rsosroyalsocietypublishingorg

ResearchCite this article Castro MC Goin FJOrtiz-Jaureguizar E Vieytes EC Tsukui KRamezani J Batezelli A Marsola JCA LangerMC 2018 A Late Cretaceous mammal fromBrazil and the first radioisotopic age for theBauru Group R Soc open sci 5 180482httpdxdoiorg101098rsos180482

Received 23 March 2018Accepted 23 April 2018

Subject CategoryBiology (whole organism)

Subject Areaspalaeontologytaxonomy and systematics

KeywordsTribosphenida enamel reduction Bauru BasinSouth America U-Pb geochronology Mesozoic

Author for correspondenceMariela C Castroe-mail marielaccastroyahoocombr

daggerPresent address Departamento de CiecircnciasBioloacutegicas IBiotec Universidade Federal deGoiaacutes Regional Catalatildeo Catalatildeo-GO Brazil

Electronic supplementary material is availableonline at httpsdxdoiorg106084m9figsharec4097852

A Late Cretaceous mammalfrom Brazil and the firstradioisotopic age forthe Bauru GroupMariela C Castro1dagger Francisco J Goin25 Edgardo

Ortiz-Jaureguizar35 E Carolina Vieytes45 Kaori

Tsukui6 Jahandar Ramezani6 Alessandro Batezelli7

Juacutelio C A Marsola1 and Max C Langer11Laboratoacuterio de Paleontologia FFCLRP Universidade de Satildeo Paulo Ribeiratildeo Preto-SP14040-901 Brazil2Divisioacuten Paleontologiacutea Vertebrados Facultad de Ciencias Naturales y Museo 3LASBEFacultad de Ciencias Naturales y Museo and 4Divisioacuten Zoologiacutea Vertebrados Facultadde Ciencias Naturales y Museo Universidad Nacional de La Plata Paseo del BosqueSNdeg B1900FWA La Plata Argentina5CONICET Buenos Aires Argentina6Department of Earth Atmospheric and Planetary Sciences Massachusetts Instituteof Technology Cambridge MA 02139 USA7Department of Geology and Natural Resources IG Universidade Estadual deCampinas Campinas-SP Brazil

MCC 0000-0002-5734-9639

In the last three decades records of tribosphenidan mammalsfrom India continental Africa Madagascar and South Americahave challenged the notion of a strictly Laurasian distributionof the group during the Cretaceous Here we describe a lowerpremolar from the Late Cretaceous Adamantina Formation SatildeoPaulo State Brazil It differs from all known fossil mammalsexcept for a putative eutherian from the same geologic unityand Deccanolestes hislopi from the Maastrichtian of India Theincompleteness of the material precludes narrowing downits taxonomic attribution further than Tribosphenida butit is larger than most coeval mammals and shows a thinlayer of parallel crystallite enamel The new taxon helpsfilling two major gaps in the fossil record the paucity ofMesozoic mammals in more northern parts of South Americaand of tribosphenidans in the Cretaceous of that continentIn addition high-precision U-Pb geochronology provided apost-Turonian maximal age (le878 Ma) for the type stratumwhich is overlain by the dinosaur-bearing Mariacutelia Formation

2018 The Authors Published by the Royal Society under the terms of the Creative CommonsAttribution License httpcreativecommonsorglicensesby40 which permits unrestricteduse provided the original author and source are credited

on July 15 2018httprsosroyalsocietypublishingorgDownloaded from

2

rsosroyalsocietypublishingorgRSocopensci5180482

constraining the age of the Adamantina Formation at the site to late Coniacianndashlate MaastrichtianThis represents the first radioisotopic age for the Bauru Group a key stratigraphic unit for the studyof Cretaceous tetrapods in Gondwana

1 IntroductionThe mammalian clade Tribosphenida [1] (=Boreosphenida [2]) includes eutherians and metatherianswith records as old as the Early Cretaceous (Berriasian) in the Northern Hemisphere [3] The groupis characterized by the presence of a prominent protocone occluding on a basined talonid combiningshearing and grinding in a single chewing stroke [45] In the last three decades records from India [6ndash8]Madagascar [910] continental Africa [11] and South America [12ndash14] have challenged the notion that thegroup was restricted to Laurasia during the Cretaceous Unfortunately despite few exceptions like theIndian Deccanolestes hislopi [1516] these occurrences are fragmentary andor scarce hampering moreprecise phylogenetic placements

Here we describe a new mammal species based on an isolated tooth collected during 2015 (inaccordance with Brazilian laws) from the Late Cretaceous Adamantina Formation in the area of GeneralSalgado Satildeo Paulo State Brazil (figure 1 electronic supplementary material figures S4ndashS7) We alsoprovide the first radioisotopic age for Bauru Group deposits which is an important window to thebiodiversity of Western Gondwana in the Late Cretaceous especially due to its tetrapod records

2 Material and methods21 U-Pb geochronologyZircons were isolated from an approximately 1 kg sample of the bed where the mammalian remain wascollected using standard crushing as well as magnetic and density separation techniques Preliminaryanalyses of selected zircons (z1 to z17) from this sample by the high-precision chemical-abrasion isotope-dilution thermal-ionization mass spectrometry (CA-ID-TIMS) technique at MIT revealed a wide rangeof detrital zircon ages from Neoproterozoic to ca 878 Ma To isolate the youngest population of zirconsthat would most closely approximate the depositional age 100 additional zircon grains were placed in agrain mount and analysed by laser-ablation inductively coupled plasma mass spectrometry (LA-ICP-MS)at the University of Kansas following the procedures described in Cioffi et al [18] Based on the LA-ICP-MS results seven youngest dated zircons were removed from the mount and subsequently analysed bythe CA-ID-TIMS method at MIT

