cursorial. andnon-macropodmarsupials,thatareconsideredtobe obviously different fromothernon-unguligrademammals,suchas example, bothconsideredtobecursorial,donotdisplayMT:F ratios mammals, itwouldseemthatcarnivoresandlagomorphs,for test forinterordinaldifferences withinthenon-unguligrade unguligrade ornon-unguligradedistributions.Althoughwedidnot 1320 RESEARCH ARTICLE for regressionstatistics). regression ofthephylogeneticallyindependent linearcontrasts(seeTable obtained fromaBrownianmotionPGLS model,equivalenttoanOLS (short dashedlines)andprediction(long dashedlines)intervalswere mammals <20 Fig. allometries oftheMT:F ratioandlog Mammals <20 Statistics mass (kg) Table R Intercept Slope R Intercept Slope AIC Pagel’s lambda Pagel’s lambda AIC The negativeslopeoftheallometryforunguligrademammalsthat These dataemphasizethattheMT:F ratioofmammalssmaller 2 2

.Maximum runningspeedasafunctionofbodymassfor80 3. –1 log10 Maximum running speed (km h ) 5. Resultsofvariousregressionmodelsfittedtothesmallmammaldatasetlog 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 kg neverexceeds0.7,whereasitdoesinlarger unguligrade E. edwardii and

kg (n

E. rupestris kg. .10.1 0.01 Marsupials Carnivora Sciurid rodents Non-sciurid rodents E. rupestris The regressionline(solidline)andthe 95%confidence =80), includingalllocomotormodesexceptunguligrade

g (n =52) data, thedatafallneatlyintoeither Body mass(kg) Presbytis () Madoqua kirkii Lagomorpha 10 M b . With theexceptionof

kg (seeLovegroveand 1 10

4 the MT:F ratioinherbivores,butnotcarnivores(Garlandand when therewasadramaticaccelerationintherateofevolution divergence inMT:F ratiosoccurredintheOligoceneandMiocene was notthecaseincarnivores(LovegroveandMowoe,2013).The the fitnesscostsoflossdigitnumbersandfunctionality, which that thefitnessbenefitsofunguligradyinherbivoresfaroutweighed functionality (LovegroveandMowoe,2013).Thetrade-off posits evolutionary trade-off betweenlocomotorperformanceanddigit similar-sized unguligradeandnon-unguligrademammals:the one carnivore, lagomorph, ( (0.490–0.649) includesixrodents,oneofwhichissaltatorial data. Forexample,thespecieswithtop10highestratios seems tobenoconsistenttaxonomictrendinMT:F ratiosinthese despite beingsignificantlydifferent fromzero.Moreover, there of mentionbecausetheslopebest-fitregressionissolow 2004). running speedthanlarge ungulates(Garland,1983a;Lovegrove, speed. Consequently, smallungulates alsohaveahighermaximum because weightbearingwaslessimportantthanmaximumrunning statistics). were obtainedfromanOLSregression model(seeTable confidence (longdashedlines)and prediction (shortdashedlines)intervals g. plantigrade mammals<500 Fig. Dipodomys), thesaltatorialmarsupial We offer oneexplanationforthedifference intheMT:F ratiosof The allometryofthenon-unguligrademammalsishardlyworthy

.Maximum runningspeedasafunctionofbodymass52 4. –1 log10 Maximum running speed (km h ) 0.6 0.8 1.0 1.2 1.4 1.6 0 1.305 0.138 0 .3 .8 1.393 0.167 PGLSML 1.384 0.150 PGLSBrownian 1.436 0.232 OLS − 0.254 0.685 − 77.10 68.44 The JournalofExperimentalBiology(2014)doi:10.1242/jeb.095737 Lepus americanus,onelipotyphlan, Felis nigripes .1011 0.1 0.01 10 MRS (km E. edwardii E. rupestris 1.172 0.052 1 1 − 0.033 0.255 − 65.18 87.32

h The regressionline(solidline)andthe 95% . − Body mass(kg) 1 ) asafunctionoflog Bettongia penicullta Sorex cinereus

4 forregression 10 1.304 0.138 0 0.905 − 0.254 0.328 − 79.10 93.46 body , one , and

The Journal of Experimental Biology landscapes andC was drivenbyCenozoiccoolingandtheappearanceofmore open The evolutionofwidespreadcursoriality, especiallyunguligrady, mammals. running speeds,thatis,micro-cursoriality, hasevolvedinsmall because theyaretheonlyexamplesinwhichdigitigradyandhigh mammalian generalizationsconcerningbodysizeandrunningspeed elephant-shrews aresomewhatenigmaticintermsofthese occurs intheLagomorpha(Garland,1983a;Lovegrove,2001).But an aquaticlifestyle,orexceptionallyfastrunningspeeds,suchas Exceptions occurredwiththeevolutionofbodyarmour, arboriality, (Lovegrove, 2000;Lovegrove,2001;LovegroveandMowoe,2013). constrained fromevolvingtobodysizeslarger than~0.5 ThermalMaximum,mostplantigrademammalswere oflarger, fasterdigitigradecarnivoresfollowingthe 2013). Forexample,theBowtiemodelargued that,followingthe and Haines,2004;Lovegrove,2012b;LovegroveMowoe, (Lovegrove, 2000;Lovegrove,2001;2004;Lovegrove morphology andbodysizeinmammalsduringtheCenozoic hypotheses ontherelationshipbetweenevolutionoflimb mammals smallerthan1 well astheobservationofcomparativelyslowMRSsinplantigrade quantified ortestedstatistically. Nevertheless,theseobservations,as as thoseoflagomorphs,althoughthesedifferences couldnotbe with thoseoflarger digitigradecarnivores,buttheywerenotasfast In absoluteterms,theMRSsofelephant-shrewswerecomparable undoubtedly beattributedtothedigitigradyof for alarger squirrelandamarsupial.Thesefastrunningspeedscan speeds fasterthanthoseofallmammalssmaller1 Elephantulus ratios, namelythoseofthecursorialArtiodactyla>1 <1 double theaveragefor24otherspecies(0.42)ofsmallmammals the limbs.TheMT:F ratiosoftheelephant-shrewsweremorethan as (Lovegrove andMowoe,2013). andcarnivoreswerealsoincreasinginbodysize Janis, 1993;JanisandWilhelm, 1993),despitethefactthatboth RESEARCH ARTICLE digitigrady occurred surprisinglyquicklyin ‘condoylarths’ (e.g. of theK–Pgboundary(O’Leary etal.,2013),theevolutionof the dramaticradiationofmammals withinseveral100,000 plesiomorphic plantigrady(Lovegrove andMowoe,2013).During Artiodactyla, Perissodactyla),the mostcommonconditionremained digitigrady weresynonymouswith severalcrownorders(Carnivora, Jardine etal.,2012).Nevertheless, althoughunguligradyand increased hypsodontyinunguligrademammals(MacFadden, 2000; Smith andLyons, 2011; Lovegrove and Mowoe,2013) digitigrade mammalsshowedbodysizeincreases(Alroy, 1998; 2013). Moreover, duringtheLateCenozoic, bothunguligradeand Lovegrove, 2012b;Secordetal.,2012;LovegroveandMowoe, Wilhelm, 1993;Edwardsetal.,2010;Figueirido etal.,2012; Maximum (ca.55 The evolution of micro-cursoriality inelephant-shrews micro-cursoriality evolutionThe of speeds Maximum running smaller than1 cursors, theycanalsorunfasterthanthemajorityofmammals 0.7. Thuselephant-shrewsarenotonlythesmallestmammalian No othermammalsmallerthan1 quantified bytheirveryhighMT:F ratiosforsuchsmallmammals. Relative toothersimilar-sized mammals,elephant-shrews,such E. ruprestris, displayremarkabledigitigrade-likeadaptationsof kg, butwerecomparabletosomeofthehighestallmammalian

elephant-shrews alsodisplayedmaximumrunning kg.

MYA), especiallyduring theMiocene(Janisand 4 grasslands followingtheEoceneThermal

kg (Lovegrove,2004),supportseveral

kg hasanMT:F ratiothatexceeds Elephantulus

kg, except

years

kg.

kg as 65 splitbetweentheLeptictideaandMacroscelidea, ca. be separatedfromthatoftheAfrotherianoriginor Early Paleocene ancestor. and Leptictidae,orwhetheritwasinheritedfromacommon Early micro-cursoriality wasderivedindependentlyintheMacroscelidea Elephantulus elephant-shrews, forexample 2006) thatisremarkablysimilartotheconditioninmodern Leptiptidium of theMacroscelideatoK–Pgboundary~65 Amphilemuridae andAdapisoridae.Theydatedthebasaldivergence Aphelescidae withinMacroscelidea,togetherwithLouisinidae, et al.,2005b;HookerandRussell,2012).Theysuggestedplacing 5)(Zack many ofwhich,theyargue, arebasalmacroscelideans(Fig. Macroscelidea evolvedfromPaleocene,Holarctic‘’, Hooker andRussell(HookerRussell,2012)argued thatthe Russell, 2012;O’Learyetal.,2013).Baseduponcladisticanalyses, Adapisoridae) (Zacketal.,2005b;Zack2005a;Hookerand ‘condylarths’ (Aphelescidae,Louisinidae,Amphilemuridaeand relationships betweentheMacroscelideaandNorthAmerican shrews iscomplicatedbyuncertaintyinthephylogenetic perhaps theMacroscelidea. Phenacodontidae), ancestrallagomorphs(e.g. evolved forthe first timeintheEocene. 2006). Thusintheleptictidlineage, micro-cursorialityseemstohave non-cursorial hindlimb)compared withEoceneleptictids(Rose, below midshaft…’,indicatingan ancestralsynostosticcondition(i.e. show fusionofthetibiaandfibula atthedistalendonly, ‘…well K–Pg extinctionevent.However, thepostcraniaof would haveevolvedintheEarly Paleocene,verysoonafterthe was aninheritedtraitinthemacroscelidandleptictidlineages, it O’Leary etal.,2013)(Fig. are thesistercladetoMacroscelidea(HookerandRussell,2012; small, insectivoroussaltatorial()and/orcursorialmammals, a Paleoceneoriginofmicro-cursorialityinMacroscelidea. development ofourargument foranEarlyEoceneorperhapseven placement oftheAphelescidae,doesnotdetractfrom origin ofApheliscus out. ThepointwewishtoemphasizehereisthattheEarlyEocene euungulate ‘’ mayhavebeendigitigradecannotberuled cursoriality (Thewissen,1990),sothepossibilitythatancestral Phenacodontidae andDidolodontidae,tendedtowardsdigitigrade 2005b). However, othercloselyrelatedfamilies,suchas and nomorphologicalevidenceofcursorialcapacity(Zacketal., () hadalongdachshund-likebodywithshortlegs 63.3 55.8 capacity (Zacketal.,2005b).The particular, isassociatedwithenhancedparasagittal,cursorial Haplomylus and thecrusofPaleoceneapheliscines 6), specializations ofthefemur, tibiofibula(distalsynostosis;Fig. Artiodactyla). Rarepostcranialskeletonsshowcursorial American ungulatebasaltoEuungulata(=crownPerrisodactylaand phylogeny consider phenomic/genomic phylogenyandZacketal.’s (Zacketal.,2005b) (Hooker andRussell,2012). The dateoftheoldestleptictid, Understanding theevolutionofmicro-cursorialityinelephant- There seemstobeagreementthattheLeptictidae,whichwere In contrast,bothO’Learyetal.’s (O’Learyetal.,2013)combined MYA (Fig. MYA (O’Learyetal.,2013).Interestingly, MYA andthesplitofAphelescidaewithHyopsodontidaeat The JournalofExperimentalBiology(2014)doi:10.1242/jeb.095737 show adegreeoffibiotibularfusion(Rose,1999;Rose, (present study)(Fig. (Zack etal.,2005b).Tibiofibular synostosis, in 5) (O’Learyetal.,2013).Thusifmicro-cursoriality , aswellthequestionablephylogenetic Apheliscus

Rhynchocyon 5). Thehindlimbsof Prodiacodon crustuluam,cannot 6). Thusweneedtoaskwhether (Aphelescidae) tobeaNorth Apheliscus (Zack etal.,2005b)and isdatedat Apheliscus Gomphos) and Prododiacodon

MYA (Fig. 1321 and and 5)