The CA-ID-TIMS analyses were carried out at the MIT Isotope Lab following the analyticalprocedures described in Ramezani et al [19] The selected grains were pre-treated by a chemical-abrasionmethod modified after Mattinson [20] involving thermal annealing at 900degC for 60 h and leachingin concentrated HF at 210degC for 12 h in order to mitigate the effects of radiation-induced Pb loss inzircon The EARTHTIME mixed 205Pbndash233Undash235U (ET535) tracer [2122] was used in the analyses andisotopic measurements were made either on the VG Sector 54 or on Isotopx X62 multi-collector massspectrometers equipped with Daly photomultiplier ion-counting systems at MIT Reduction of massspectrometric data as well as calculation of dates and propagation of uncertainties was done usingthe Tripoli and U-Pb_Redux software and associated algorithms [2324]

Complete U-Pb data are given in electronic supplementary material tables S1 (ICP-MS) and S2(ID-TIMS) and the analyses used in age calculation are illustrated in the date distribution plot ofelectronic supplementary material figure S8 All tabulated U-Pb data are reported with 2σ analytical(internal) uncertainties whereas the maximum depositional age stated in the text includes all sources ofuncertainty The youngest measured ID-TIMS 206Pb238U dates are considered a maximum constrainton the age of deposition of the corresponding bed and its associated in situ fauna

22 Enamel analysesThe labial face of the tooth and the tip of the main cusp were prepared for enamel study accordingto the protocol described in Flynn amp Wahlert [25] Quantitative microanalyses on the uncoated premolarwere performed by energy-dispersive X-ray spectrometry (EDS) using a JEOL 6510 LV scanning electronmicroscope (SEM) coupled to Oxford Instrument X-Max EDS detector operating at 30 kV

on July 15 2018httprsosroyalsocietypublishingorgDownloaded from

3

rsosroyalsocietypublishingorgRSocopensci5180482

(a)

(b)

(c)

0 50scale

legend

100 km

Marilia Formation

topsoil

legend

very fine- to fine grained sandstone

calcrete (carbonate palaeosol)tabular fine- to medium grained

sandstone

Palaeosol 1P1

P2

TSF

AF1

AF2

AF2

P2

TSF

AF1

TSF

P1

0C S G

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

(m)

Palaeosol 2

tabular sandstone facies

Aeolian facies 1

Aeolian facies 2low-angle cross stratification

subtle parallel lamination

root marks

carbonate nodules

blocky structure

bioturbation

carbonate cimentation

wave ripples

flaser bedding

Gondwanasuchus scabrosus

Caipirasuchus stenognathus

Aplestosuchus sordidus

Baurusuchus salgadoensis

Baurusuchus albertoi

Brasilestes stardusti

snake

detail of the lenticular sandstones (1 m)

Bauru GroupLate Cretaceous

Uberaba Formation

Adamantina Formation

Serra Geral Formation (Early Cretaceous)vGO - Goiaacutes StateMG - Minas Gerais StateMS - Mato Grosso do Sul StateSP - Satildeo Paulo StatePR - Paranaacute State

State boundary

fossil occurrenceat buriti farm

50deg28cent25le

20deg3

3cent51

le24

deg34cent

10le

50deg27cent53le

Araccedilatuba Formation

Caiuaacute Group (Early Cretaceous)

Figure 1 Geological settings (a) Surface distribution map of Bauru Basin stratigraphic units (modified from ref [17] using CorelDRAW2017 wwwcoreldrawcom) (b) Google Earth image (2016) marking fossil occurrences at Buriti Farm Satildeo Paulo State Brazil (c)Stratigraphic columnof the studied site AndashB transect of the stratigraphic column BaAsBaurusuchus albertoi andAplestosuchus sordidustype-locality Bs (within the AndashB transect) B stardusti type-locality Bs Baurusuchus salgadoensis type-locality Cs Caipirasuchusstenognathus type-locality Gs Gondwanasuchus scabrosus type-locality

3 Results31 U-Pb geochronologyThe LA-ICP-MS 206Pb238U dates have an average 2σ uncertainty of 26 and range fromapproximately 33 Ga to 852 plusmn 27 Ma indicating a highly mixed detrital zircon population (seeelectronic supplementary material) The ID-TIMS 206Pb238U dates are characterized by 2σ uncertaintiesof the order of 02 and range from 6204 plusmn 16 Ma to 87782 plusmn 0062 Ma The youngest two high-precisionID-TIMS analyses (z17 and z18) overlap within uncertainty (and with the youngest five LA-ICP-MSdates) probably representing the youngest zircons that exist in the sample (electronic supplementarymaterial figure S8) The data are not sufficient to calculate a weighted mean date therefore the youngestID-TIMS date of 87782 plusmn 0062 Ma (plusmn012 Ma including total propagated uncertainties) best representsthe maximum age of deposition of the fossil-bearing bed

32 Systematic palaeontologyTribosphenida McKenna 1975

Brasilestes stardusti gen et sp nov(figure 2 electronic supplementary material figures S1ndashS3)

Etymology The genus name combines the words Brasil and ληστ ησ (Greek for robber) Species epithetafter the fictional character Ziggy Stardust honouring the deceased artist David BowieHolotype LPRPUSP 0751 isolated lower right premolar probably p3 or p4 (figure 2)Type locality and horizon Buriti Farm General Salgado Satildeo Paulo Brazil (20deg33rsquo56primeprime S 50deg28rsquo09primeprime W)Adamantina Formation Bauru Group Bauru Basin (figure 1)

on July 15 2018httprsosroyalsocietypublishingorgDownloaded from

4

rsosroyalsocietypublishingorgRSocopensci5180482

05 mm

05 mm

05 mm

En

EDJ

10 microm

(a)

(c) (d )

(b)

(e) ( f )

Figure 2 Brasilestes stardusti gen et sp nov (LPRPUSP 0751 holotype) Lower right premolar (probably p3 or p4) in different views(a) labial (b) occlusal (c) lingual (d) posterior and (e) anterior (f ) thin enamel layer with no prisms detected on the labial side of thetooth En enamel EDJ enamelndashdentine junction