The Journal of Experimental Biology 1322 RESEARCH ARTICLE Nearctic leptictid(Rose,1999), Fig. mammals were body sizes(Alroy, 1998;Alroyetal.,2000),theearliestPaleocene 1998; Yuanqing etal.,2007),someofwhichattainedverylarge reliably datedtoearlierthantheEarlyEocene. the originofmicro-cursorialityinmacroscelidscannotatpresentbe al., 2007).However, withoutpostcranialdataforstemmacroscelids, characteristics offastrunningspeedsandlateralstability(Tabuce et lengthened proximallyanddistallytosubtarsaljoints,are semicircular crestsoftheastragalartrochlea,andcalcaneus (Tabuce etal.,2007).Forexample,thepulley-shapeand (Tabuce etal.,2007).Postcranialremainsindicatemicro-cursoriality (late EarlyEocene,Tunisia), estimatedto haveweighed~13 Apheliscus but alllimbsarescaled tothesamelength. synostosis (fusion) isindicativeofincreasedcursorial capacity. Nottoscale, and Elephantulusrupestris aecn oeeOioeeMiocene Oligocene Eocene Paleocene Prior totheflourishingofPaleocenearchaicmammals(Alroy, The oldestmacroscelidisthoughttobeChambiuskasserinensis Eocene Nearctic

.Digitized outlinesofthesynostosedtibiofibulaanunnamed 6. 05 03 01 0 10 20 30 40 50 60 leptictid (Zack etal.,2005b),anextant de facto dakotensis Leptictis Millions ofyearsago small-bodied forestdwellers.Smallbody (present study).Thedegreeofdistal Adapisoricidae Leptictis dakaotensis Apheliscus Macroscelidinae Rhynchocyoninae Rhynchocyon Rhynchocyon Amphilemuridae Herodotinae Metoldobotinae Apheliscidae Louisinidae (Rose, 2006), (Rose, 1999) Elephantulus Lepticida rupestris

Myohyracinae –

g savannas andC creation ofsub-Saharanaridcorridors,andtheemergence of southwestern AfricadrivenbythearidificationofSaharaand commenced ~11.5 forest dwellers(Rathbun,2009).SpeciationwithinMacroscelidinae 2011) (Fig. Miocene forestfragmentationinducedbyaridification(Smitetal., Macroscelides Smit etal.,2011) 26–43 shrews, Macroscelidinae( et al.,2007).Itisestimatedthatthetwosubfamiliesofelephant- appearance ofmicro-cursorialityinacrownEutherianorder(Tabuce for sizes wereretainedbymacroscelidsintotheEocene,asdiscussed nraei h TFrto(al 6,Fig. increase intheMT:F ratio(Table cursorial specializationsthan 2009; Smitetal.,2011). TheMacroscelidinaedisplaymorederived body sizereduction andmoreopenhabitats(Rathbun, 2009). which theMacroscelidinaerun atgreatspeedalsoevolvedwith 2001). Theestablishmentofa systemofmaintainedtrailsalong Damme andVan Dooren,1999;Blanckenhorn,2000; Lovegrove, distance requirementsandtherisk ofpredation(Garland,1983a;Van demands andsmallerhomeranges, thusreducingdailymovement Mowoe, 2013).Smallerbody sizesrequirelowertotalenergy evolution oftheirsmallersizes(Lovegrove,2001;Lovegrove and constraints onthebodysizesofmacroscelilididsresulting in the larger, faster carnivoresandavianpredatorsmayhaveplacedupper which undoubtedlyalsoelevatedavianpredation.Thepresence of landscapes withlesscanopycoverandshelterprovidedby trees, cursoriality waspresumablypre-adaptiveinnewlyemerging open Africa dockedwithEuropeandAsia(Hedges,2001).Micro- intensified withtheinfluxofmodernCarnivora~30 (<300 The Rhynchocyoninaespeciated~8–10 C. kasserinensis.Thusthemacroscelidsshowfirst g). InthenewopenAfricanlandscapes,predatorypressures MYA fromaforest-adaptedancestor(Douadyetal.,2003; (=crown PerrisodactylaandArtiodactyla)notasmacroscelids. recognized asNorthAmericanungulatesbasaltoEuungulata al., 2011). NotethattheAphelisidae haverecentlybeen follows Tabuce etal.(Tabuce etal.,2007)andSmital.(Smit whereas thatofthetraditionallyrecognizedmacroscelidfamilies Butler (Butler, 1995)andTabuce etal.(Tabuce etal.,2001), Russell, 2012),Zacketal.(Zackal.,2005b;2005a), the basalmacroscelidsfollowsHookerandRussell(Hooker Macroscelidae andtheirleptictidancestors. Fig. The JournalofExperimentalBiology(2014)doi:10.1242/jeb.095737 5). Thefourrecognizedspeciesof

.A workingmodelofthephylogeny 5. ) andRhynchocyoninae( 4 grasslands (Fig.

MYA followingdispersalfromeastAfricato Rhychocyon Elephantulus 5) (Douadyetal.,2003;Rathbun,