Age High-precision U-Pb geochronology of detrital zircons from the fossil-bearing bed yielded amaximum depositional age of 8778 plusmn 012 Ma (see electronic supplementary material)Diagnosis Brasilestes stardusti differs from other described mammals by a unique suite of traits of itstwo-rooted premolar strongly asymmetric profile outline single and slightly recumbent main cusp withan anterior crest and a flat posterior surface small anterior cuspule flat talonid perpendicular to theposterior surface of the main cusp corresponding to one-third of the total length of the crown extremelythin enamel layer (less than 20 microm)Description LPRPUSP 0751 is large (max crown length = 35 mm max crown width = 18 mm)compared to most Mesozoic mammal premolars Although broken at the rootndashcrown junction the toothis clearly two-rooted and its outline is strongly asymmetric in the lateral view It has a single slightlyposteriorly recumbent main cusp with an anterior crest and a flat posterior surface perpendicular to arelatively wide talonid Although the apex is worn the anterior and posterior borders of the main cuspsuggest it was pointed The anterior crest ends at the base of the crown in a barely differentiated verysmall cuspule that defines the anterior-most end of the crown A somewhat concave vertical facet isseen in the mesio-lingual surface of the main cusp No posterior cingulum is present The labial surface

on July 15 2018httprsosroyalsocietypublishingorgDownloaded from

5

rsosroyalsocietypublishingorgRSocopensci5180482

is more salient and more convex at the contact between the main cusp and the talonid whereas thelingual surface is flatter The talonid corresponds to approximately one-third of the total length of thecrown It has a nearly triangular outline in the occlusal view narrowing towards the posterior end andbears no cusps Its surface is flat with signs of wear A notch marks a tiny cuspule at the lingual borderhalfway between the posterior end of the tooth and the posterior surface of the main cusp The preservedmorphology suggests there were well-developed roots the anterior slightly larger than the posterior

Enamel microstructure and EDS analyses (figure 1 figures S1ndashS3) the enamel layer is extremely thin(less than 20 microm thick) it does not homogeneously cover the surface of the crown being absent in someareas (mostly at the tip and the anterior slope of the main cusp) As the enamel shows numerous crackswe regard these absences as due to post-mortem processes (probably spalling of the tooth exposed tosub-aerial weathering andor desiccation [26]) Processes of enamel loss or thinning due to digestionby predators and abrasion during transport were discarded (see below) Regarding its microstructurethe enamel is prismless being composed of crystals that are displayed parallel to each other andlie perpendicular to the enamelndashdentine junction (EDJ) The columns that characterize the SynapsidColumnar Enamel were not observed

The results of the elemental analyses (electronic supplementary material figure S3) are similar tothose of previous studies on fossil bones and teeth from several fossil sites [2728] The major elementsinclude oxygen (356 wt) calcium (296 wt) and phosphorus (102 wt) These are components ofhydroxyapatite the main mineral that forms dental enamel and dentine and may correspond to theoriginal composition of the tooth The calcium could also have been originated from paedogeneticprecipitation of calcium carbonate during the calcrete formation (see above) Likewise this paedogenicevent could be responsible for the great amount of carbon in the sample (199 wt) Among the minorelements magnesium (05 wt) sodium (04 wt) potassium (02 wt) and chlorine (01 wt) may alsocompose the dental tissues [29] Silicon (11 wt) and iron (05 wt) as well as the aluminium (097wt) in lesser degree are more concentrated in the striations on the surface where the sediment isaccumulated and might be related to oxidation processes during the current paedogenesis

4 Discussion41 Age of the fossil bedThe Adamantina and correlated formations can be conservatively considered younger than theunderlying Valanginian-Barremian Serra Geral basalts (part of the Paranaacute-Etendeka Volcanic Province40Ar-39Ar dated at ca 137ndash127 Ma [30]) and older than the Cenozoic as they are covered over nearlythe entire Bauru Basin by the dinosaur-bearing beds of the Mariacutelia Formation (usually accepted asMaastrichtian in age based on both vertebrates and microfossils [31ndash33]) As such the maximumdepositional age of 8778 plusmn 012 Ma provided by the U-Pb zircon geochronology (see above) furtherconstrains the age of the Buriti Farm fossil site as late Coniacian to late (although probably not the latest)Maastrichtian Previous age estimates for the Adamantina Formation were biochronology-based andrather inconsistent Based on ostracods and charophytes a TuronianndashSantonian age has been suggested[3133] but a contrasting CampanianndashMaastrichtian age also based on ostracods and charophytes wasproposed by Gobbo-Rodrigues et al [34] and corroborated by the record of sauropod dinosaurs ([32] butsee [33]) Finally a broader CenomanianndashCampanian age range was proposed based on a review of thecollective fossil content of the Adamantina and correlated formations [35]

The correlation data provided by fossils of the Buriti Farm are inconclusive The dinosaur andsnake remains are either too fragmentary or poorly studied to allow a precise taxonomic identificationSphagesaurids are largely endemic to the Bauru Basin and the only exception Yacarerani boliviensis [36]comes from the presumably Late Cretaceous (Maastrichtian) Cajones Formation of Bolivia [37] Likewisewell-known Baurusuchinae the clade that includes three of the four baurusuchids of the Buriti Farm arerestricted to the Adamantina Formation [38] As for Gondwanasuchus scabrosus it derives from an earlierphylogenetic split within Baurusuchidae [38] and could potentially indicate an older age

Our new chronostratigraphy is in agreement to some of the previous age estimates [313334] but rulesout any pre-Coniacian age for the B stardusti fossil bed In addition magnetostratigraphic studies in theUberaba Formation [39] an inferred lateral correlate of the Adamantina Formation [17] indicate that itis younger than the Cretaceous Normal Superchron (ca 121ndash83 Ma [40]) Such a Campanian maximalage fits the results of the U-Pb geochronology presented here and could be tentatively extended tothe Adamantina Formation following its inferred synchronicity to the Uberaba Formation Yet it is

on July 15 2018httprsosroyalsocietypublishingorgDownloaded from

6

rsosroyalsocietypublishingorgRSocopensci5180482

important to consider that previous workers assigned ages to the Adamantina Formation and its set ofpossible correlates like the Uberaba Formation as a whole but there is presently no evidence that thesestratigraphic units could not actually span a significant portion of their assigned age bracket Furtherstratigraphic and geochronologic work is needed to fully understand the depositional history of theBauru Basin and its extensive fossil record

42 EnamelThe presence of a thin layer of enamel partially covering the premolar of B stardusti is particularlyinteresting As mentioned the absence of enamel on the main cusp as well as the numerous crackson the dental surface could be a result of spalling related to sub-aerial weathering and desiccation [26]On the other hand the specimen does not show signs of digestion by predators as a cause of the loss orthinning of the enamel Considering the fossil fauna of the locality (see above) the most likely predatorof a small- to medium-sized mammal like Brasilestes would be a crocodile The crocodile remains areabundant in the Bauru Group and the described forms (approx 25 species [41]) range from 05 to 4 mlong approximately The digestion in modern crocodiles decalcifies tooth tissues completely removingthe enamel (except by residual interdental patches) leaving the organic matrix of the dentine and cementwhich decompose within days under aerobic conditions [26] The presence of a tooth with corrodedenamel in a crocodylomorph coprolite from the Bauru Group illustrates such a process [42] this is notthe case of specimen LPRPUSP 0751 Other possible predators in the site would be medium-sized avianand non-avian theropods but their remains are relatively scarce in the Bauru Group [4344] Digestion bybirds (except falconiforms) and mammals does not significantly alter the enamel distribution [26] Yetartificial digestion of teeth under experimental conditions shows that digestion corrodes the enamel in acharacteristic irregular pattern [45] different from the one observed in Brasilestes

43 Affinities of Brasilestes stardustiBrasilestes stardusti was compared to a wide range of Mesozoic and Cenozoic species (see electronicsupplementary material for an expanded comparison) It differs from nearly all of them except for twoeutherians a possible placental briefly described from the same geological unit as B stardusti [13] andDeccanolestes hislopi from the Maastrichtian of India [6] as detailed below

The main differences of the posterior lower premolars in the analysed taxa relative to that of Bstardusti are an incipient to absent basined talonid andor the presence of more cusps This is thecase of Australosphenida (see below) Eutriconodonta Allotheria lsquoSymmetrodontarsquo and lsquoEupantotheriarsquo(including Dryolestoidea Meridiolestida and Zatheria) Some deciduous premolars of Nanolestes have aposterior talonid as developed as in B stardusti but are anteriorly multicuspate and have a basin amongthe cusps [46] In the early tribosphenidan Slaughteria p3 has a talonid on the posterior third of the toothbut it ends in a posterior cusp and a labial cuspule whereas the main cusp is flanked by an anteriorand a posterior crest dp4 and dp5 are molariforms with fully developed trigonids [5] The posteriorpremolars of metatherians differ from that of B stardusti because they are labiolingually compressedand relatively more symmetric (Didelphimorphia Paucituberculata Microbiotheria Sparassodonta andDasyuromorpha) or they are robust molariform or have plagiaulacoid aspect (Polydolopimorphia andDiprotodontia) The Australian marsupial Naboryctes shows asymmetric posterior premolars but themain cusp is procumbent and the incipient talonid ends in a cusp Among Eutheria most taxa haveeither labiolingually compressed teeth no or incipient talonid more cuspscrests than B stardusti or(sub)molariform lower premolars In Gypsonictops although p4 has a talonid and is asymmetric in lateralview as in B stardusti there is an anterior cingulum one to three posterior cuspules and the posteriorwall of the main cusp develops into a crest [47]

As for the reduced enamel when present in Palaeanodonta this tissue is composed of a relatively thinlayer (70ndash140 microm [48]) However the enamel is much thicker than that of B stardusti as well as prismaticand the morphology of their posterior lower premolars differs in having either more crestscuspsor absentincipient talonid (eg Ernanodon [4950] Melaniella [51] Tubulodon [52]) In Mylanodon p4develops a talonid that narrows posteriorly as in B stardusti but it lacks an anterior crest and the maincusp shows a vertical lingual groove [53]

In Xenarthra even though most taxa completely lack enamel a thin layer is retained in the extinctarmadillos Astegotherium and Utaetus and in the extant long-nosed armadillo Dasypus [54] but theirhomodont dentition is composed of subcylindric teeth Work in progress on the microstructure ofpermanent molars of Dasypus (M Ciancio ECV MCC and AA Carlini in preparation) shows a thin

on July 15 2018httprsosroyalsocietypublishingorgDownloaded from

7

rsosroyalsocietypublishingorgRSocopensci5180482

enamel layer (less than 25 microm) with or without prisms depending on the species a condition strikinglysimilar to that of B stardusti (see electronic supplementary material figure S2) Different lines of evidenceindicate that xenarthrans had a South American origin no younger than 85 Ma [5556] but the oldestfossils of the group come from the early Palaeogene [56] Although the age and provenance of B stardusti(ConiacianndashMaastrichtian of tropical South America) match previous molecular hypotheses for the originof Xenarthra inference of taxonomic affinity is premature in the face of the morphologic differencesscarcity of remains and plasticity of enamel reduction among different mammalian lineages