1) andsmallerbodysizes Rhynchocyon , suchasaneartwofold

, MYA coincidentwith Petrodromus Rhynchocyon The topologyof

MYA when ), diverged and are

The Journal of Experimental Biology of cursorialityinlarger mammalsduringtheMiocene. perhaps asearlythePaleocene,butlongbeforeproliferation cursoriality evolvedinsmall,ground-dwellingforestmacroscelids cursoriality withintheMacroscelidea.We proposethatmicro- characteristics areplesiomorphic,indicatingalonghistoryofmicro- through dailyheterothermy. Uniquelytoo,bothofthese only cursorialanimalscapableofoffsetting highlocomotorcosts relative tosimilar-sized mammals,andalsobecausetheyarethe their dramaticmorphologicalspecializationsforfastrunningspeeds inelephant-shrewstooffset thecostsofmicro-cursoriality. (Lovegrove, 2012a)andmayhavebeenretainedthroughoutthe demands. Dailytorporisaplesiomorphiccharacteristicinmammals Lovegrove, 2004),whichprofoundlydecreasesdailyenergy Lovegrove etal.,2001b;Mzilikazi2002;and daily torpor(Lovegroveetal.,1999;Lovegrove2001a; metabolic demands,thistrade-off wasoptimizedthroughtheuseof small Macroscelilinae,thatis,thosewiththehighestmass-specific metabolic demandsassociatedwithmicro-cursoriality. However, in locomotor capacitywerebalancedbythecostsofincreased 2012a) presumablycontinueduntilthefitnessbenefitsofenhanced Extant macroscelids gravel. Thetunnel (1.2 down a30 with waterandtinneddogfood. cages providedwithpapertoweling andarefugetube.Theywereprovided were housedatroomtemperaturein afarmbuildingduringcaptivityinrodent after 48 morning ofcaptureandonthefollowing morning,theanimalswerereleased identifications. Aftermeasuringbodyweightsandrunningspeedsonthe same latter study, SmitandMcKechnie,assistedusinthefieldwithspecies analyses atthesamestudysite(Boylesetal.,2012).Two oftheauthors ofthis on thebasisofmorphologicalcharactersthathadbeenverifiedwithgenetic Kamiesberg Mountains(Boyles etal.,2012).Thetwospecieswereidentified are synopticinthisarid,ruggedandhighlyheterogeneousenvironment inthe weretrappedonthefarmNoHeep(30°02 tending towardssupraendothermy( Species July 2011. Thesetwospeciesof using Elliottrapsbaitedwithamixtureofpeanutbutterandrolledoats during 600–1000 RESEARCH ARTICLE Table of muscleperformance(ClarkeandPörtner, 2010).Selectionfor linked withcursorialityandtheproposedtemperaturedependence sister (Tenrecidae andChrysochloridae)arethoughttobe body temperaturesofelephant-shrewsrelativetotheirAfrotherian MATERIALS ANDMETHODS Leptictis dakotensis Rhynchocyon cirnei Rhynchocyon petersi Petrodromus tetradactylus Elephantulus brachyrhyncus Elephantulus rozetti Elephantulus proboscideus Elephantulus rupestris Elephantulus edwardii Prodiacodon tauricinerei Maximum runningspeedswereobtained bytiminganimalsastheyran In conclusion,elephant-shrewsareuniquemammalsintermsof Like othercursorialanimals(Lovegrove,2012b),theelevated 6. Metatarsal:femur(MT:F) ratiosandhabitatsofextantmacroscelidstheirputativecondylarthleptictidancestors

h intheeveningattheirexactplace ofcapture.Theelephantshrews m), 22 m tunnelrunway erectedonaflatsectionofcompacted fine km northeastofKamieskroon,Namaqualand,SouthAfrica,

m wide,1.4 Elephantulus

m high)wasformed by T b MT:F 0.47 0.41 0.55 0.56 0.61 0.62 0.71 1.00 1.07 1.08 .6EryEcn oet Rose,1999 EarlyEoceneforests 0.46 >37.9°C) (sensu ( E. edwardii ′ S, 17°59′ and U Eocene Oligocene forests Closed forest Closed forest Closed forest Savanna, woodland Open, rockydesert Open, desert Open, rockydesert Open, rockydesert Habitat Lovegrove, -shaped iron E. rupestris) E, altitude; T b MRS ofD.merriami timed astheyranoveracarefully measured, uniformdistance.Second,the is notcomparablewithMRSobtained morepreciselywhenanimalswere were releasedfromtraps,usingastopwatch (Kenagy, 1973).Thustheestimate perse.First,theMRSsofthiskangarooratweremeasuredwhen they of thedatum,althoughnotnecessarilypotentiallyhighMRSskangaroo , we discardedfromthedatasetwasthatofsaltatorialMerriam’s correlations…’ ofMRS with morphologicalvariables.Theonlyspecieswhich mammals smaller than1 shrews displaythe mosthighlyderived,digitigrade cursoriallimbsofall we measured. (Rathbun, 2009)probablyallowsthemtoattainfasterMRSsthanthosethat employment oftheirtrailsystemsthattheycreatewithinterritories We suspectthatthelocalknowledgeexploitedbyelephant-shrewsin be measuredinelephant-shrewsundermorenatural,free-rangingconditions. do notprecludetheveryrealpossibilitythathighermaximumspeedsmay were thefastestspeedsthatwemeasuredusingourmethod.Ourestimates run downthefulllengthoftunnel. individual, the‘best’ run,ortheMRS,wastakenasfastest,uninterrupted shade clothtunnelortriedtoclimbthesidesoftunnel.Foreach middle ofthetunnel.Inothercasesanimals‘bounced’ off thesideof varied. Insomecasestheanimalsranwell,butstoppedrunningin calculated fromplaybacksofvideorecordings.Thesuccesstherunswas each occasiontheindividualcompletedthreeruns.Runningspeedswere tested duringtworunningsessionsonconsecutivemornings,and animals wereinducedtostartrunningwithhand-clapping.Eachanimalwas the rockpile,theyrandowntunneltowardscover. Some release, butoncetheyhadorientatedthemselvesandvisuallylocated the rocksatoppositeendoftunnel.Typically, theanimalsfrozeon pile ofrocksattheendtunnelandreleasedanimals~20 linked toaneight-channelJPEG2000digitalvideorecorder. We placeda three pairsofcolourCCDcameras3 green shadecloth.Alongthelengthoflast20 rods (10 data, itis‘…perhapssurprisingthatweareabletoshow that thedataarebiasedinanysystemicwayandthat,given‘noise’ inthe Garland andJanis(GarlandJanis,1993)makethepointthatitisunlikely have beenusedinpastanalyses(seeGarlandandJanis,1993).However, measure runningspeedandhencethequalityofdata that There hasbeenmuchdiscussionaboutthequestionablemethodsused to 2002; Lovegrove,2003;Rojasetal.,2010)(supplementarymaterialTable 1985; RobinsonandRedford,1986;SteudelBeattie,1993;Iriarte-Díaz, mammals obtainedfromtheliterature(Garland,1983a;HayssenandLacy, Our runningspeedandbodymassdatawerecomparedwiththoseofother (16.0 Maximum running speed Maximum running It becameobviousduringthepreliminary dataanalysisthatelephant- We termourdata‘maximumrunningspeed’ onlyinthesensethatthese Dipodomys merriami.We havecausetoquestiontheunusuallyhighvalue km mm diametermildsteel)placedevery3 h − 1 ) forthreeotherspeciesof The JournalofExperimentalBiology(2014)doi:10.1242/jeb.095737 Carrano, 1999 Rose, 2006 Rose, 2006 Evans, 1942;Rathbun,2009 Carrano, 1999;Rathbun,2009 Evans, 1942;Rathbun,2009 Evans, 1942;Rathbun,2009 Carrano, 1999;Rathbun,2009 Present study Present study Reference (31.2

kg. Thusthemammalian runningspeedmodelwith km h − 1 ) wasdoublethatoftheaverage MRS

m apart,eachpairfacingother, Dipodomys

m ofthetunnel,weplaced

m andcoveredwith30% in thedataset. any significant

m from 1323

S1).