Conspicuous similarities were found only between B stardusti and two other taxa both referred toEutheria The first is Deccanolestes hislopi (Maastrichtian of India) particularly the p3 figured in theoriginal description [6] (fig 3 NKIM 12 also in [57] Plate III) As in B stardusti there is a clear asymmetrybetween the anterior and the posterior halves a single large cusp and a small notch in the lingual borderof the talonid delimiting a minute cuspule In fact B stardusti is more similar to that specimen of D hislopi(NKIM 12) than to any other material of our knowledge They differ on the talonid which is flatter inB stardusti on the angle formed by the posterior surface of the main cusp and the talonid (perpendicularin B stardusti and obtuse in D hislopi) and on the presence of one or two talonid cusps in the p2 andp3 of D hislopi [58] Also there is a considerable difference in the size of both teeth the p3 of D hislopiis 041 mm long whereas the premolar of B stardusti is 35 mm long Among other species of the genusD robustus has larger teeth (the m1 is approximately 15 mm long [59]) but still much smaller than thatof B stardusti Deccanolestes has been regarded as an Adapisoriculdae a group that is considered either anon-placental Eutheria or an Euarchonta [15166061]

Similarities are also found between B stardusti and a possible Placentalia that was previously collectedin the Adamantina Formation [13] especially in terms of the talonid development However besides thesize difference (the length of that p3 is 12 mm) its talonid ends in a posterior cusp Also the main cusphas an anterior and a posterior crest is proportionally smaller and more labiolingually compressed thanthat of B stardusti

The incompleteness of B stardusti precludes a more precise taxonomic assignment As for theinclusion of the taxon in a cladistic framework the data matrix of [62] for example focuses on SouthAmerican Mesozoic mammals and presents 317 craniodental characters Out of those the characters thatcan be scored for B stardusti (ie related to isolated premolars) refer to the morphology of the penultimate(characters 43 and 149) or the ultimate (characters 44 45 47 48 50ndash52 and 150) lower premolar Althoughwe infer that B stardusti is represented by a posterior lower premolar the total number of premolars andthe exact locus occupied by this tooth are uncertain thus the inclusion of B stardusti in the availablephylogenetic studies will be of limited use regarding the inference of its taxonomic affinities

The diagnoses of Tribosphenida (=Boreosphenida see [45]) and Australosphenida presented by Luoet al [2] include features of the last lower premolar This tooth is defined as labiolingually wide in theposterior part and having a fully triangulated trigonid in Australosphenida whereas those featuresare absent in Tribosphenida Also an asymmetric main cusp is not observed in the last premolar ofaustralosphenidans [2] a condition present in B stardusti In this scenario as determining the positionof its premolar in the lower tooth row is unfeasible the affinity of B stardusti with Australosphenidacannot be completely ruled out However based on the morphological features described above and onthe similarities with two purported eutherians we attribute B stardusti to Tribosphenida

44 Biogeographic implicationsThe evolutionary history of mammals during the Late CretaceousndashEarly Palaeocene in South America[63] is marked by the almost complete extinction of endemic non-tribosphenidan lineages (leading tothe end of the lsquoGondwanan Episodersquo) and the expansion of Tribosphenida (the so-called South AmericanEpisode [64]) This remarkable change in land-mammal communities (First Great Turnover) is howeverbased primarily on the well-studied Late Cretaceous faunas of Patagonia [64] Along with previousrecords of putative tribosphenidans in Peru [12] Bolivia [14] and Brazil [13] B stardusti partially fills twomajor gaps the paucity of mammals in more northern parts of South America and of tribosphenidans inthe Cretaceous of the continent [65]

During the Late Cretaceous South America was divided into northeastern and southwestern portionsby a sea corridor ([636566] figure 3) which probably acted as a barrier or filter for some organismsMoreover throughout this period those regions had different climatic conditions with the warm-temperate Patagonia separated from the arid areas of central and northeastern South America bya subtropical seasonal dry belt [6768] Indeed the Late Cretaceousndashearliest Palaeocene terrestrialbiotas of Patagonia have greater similarity to those of East Gondwana (ie Antarctica and Australia)

on July 15 2018httprsosroyalsocietypublishingorgDownloaded from

8

rsosroyalsocietypublishingorgRSocopensci5180482

Figure 3 Distribution of Gondwanan landmasses in the Late Cretaceous (80 Ma) and known mammalian records Circle eutheriansquare indeterminate tribosphenidan plus non-tribosphenidan (map generated using the softwares GPlates 200 wwwgplatesorgand PaleoData Plotter httpswwwearthbyteorgpaleomap-paleoatlas-for-gplates [66])

whereas the faunal assemblage of northern South America suggests a closer biogeographic relation withAfrica [6365]

Brasilestes stardusti and the three other records mentioned above [12ndash14] indicate the presence oftribosphenidans in the Late Cretaceous of northern South America (figure 3) Biogeographically thesetribosphenidans may have either dispersed from the North American part of Laurasia during the LateCretaceous (as part of the so-called First American Biotic Interchange [65]) or they may have had aGondwanic origin [69]

The presence of northern immigrants like hadrosaurs and probably marsupials in South America[65] and conversely of southern immigrants like titanosaurs in North America [70] has been explainedby a transient inter-American connection during the Late Cretaceous In this biogeographic contextB stardusti and the other tribosphenidans recorded in northern South America (figure 3) could beinterpreted as mammals with a northern ancestry The tribosphenid mammals of Peru and Boliviahave been considered as related to North American immigrants (see [3] and references therein) but Bstardusti raises more questions If B stardusti is related to Adapisoriculidae (as inferred for Deccanolestes)or to Xenarthra (hinted by the reduced enamel) the lack of comparable coeval Laurasian forms isindirect evidence against the hypothesis that the taxon had a northern ancestry According to the knownfossil record the oldest adapisoriculids have been recorded in Late Cretaceous beds of India [60] anddifferent lines of evidence indicate that xenarthrans had a South American origin no younger than85 Ma [557172]