The Journal of Experimental Biology we employedstep-wiseanalysesoftheallometricrelationship Lovegrove, 2004).To selectthebestmodelwithwhichtotesthypothesis with masscomparedotherlocomotormodes(Iriarte-Díaz,2002; study, unguligrademammalsdonotshowthesamescalingpatternofMRS comparison withunguligrademammals,although,asweshowinthepresent bounds oftheregressiondata.Indeed,sameargument appliestoa extrapolation ofthedigitigraderegressionmodelswaybeyondlower mostly larger than1 mammals, suchascarnivoresandlagomorphs(rabbitshares),whichare establish whetheraninflectionexistedintherelationshipbetweenlog (Iriarte-Díaz, 2002),weusedpiecewiseregression(Crawley, 2007)to log digitigrade distributions(Lovegrove,2000;Lovegrove,2001). frequency distributionandtheintersectionbetweenplantigrade and body sizelimitapproximatesthe95thpercentileofplantigrade mass pure BrownianmotionPGLSwithbranchlengthtransformationsetto a standardMANCOVA. either plantigrade,lagomorph-like,saltatorialordigitigrade,andthenused al., 2013).We createdafactorvariable(‘foot’)whichcodedthedataas (Garland andIves,2000)asimplementedbyOutomuroetal.(Outomuro phylogenetically transformedfollowingthemethodofGarlandandIves model (PGLS)inwhichthedependentandindependentdatawerefirst mammal dataset,weusedamultivariatephylogeneticgeneralizedlinear 1324 RESEARCH ARTICLE Beattie, 1993;Carrano,1999)(supplementary materialTable mammal obtainedfromtheliterature (GarlandandJanis,1993;Steudel E. edwardii were euthanizedforageneticstudy(Boylesetal.,2012).TheMT:F datafor MT:F ratiosweremeasuredfromanimalsobtainedthesamesitethat Elephantulus equivalent bodysizeirrespectiveoflimbmorphology. TheMRSsof simply thattheMRSofelephant-shrewsexceedsthosemammals here thatthereisonlyonehypothesisreasonablytestable,which which tocompareelephant-shrewswasnotintuitivelyobvious.We argue and log al., 2004).OLSandPGLSmodelswerefittedtolog using theRpackages‘picante’ (Kembeletal.,2010)and‘ape’ (Paradiset (Nunn, 2011), andwithBlomberg etal.’s (Blomberg etal.,2003)K estimated usingPagel’s lambda( calculated fromOLSregressions.Evidenceofphylogeneticsignalwas small mammaldatawereidentifiedusingCook’s distance(Cook,1977) Outliers intheregressionanalysisofconventionalspeciesdata because thisbodysizerangeembracedtheelephant-shrewsizes. termed hereafterthesmallmammaldataset,wereusedforfurtheranalyses observed inthescalingofM Development CoreTeam, 2012).Becausestronginflectionshavebeen material Appendix 2008; Montgelardetal.,Lovegrove,2012a)(seesupplementary 2004; Bininda-Emondsetal.,2007;BradleyMeredith Masuda, 2000;DeBryandSagel,2001;Herronetal.,2004;Steppan of sources(DeWalt etal.,1993; Kruckenhauser etal.,1999;Oshidaand using Mesquiteversion2.74(MaddisonandMaddison,2009)fromavariety MRS inaphylogeneticcontext. Metatarsal:femur ratios smaller than500 mammals, wealsocomputedsimilarPGLSmodelsforthedatamammals model wasdeterminedasthewithlowestAIC. and aPGLSwithPagel’s ML estimationofbranchlengths.Thebestfit material Appendix in theMT:F ratioanalyses wasconstructedusingMesquite(supplementary (see Fig. outlier thathadalarge leverageinfluenceontheunguligrade regressions for thegiraffe (MT:F ratio=1.4)wasexcludedbecause itwasaverylarge For thesmallmammaldataset,twoPGLSregressionswerecalculated:a To determinewhetherMRSisinfluencedbylocomotormodeinthesmall All statisticalanalyseswereconductedusingRversion3.0.1(R A phylogenyofallthespeciesusedincomparisonswascompiled To comparetheMRSofelephant-shrewswiththosesimilar-sized 10 M b 10 using theRpackage‘caper’ (Nunn,2011). M 1). AsdescribedearlierforMRS,a phylogenyofthespeciesused b . Dataformammalssmallerthanandequaltotheinflection and (40–60

g (n E. rupestriswerecomparedwiththosefor135 speciesof

S1). S2). =52), termedhereaftertheplantigradedataset.Thisupper

kg, becausetheirsmallerbodysizeswouldrequire g) cannotbecomparedwiththoseofdigitigrade b with bothMRS(Garland,1983a)andRRS λ) calculatedwiththeRpackage‘caper’ 10 MRS asafunctionof