Consequently the hypothesis of a Gondwanan ancestry of B stardusti appears to be better supportedthan the Laurasian one Another indirect evidence of a potential Gondwanan origin of B stardusticomes from the record of the advanced theriiform Vincelestes neuquenianus in the Early Cretaceous ofPatagonia In spite of V neuquenianus not being closely related to therians (eg [73]) its record opens thepossibility that forms like B stardusti in northern South America the adapisoriculid Deccanolestes in Indiaor perhaps xenarthrans in South America could represent late and derived survivors of that theriiformlineage in Gondwana

A more precise taxonomic placement of B stardusti depends upon additional discoveries in the BauruBasin but this finding underscores the importance of northern South America in further understandingthe history of mammals in Gondwanan landmasses during the Late Cretaceous

5 ConclusionBased on an isolated lower premolar B stardusti represents a new tribosphenidan mammal a raregroup in Gondwanan deposits of Cretaceous age The tooth is larger than those of most coeval

on July 15 2018httprsosroyalsocietypublishingorgDownloaded from

9

rsosroyalsocietypublishingorgRSocopensci5180482

mammals but the definition of more precise taxonomic affinities is hampered by the incompletenessof the type-specimen It shows a thin layer of parallel crystallite enamel as previously unrecognizedamong Cretaceous tribosphenidans Geologic interpretation of the fossil site indicates intermittentsedimentation events probably in fluvial environments under a semi-arid climate and low transport ofbioclasts The maximum age of its type-stratum is constrained by a high-precision U-Pb geochronologyas late Coniacian age and by the superposed dinosaur-bearing beds of the Mariacutelia Formation indicatinga late Coniacian to late (although probably not latest) Maastrichtian time range for the AdamantinaFormation The occurrence of B stardusti along with two morphologically similar forms a putativeeutherian from the Adamantina Formation and Deccanolestes hislopi from India indicates the existenceof a still poorly sampled tribosphenidan diversity in the Late Cretaceous of Gondwanan landmasses

Data accessibility The datasets supporting this article have been uploaded as part of the electronic supplementarymaterial including expanded comparison of Brasilestes stardusti additional data on the enamel geological settingsand U-Pb geochronologyAuthorsrsquo contributions JCAM collected the specimen FJG EO-J and ECV performed the enamel microstructureanalysis KT and JR conducted the geochronological study of the type stratum AB analysed the geology of thefossil site MCC FJG EO-J and MCL interpreted the data All authors wrote the main manuscriptCompeting interests We have no competing interestsFunding This study was supported by Satildeo Paulo Research Foundation (FAPESP) grants nos 201423815-2 to MCC201517632-5 to AB 201323114-1 and 201602473-1 to JCAM and 201403825-3 to MCLAcknowledgements The authors thank M Tomeo for handling the images of Brasilestes stardusti Zhe-Xi Luo for his kindsuggestions GW Rougier for detailed review of a previous version of the manuscript GS Ferreira for phylogenetictests BW McPhee for improving the English TS Marinho for the information on the provenance of fossils withinthe Buriti Farm

References1 McKenna MC 1975 Toward a phylogenetic

classification of the Mammalia In Phylogeny of theprimates (eds WP Luckett FS Szalay) pp 21ndash46New York NY Plenum Press

2 Luo Z-X Cifelli RL Kielan-Jaworowska Z 2001 Dualorigin of tribosphenic mammals Nature 40953ndash57 (doi10103835051023)

3 Kielan-Jaworowska Z Cifelli RL Luo Z-X 2004Mammals from the age of dinosaurs originsevolution and structure New York NY ColumbiaUniversity Press

4 Rougier GW Martinelli AG Forasiepi AM NovacekMJ 2007 New Jurassic mammals from PatagoniaArgentina a reappraisal of australosphenidanmorphology and interrelationships Am Mus Novit3566 1ndash54 (doi1012060003-0082(2007)507[1NJMFPA]20CO2)

5 Davis BM 2011 A novel interpretation of thetribosphenidan mammal Slaughteria eruptens fromthe Lower Cretaceous Trinity Group andimplications for dental formula in early mammalsJ Vert Paleontol 31 676ndash683(doi101080027246342011558149)

6 Prasad GVR Sahni A 1988 First Cretaceous mammalfrom India Nature 332 638ndash640 (doi101038332638a0)

7 Rana RS Wilson GP 2003 New Late Cretaceousmammals from the Intertrappean beds ofRangapur India and paleobiogeographicframework Acta Palaeontol Pol 48 331ndash348

8 Prasad GVR Verma O Sahni A Parmar V Khosla AA2007 Cretaceous hoofed mammal from IndiaScience 318 937 (doi101126science1149267)

9 Krause DW 2001 Fossil molar from a Madagascanmarsupial Nature 412 497ndash498 (doi10103835087649)

10 Averianov AO Archibald JD Martin T 2003Placental nature of the alleged marsupial from the

Cretaceous of Madagascar Acta Palaeontol Pol 48149ndash151

11 Sigogneau-Russell D 1995 Further data andreflexions on the tribosphenid mammals(Tribotheria) from the Early Cretaceous of MoroccoBull Mus Natl Hist Nat Paris 16 291ndash312

12 Mourier T et al 1986 Deacutecouverte de de restesdinosauriens et mammalien drsquoacircge creacutetaceacute supeacuterieuragrave la base des couches rouges du synclinal de Bagua(Andes nord-peacuteruviennes) aspectsstratigraphiques seacutedimentologiques etpaleacuteogeacuteographiques concernant la reacutegressionfini-creacutetaceacutee Bull Soc geacuteol France 2 171ndash175

13 Bertini RJ Marshall LG Gayet M Brito P 1993Vertebrate faunas from the Adamantina and Mariacuteliaformations (Upper Bauru Group Late CretaceousBrazil) in their stratigraphic andpaleobiogeographic context N Jb Geol PalaontolAbh 188 71ndash101

14 Gayet M et al 2001 Middle Maastrichtianvertebrates (fishes amphibians dinosaurs andother reptiles mammals) from Pajcha Pata(Bolivia) biostratigraphic palaeoecologic andpalaeobiogeographic implications PalaeogeogPalaeoclimatol Palaeoecol 169 39ndash68(doi101016S0031-0182(01)00214-0)

15 Boyer DM et al 2010 New postcrania ofDeccanolestes from the Late Cretaceous of India andtheir bearing on the evolutionary andbiogeographic history of euarchontan mammalsNaturwissenschaften 97 365ndash377(doi101007s00114-010-0648-0)

16 Smith T De Bast E Sigeacute B 2010 Euarchontan affinityof Paleocene Afro-European adapisoriculidmammals and their origin in the Late CretaceousDeccan Traps of India Naturwissenschaften 97417ndash422 (doi101007s00114-010-0651-5)

17 Batezelli A 2015 Continental systems tracts of theBrazilian Cretaceous Bauru Basin and theirrelationship with the tectonic and climaticevolution of South America Basin Res 27 1ndash25(doi101111bre12102)

18 Cioffi CR Neto MDC Moller A Rocha BC 2016Paleoproterozoic continental crust generationevents at 215 and 208 Ga in the basement of thesouthern Brasilia Orogen SE Brazil PrecambrianRes 275 176ndash196 (doi101016jprecamres201601007)

19 Ramezani J et al 2011 High-precision U-Pb zircongeochronology of the Late Triassic ChinleFormation Petrified Forest National Park (ArizonaUSA) temporal constraints on the early evolution ofdinosaurs Geol Soc Am Bull 123 2142ndash2159(doi101130B304331)

20 Mattinson JM 2005 Zircon UPb chemical abrasion(CA-TIMS) method combined annealing andmulti-step partial dissolution analysis for improvedprecision and accuracy of zircon ages Chem Geol220 47ndash66 (doi101016jchemgeo200503011)

21 Condon DJ Schoene B McLean NM Bowring SAParrish RR 2015 Metrology and traceability of U-Pbisotope dilution geochronology (EARTHTIME TracerCalibration Part I) Geochim Cosmochim Acta 164464ndash480 (doi101016jgca201505026)

22 McLean NM Condon DJ Schoene B Bowring SA2015 Evaluating uncertainties in the calibration ofisotopic reference materials and multi-elementisotopic tracers (EARTHTIME Tracer Calibration PartII) Geochim Cosmochim Acta 164 481ndash501(doi101016jgca201502040)

23 Bowring JF McLean NM Bowring SA 2011Engineering cyber infrastructure for U-Pbgeochronology Tripoli and U-Pb_Redux GeochemGeophys Geosyst 12 Q0AA19 (doi1010292010GC003479)

on July 15 2018httprsosroyalsocietypublishingorgDownloaded from

10

rsosroyalsocietypublishingorgRSocopensci5180482

24 McLean NM Bowring JF Bowring SA 2011 An

algorithm for U-Pb isotope dilution data reductionand uncertainty propagation Geochem GeophysGeosyst 12 1ndash26 (doi1010292010GC003478)

25 Flynn LJ Wahlert JH 1978 SEM study of rodentincisors preparation and viewing Curator 21303ndash310 (doi101111j2151-69521978tb00550x)

26 Fisher JA Nichols GJ Waltham DA 2007 Unconfinedflow deposits in distal sectors of fluvial distributarysystems examples from the Miocene Luna andHuesca Systems northern Spain Sediment Geol195 75ndash90 (doi101016jsedgeo200607005)

27 Previtera E Drsquoangelo JA Mancuso AC 2013Preliminary chemometric study of bone diagenesisin Early Triassic cynodonts fromMendozaArgentina Ameghiniana 50 460ndash468(doi105710AMGH272013623)

28 Smith T Codrea V 2015 Red iron-pigmented toothenamel in a multituberculate mammal from theLate Cretaceous Transylvanian lsquoHaţeg Islandrsquo PLoSONE 10 e0132550

29 Dauphin Y Williams CT 2007 The chemicalcompositions of dentine and enamel from recentreptile and mammal teethmdashvariability in thediagenetic changes of fossil teeth CrystEngComm 91252ndash1261 (doi101039b708985f)

30 Turner S Regelons M Kelley S Jawkesworth CMantovani MSM 1994 Magmatism and continentalbreak-up in the South Atlantic high precisiongeochronology Earth Planet Sci Lett 121 333ndash348(doi1010160012-821X(94)90076-0)

31 Dias-Brito D et al 2001 Grupo Bauru uma unidadecontinental Cretaacutecea no Brasilndashconcepccedilotildeesbaseadas em dados micropaleontoloacutegicosisotoacutepicos e estratigraacuteacuteficos Rev Paleacuteobiol 20245ndash304

32 Santucci RM Bertini RJ 2001 Distribuiccedilatildeopaleogeograacutefica e biocronoloacutegica dos titanossauros(Saurischia Sauropoda) do Grupo Bauru CretaacuteceoSuperior do Sudeste brasileiro Braz J Geol 31307ndash315

33 Martinelli AG Riff D Lopes RP 2011 Discussionabout the occurrence of the genus AeolosaurusPowell 1987 (Dinosauria Titanosauria) in the upperCretaceous of Brazil Gaea-J Geosci 7 34ndash40(doi104013gaea20117103)

34 Gobbo-Rodrigues SR Petri S Bertini RJ 1999Ocorrecircncias de ostraacutecodes na FormaccedilatildeoAdamantina do Grupo Bauru Cretaacuteceo Superior daBacia do Paranaacute e possibilidades de correlaccedilatildeo comdepoacutesitos isoacutecronos argentinos Parte I-FamiacuteliaIlyocyprididae Acta Geol Leopoldensia 233ndash13

35 Menegazzo MC Catuneanu O Chang HK 2016 TheSouth American retroarc foreland system thedevelopment of the Bauru Basin in the back-bulgeprovinceMar Petrol Geol 73 131ndash156(doi101016jmarpetgeo201602027)