S2). Thedatum statistic M 10 b MRS λ and M =1, b , photograph takenduringthemeasurementofMT:F ratios. published graphics.Theoutlinesfor data separately, asdescribedabovefortheMRSanalyses. fitted tothecompletedataset,andunguligradenon-unguligrade , marsupialsandamonotreme.OLSPGLSregressionswere mammals, whichinthisdatasetincludedCarnivora,Rodentia,Lagomorpha, mammals (ArtiodactylaandPerrisodactyla),anothertonon-unguligrade obviously dichotomousallometricrelationships:oneuniquetounguligrade the manuscript. B.G.L. andM.O.M.collectedanalyzedthedata,wrotebulkof The authorsdeclarenocompetingfinancialinterests. Levesque forprovidingcommentsonthedraftmanuscript. their farmNoheep,Namaqualand.We aregratefultoGalenRathbunandDanielle and EzitMalan.We thankPieterandVerencia Benadeforpermissiontoworkon Namaqualand: AndrewMcKechnie,BenSmit,DanielleLevesque,KeriLobban identifying elephant-shrewsandmeasuringtheirrunningperformancesin We areverygratefultothoseentertainingvolunteerswhoassistedintrappingand 2005b) andanextant (Rose, 1999),Leptictisdakaotensis The outlinesofthesynostosedtibiofibulaanunnamedNearcticleptictid http://jeb.biologists.org/lookup/suppl/doi:10.1242/jeb.095737/-/DC1 Supplementary materialavailableonlineat KwaZulu-Natal andtheNationalResearchFoundation. This studywasfinancedbyincentivegrantstoB.G.L.fromtheUniversityof rde,R . uih .D,Rgr,D . ilr .R,Egto,M .and M.D. Engstrom, J.R., Miller, D.S., Rogers, N.D., Durish, R.D., Bradley, A.E. McKechnie, and C.L. , B., Smit, J.G., Boyles, iid-mns .R . adlo . oe,K . aPe,R .E,Bc,R. Beck, R.D.E., MacPhee, K.E., Jones, M., Cardillo, O.R.P., Bininda-Emonds, lo,J. Alroy, Author contributions Competing interests Acknowledgements Tibiofibula outlines References Supplementary material Funding ok R.D. Cook, M. Waterhouse, and J.A. Wells, J.G., Lefevre, A.R. S.P., Ives, Blomberg, and Jr T., Garland, S.P., Blomberg, lre .adPrnr H.-O. Pörtner, and A. Clarke, M.T. Carrano, P. M. Butler, W. U. Blanckenhorn, J.C. Zachos, and P. L. Koch, J., Alroy, oay .J,Ctels . aa,J,Srne,M .adSahp,M.J. Stanhope, and M.S. Springer, J., Raman, F., Catzeflis, C.J., Douady, R.M. Sagel, and R.W. J.V. DeBry, Planz, and E.G. Zimmerman, T. S., DeWalt, M.J. Crawley, Edwards, hitn .A,Cuis .B,Dvl,M . o,D . rcltn .P etal.; R.P. Freckleton, D.L., Fox, M.R., Duvall, A.B., Cousins, P. A., Christin, iptik C. W. Kilpatrick, Physiol. patterns intwosyntopicelephantshrewspeciesduringwinter. 61, 382-391. Independent contrastsandPGLSregressionestimatorsareequivalent. small? A. Purvis, and J.L. Gittleman, R.A., Vos, The delayedriseofpresent-daymammals. S.A., Price, R., Grenyer, M. D., American mammalianevolution. American fossilmammals. Technometrics of endothermy. from mitochondrialcytochrome-bsequences. signal incomparativedata:behavioraltraitsaremorelabile. determining locomotorhabitinmammalsanddinosaurs. Acad. Sci.USA diversification ofthemammalianorderMacroscelidea (elephantshrews). The Saharaasavicariantagent,and the roleofMioceneclimaticevents,in from thenuclear-encodedgeneIRBP. phylogeny ofspeciesthe Initial plotsoftherelationshipbetweenMT:F ratioand log Q. Rev. Biol. (1998). Cope’s ruleandthe dynamicsofbodymassevolutioninNorth 161A .J,Obre .P,Srmeg .A . mt,S . od W. J., Bond, S.A., Smith, C.A.E., Strömberg, C.P., Osborne, E. J., (1995). FossilMacroscelidea. The JournalofExperimentalBiology(2014)doi:10.1242/jeb.095737 74, 352-362. (2007). (1999). What,ifanything,isacursor?Categoriesversuscontinuafor (1977). Detectionofinfluentialobservationinlinearregression. , 89-94. 19, 15-18. Biol. Rev. Camb.Philos.Soc. 100 (2007). Toward amolecularphylogenyfor , 8325-8330. 75, 385-407. The RBook.Sussex,England:JohnWiley&SonsLtd. (2000). Theevolutionofbodysize:whatkeepsorganisms Rhynchocyon Science (2010). Temperature, metabolicpowerandtheevolution boylii (2001). PhylogenyofRodentia(Mammalia) inferred Paleobiology 280 and Mol. Phylogenet.Evol. , 731-734. (Rose, 2006),Apheliscus Mammal Review truei (2000). GlobalclimatechangeandNorth E. rupestris (Rose, 1999)weredigitizedfrom Nature 85, 703-727. J. Mammalogy 26, 259-288. groups ofthegenus 446 (2003). Testing forphylogenetic , 507-512. J. Zool.247 (1993). Mitochondrial-DNA were digitizedfroma 25, 3-14. (2012). Bodytemperature Evolution 88, 1146-1159. 19, 290-301. Peromyscus Comp. Biochem. , 29-42. 10 Peromyscus 57, 717-745. (Zack etal., M b Proc. Natl. : evidence Syst. Biol. revealed (2012). (2007). (2003). . J.

The Journal of Experimental Biology oerv,B . aa,J n ern M.R. Perrin, and J. Raman, B.G., Lovegrove, adsn .P n adsn D.R. Maddison, and W. P. Maddison, M.R. Perrin, and J. Raman, B.G., Lovegrove, eeih .W,Wsemn .adSrne,M.S. Springer, and M. Westerman, R.W., Meredith, aFde,B.J. MacFadden, oerv,B . ae,M .adRxug,L. Roxburgh, and M.J. Lawes, B.G., Lovegrove, M.O. Mowoe, and B.G. Lovegrove, B. G. Lovegrove, G.J. Kenagy, RESEARCH ARTICLE otead . ot,E,Anl .adMthe C.A. Matthee, and V. Arnal, E., Forty, C., Montgelard, rcehue,L,Pnkr .adAnl,W. Arnold, and W. Pinsker, L., Kruckenhauser, oerv,B.G. Lovegrove, B.G. Lovegrove, B.G. Lovegrove, B.G. Lovegrove, oerv,B .adHie,L. Haines, and B.G. Lovegrove, B.G. Lovegrove, vn,F. G. Evans, iuiio . ai,C . ée-lrs .A,D ez,M n amvs,P. Palmqvist, and M. DeRenzi, J.A., Pérez-Claros, C.M., Janis, B., Figueirido, rat-íz J. Iriarte-Díaz, okr .J n usl,D.E. Russell, and J. Hooker, M. Hildebrand, S.B. Hedges, R.C. Lacy, and V. Hayssen, aln,T. Garland, ai,C.M. Janis, ern .D,Cso,T .adPrisn C.L. Parkinson, and T. A. Castoe, M.D., Herron, aln,T,J n vs A.R. Ives, and Jr T., Garland, T. Garland, ai,C .adWlem P. D. Wilhelm, and C.M. Janis, aln,T n ai,C.M. Janis, and T. Garland, Kembel, adn,P . ai,C . any .adBno,M.J. Benton, and S. Sahney, C.M., Janis, P. E., Jardine, oe,K .adPri,A. Purvis, and K.E. Jones, Physiol. B shrews, 33-59. Reconstructing ancientdietsandterrestrial communities. Comp. Physiol.B shrews (Macroscelidea: Macroscelides proboscideus pleisiomorphic dailytorporinmammals:theround-earedelephantshrew nucleotides inmitochondrial, exonandintronfragments. relationships among Rodentiaandthephylogeneticeffect ofremovingfast-evolving based onnuclearDNA sequences. timescale forthelivinggeneraofkangaroos andkin(Macropodiformes:Marsupialia) Evolutionary Analysis,Version 1.12 26, 1317-1329. responses toCenozoicclimatechangeinNorthAmericanmammals. Nat. 49-56. molecular evidenceforadiphyleticoriginofsociality. heteromyid rodentcommunity. C.O. Webb, and phylogenies andecology. S.P. Blomberg, D. D., sizes: evolutionaryhistory, form,andfunction. mammals: aslow–fastmetaboliccontinuum. Cenozoic supraendothermicpulses. C4 GrassesConsortium (Macroscelididae). evolutionary andecosystemscience. metabolic rateinplacentalmammals. constraints oflargebodysize. plesiomorphic–apomorphic continuum. Linn. Soc. (Macroscelidea, Mammalia),theirrelationshipsandnorthEuropeandiversity. Proc. Natl.Acad.Sci.USA (2012). Cenozoicclimatechangeinfluencesmammalianevolutionarydynamics. terrestrial mammals. Sons. 1015-1030. ofHolarcticgroundsquirrels( USA Sci. 741-754. differences intheallometryofBMRandbodymass. Am. Nat. confidence intervalsforregressionequationsinphylogeneticcomparativemethods. terrestrial mammals. vegetation, andtectonicevents. running speedincursorialmammals? mammals. Am.Nat. the Tertiary? Danceswithwolfavatars. Comparative evidencefrombats. Glires. Palaeogeogr. Palaeoclimatol.Palaeoecol. concordant patternsofearlyoriginhypsodontyinGreatPlainsungulatesand 156 .W,Cwn .D,Hlu,M . onel .K,Mro,H,Ackerly, H., Morlon, W. K., Cornwell, M.R., Helmus, P. D., Cowan, S. W., , 201-219. Elephantulus 155 98, 1-2. (1983b). Scalingtheecologicalcostoftransporttobodymassinterrestrial (1983a). Therelationbetweenmaximalrunningspeedandbodymassin 171 164 (1993). Tertiary mammalevolutioninthecontextofchangingclimates, (2002). Differential scaling of locomotorperformanceinsmallandlarge , 346-364. (2001). :platetectonicsmeetsgenomics. (1942). Theosteologyandrelationshipsoftheelephantshrews (1973). Dailyandseasonalpatternsofactivityenergeticsina (1974). , 1-10. , 856-936. (2003). Theinfluenceofclimateonthebasalmetabolicratesmall (2000). Thezoogeographyofmammalianbasalmetabolicrate. (2012a). TheevolutionofendothermyinCenozoicmammals:a 171 (2004). Locomotormode,maximumrunningspeed,andbasal (2001). Theevolutionofbodyarmorinmammals:plantigrade (2012b). Theevolutionofmammalianbodytemperature:the Bull. Am.Mus.Nat.Hist. (2000). CenozoicmammalianherbivoresfromtheAmericas: 121 J. Exp.Biol. J. Zool. , 11-21. Analysis ofVertebrate Structure spp. (Macroscelidea):dailytorpororhibernation? , 571-587. Bioinformatics Elephantulus 109 199 (Macroscelidea). (1985). Basalmetabolicratesinmammals:taxonomic (2010). TheoriginsofC (1993). Doesmetatarsal/femurratiopredictmaximal Evolution Ecology , 722-727. (1993). Were theremammalian pursuitpredatorsin Annu. Rev. Ecol.Syst. , 157-170. (1997). Anoptimumbodysizeformammals? 205 Funct. Ecol. (2004). Theevolutionofplacentalmammalbody (2000). Usingthepasttopredictpresent: . Available athttp://mesquiteproject.org. Aust. J.Zool. J. Comp.Physiol.B , 2897-2908. Science Physiol. Biochem.Zool. (2013). Theevolutionofmammalbodysizes: J. Zool.229 J. Mamm.Evol. Biol. Rev. Camb.Philos.Soc. 54, 1201-1219. 55, 1464-1473. spp.) inresponsetofooddeprivation. 26, 1463-1464. Spermophilus (2009). (2012). EarlyPalaeogeneLouisinidae 80, 80-125. (2010). Picante:Rtoolsforintegrating J. Comp.Physiol.B 328 Oecologia 11 J. Comp.Physiol.B (1999). Marmotphylogenyrevisited: , 751-756. 56, 395-410. (2001a). Heterothermyinelephant 365-366 , 133-151. , 587-591. Mesquite: A (2004). Sciuridphylogenyandthe (2001b). Dailytorporinelephant Comp. Biochem.Physiol. J. Zoolog.Syst.Evol.Res. . NewYork, NY: JohnWiley& ). 24, 467-500. 1 BMC Evol.Biol. , 103-125. 182 138 Mol. Phylogenet.Evol. , 1-10. Annu. Rev. Ecol.Syst. 4 (2008). A phylogenyand (1999). Confirmationof , 579-589. , 13-27. grasslands: integrating (2012). Gritnotgrass: 77, 916-928. ModularSystemfor (2008). Suprafamilial 173 169 Proc. Natl.Acad. , 87-112. 87, 128-162. , 453-460. J. Evol.Biol. 8 , 321. J. Comp. Zool. J. 81A Am. 31, 37, 31, J. , okae,L.C. Rookmaker, ahu,G.B. Rathbun, G.B. Rathbun, R DevelopmentCoreTeam F. Johansson, and D.C. Adams, D., Outomuro, oe K.D. Rose, A.D.,Kӧ Rojas, K.H. Redford, and J.G. Robinson, K. Strimmer, and J. Claude, E., Paradis, Secord, R. Masuda, and T. Oshida, Giannini, A., Giallombardo, T. J., Gaudin, J.J., Flynn, J.I., Bloch, M.A., O’Leary, oe K.D. Rose, un C. Nunn, D.O. Ribble, and B.G. Lovegrove, N., Mzilikazi, zlkz,N n oerv,B.G. Lovegrove, and N. Mzilikazi, mt,F .adLos S.K. Lyons, and F. A. Smith, and F. Yang, M., Ferguson-Smith, P. C.M., O’Brien, B.J., vanVuuren, H.A., Smit, aue . ofat . ofat .E,Mhob,M n agr J. Jaeger, and M. Mahboubi, P. E., Coiffait, B., Coiffait, R., Tabuce, J. Beattie, and K. Steudel, J. Anderson, and R. Adkins, S., Steppan, ti,B .adCsns A. Casinos, and B.R. Stein, Cleven, S.B., Hedges, W., deJong, O., Madsen, V. G., Waddell, M.J., Stanhope, Amrine, V. G., Waddell, W. W., deJong, O., Madsen, G.C., Cleven, M.S., Springer, alr .R,ShitNesn .adRa,J.L. Raab, and K. Schmidt-Nielsen, C.R., Taylor, Mahboubi, J.L., Hartenberger, M., Bensalah, M., Adaci, L., Marivaux, R., Tabuce, hwse,J.G.M. Thewissen, unig . i,M,Xjn .adCunu,L. Chuankui, and N. Xijun, M., Jin, W., Yuanqing, T. J.M.(1999).Absoluteversusperunitbody R.andVan Dooren, Van Damme, ak .P,Pnrt .A,Kas,D .adMa,M.C. Maas, and D.W. Krause, T. A., Penkrot, S.P., Zack, ak .P,Pnrt .A,Boh .I n oe K.D. Rose, and J.I. Bloch, T. A., Penkrot, S.P., Zack, 1364. Afrotheria: Macroscelidea)taxonomyandecology? Paul Parey. http://www.R-project.org/. Computing evolution inRlanguage. ornamented-winged damselflies. gene sequences. species ofthegenus implications forbehaviorandrelationships. A. Balkema. affect maximumrunningspeed oftwosmallmarsupials? Neotropical forestmammals. Messel. PalaeontographicaAbteilungA Leptictida), withapreliminarycomparisonto etal. J. Meng, Z.X., 662-667. Luo, B.P., Kraatz, placental mammalancestorandthepost-K-Pgradiationofplacentals. S.L., Goldberg, N. P., Biology. Oecologia during arousalinfree-rangingrockelephantshrews, 296. shrews, Science J. Krigbaum, and drivenbyclimatechangeinthePaleocene-Eocenethermalmaximum. F. A. McInerney, M.J., Kraus, 497-501. 217 242 R. Soc.B macroecological lookatmammalianbodysizeoverspaceandtime. Macroscelididae). T. J. Robinson, elephant-shrews. ofMacroscelidea(Mammalia)fromtheEoceneAlgeria:apossibleorigin for Syst. Biol. estimates ofrapidradiationsinmuroidrodentsbasedonmultiplenucleargenes. Acad. Sci.USA M.S. andforaneworderofendemicAfricaninsectivoremammals. Springer, and D. Kao, G. C., M.J. Stanhope, phylogenetic tree. and H. M. of runningtobodysizeinmammals. J. 1166. Jaeger, and P. from NorthAfricareinforcethemolecularAfrotheriaclade. Tafforeau, F., Mebrouk, M., (Mammalia, Condylarthra). mammal radiationinChina. ‘hyopsodontids’ 347-352. length speedofpreyasanestimator vulnerabilitytopredation. and thephylogenyof‘hyopsodontids’. apheliscine ‘condylarth’ mammalfromthe latePaleoceneofMontanaandAlberta , 55-63. , 185-192. . lc,J . hse,S .B,Byr .M,Wo,A . ig S.L., , A.R., Wood, D.M., Boyer, S.G.B., Chester, J.I., Bloch, R., Chicago, IL:UniversityofChicagoPress. 335 Elephantulus myurus (2011). 366 53, 533-553. (2006). ThepostcranialskeletonofearlyOligocene 133 . Vienna, Austria:RFoundationforStatisticalComputing.Available at (1999). PostcranialskeletonofEoceneLeptictidae(Mammalia),andits , 959-962. The JournalofExperimentalBiology(2014)doi:10.1242/jeb.095737 , 2364-2378. (1979). , 307-314. toelephantshrews andaHolarcticoriginofAfrotheria. 95, 9967-9972. (2009). Whyistherediscordantdiversityinsengi(Mammalia: (1989). (2011). Phylogeneticrelationshipsofelephant-shrews(Afrotheria, rtner, G.andGeiser, F. J. Vertebr. Paleontol. Zoolog. Sci. The ComparativeApproachinEvolutionaryAnthropologyand J. Zool.(Lond.) Nature (1990). EvolutionofPaleoceneandEocenePhenacodontidae The SocialStructureandEcologyofElephant-Shrews Sciurus Bioinformatics The ZoologicalExplorationofSouthernAfrica 388 Univ. MichiganPap.Paleontol. (1993). Scalingofcursorialityinmammals. Geological Journal (2012). (2000). Phylogenyandzoogeographyofsixsquirrel Am. Nat. (1997). Whatisacursorialmammal? , 61-64. : ayear-longstudy. 17, 405-409. (Mammalia, Rodentia),inferredfromcytochromeB Evol. Biol. 284 (1997). EndemicAfricanmammalsshakethe (2004). Dailytorporinfree-rangingrockelephant (1986). Bodysize,diet,andpopulationdensityof (1998). Molecularevidenceformultipleoriginsof (2011). Howbigshouldamammalbe?A R: A LanguageandEnvironmentforStatistical Am. J.Physiol. , 133-143. Acta Palaeontol.Pol. 128 21, 535-546. 20, 289-290. 278 (2004). APE:analysesofphylogeneticsand , 665-680. J. Vertebr. Paleontol. 40, 300-309. , 37-56. (2004). Phylogenyanddivergence-date Leptictidium 42, 415-430. (2010). Doimplantedtransmitters (2002). Exogenouspassiveheating (2013). Evolutionofwingshapein (2007). MajoreventsofPaleogene Physiol. Biochem.Zool. Afr. J.Ecol. (1970). Scalingofenergeticcost 219 (2007). EarlyTertiary mammals (2012). Evolutionoftheearliest , 1104-1107. from themiddleEoceneof 29, 1-120. Proc. Biol.Sci. 50, 809-830. J. Mammal. Elephantulus myurus (2005b). Affinities of 47, 1-13. Leptictis 19, 355-372. Anim. Behav. (2005a). A new J. Zool.(Lond.) (2001). A new Philos. Trans. Science . Rotterdam: (2013). The Nature (Mammalia: J. Morphol. 274 Proc. Natl. 91, 1360- 77, 285- . Berlin: , 1159- 1325 339 434 57, , . ,

The Journal of Experimental Biology