36 Novas FE Pais DF Pol D Carvalho IDS Scanferla AMones A Riglos MS 2009 Bizarre notosuchiancrocodyliform with associated eggs from the UpperCretaceous of Bolivia J Vert Paleontol 291316ndash1320 (doi1016710390290409)

37 Sempere T et al 1997 Stratigraphy and chronologyof Upper Cretaceousndashlower Paleogene strata inBolivia and northwest Argentina Geol Soc AmBull 109 709ndash727 (doi1011300016-7606(1997)109lt0709SACOUCgt23CO2)

38 Godoy PL Montefeltro FC Norell MA Langer MC2014 An additional baurusuchid from the Cretaceousof Brazil with evidence of interspecific predationamong Crocodyliformes PLoS ONE 9 e97138(doi101371journalpone0097138)

39 Tamrat E Ernesto M Fulfaro VJ Saad AR BatezelliA Oliveira AF 2002 Magnetoestratigrafia dasformaccedilotildees Uberaba e Mariacutelia (Grupo Bauru) noTriacircngulo Mineiro (MG) In Boletim do 6deg Simpoacutesiosobre o Cretaacuteceo do brasil e 2deg Simpoacutesio sobre elCreaacutecico de Ameacuterica del Sur Satildeo pedro pp 323ndash327

40 Granot R Dyment J Gallet Y 2012 Geomagneticfield variability during the Cretaceous NormalSuperchron Nat Geosci 5 220 (doi101038ngeo1404)

41 Martinelli AG Teixeira VPA 2015 The LateCretaceous vertebrate record from the Bauru Groupin the Triacircngulo Mineiro southeastern Brazil BolGeol Min 126 129ndash158

42 Souto PDF 2010 Crocodylomorph coprolites fromthe Bauru basin upper Cretaceous Brazil Bull NewMexico Mus Nat Hist Sci 51 201ndash208

43 Alvarenga H Nava WR 2005 Aves Enantiornithes doCretaacuteceo Superior da Formaccedilatildeo Adamantina doEstado de Satildeo Paulo Brasil II CongresoLatinoamericano de Paleontologiacutea de VertebradosRio de Janeiro

44 Marsola JCA Grellet-Tinner G Montefeltro FCSayatildeo JM Hsiou AS Langer MC 2014 The first fossilavian egg from Brazil Alcheringa 38 563ndash567(doi101080031155182014926449)

45 Fernaacutendez-Jalvo Y Andrews P Sevilla P Requejo V2014 Digestion versus abrasion features in rodentbones Lethaia 47 323ndash336 (doi101111let12061)

46 Martin T 2002 New stem-lineage representatives ofZatheria (Mammalia) from the Late Jurassic ofPortugal J Vert Paleontol 22 332ndash348(doi1016710272-4634(2002)022[0332NSLROZ]20CO2)

47 Lillengraven JA 1969 Latest Cretaceous mammalsof upper part of Edmonton Formation of AlbertaCanada and review of marsupial-placentaldichotomy in mammalian evolution Univ KansasPaleontol Contrib 50 1ndash122

48 Kalthoff DC Rose KD von Koenigswald W 2011Dental microstructure in Palaeanodon andTubulodon (Palaeanodonta) and bioerosionaltunneling as a widespread phenomenon in fossilmammal teeth J Vert Paleontol 31 1303ndash1313(doi101080027246342011607997)

49 Radinsky L Ting SY 1984 The skull of Ernanodon anunusual fossil mammal J Mammal 65 155ndash158(doi1023071381220)

50 Kondrashow P Agadjanian AK 2012 A nearlycomplete skeleton of Ernanodon (MammaliaPalaeanodonta) fromMongolia morphofunctionalanalysis J Vert Paleontol 32 983ndash1001(doi101080027246342012694319)

51 Fox RC 1984Melaniella timosa n gen and sp anunusual mammal from the Paleocene of AlbertaCanada Can J Earth Sci 21 1335ndash1338(doi101139e84-138)

52 Jepsen GL 1932 Tubulodon taylori a Wind RiverEocene tubulidentate fromWyoming Proc AmPhil Soc 71 255ndash274

53 Secord R Gingerich PD Bloch JI 2002Mylanodonrosei a newmetacheiromyid (MammaliaPalaeanodonta) from the Late Tiffanian (Late

Paleocene) of Northwestern Wyoming ContribMus Paleontol Univ Michigan 30 385ndash399

54 Ciancio MR Vieytes EC Carlini AA 2014 Whenxenarthrans had enamel insights on the evolutionof their hypsodonty and paleontological support forindependent evolution in armadillosNaturwissenschaften 101 715ndash725 (doi101007s00114-014-1208-9)

55 dos Reis M et al 2012 Phylogenomic datasetsprovide both precision and accuracy in estimatingthe timescale of placental mammal phylogenyProc R Soc B 279 3491ndash3500 (doi101098rspb20120683)

56 Bergqvist LP Abrantes EacuteAL Avilla LS 2004 TheXenarthra (Mammalia) of Satildeo Joseacute de Itaboraiacute Basin(upper Paleocene Itaboraian) Rio de JaneiroBrazil Geodiversitas 26 323ndash337

57 Prasad GVR Khajuria CK 1990 A record ofmicrovertebrate fauna from the intertrappean bedsof Naskal Andhra Pradesh J Palaeontol Soc India35 151ndash161

58 Kapur VV Das DP Bajpai S Prasad GVR 2017 Firstmammal of Gondwanan lineage in the early Eoceneof India CR Palevol 16 721ndash737 (doi101016jcrpv201701002)

59 Prasad GVR Jaeger JJ Sahni A Gheerbrant EKhajuria CK 1994 Eutherian mammals from theupper Cretaceous (Maastrichtian) intertrappeanbeds of Naskal Andhra Pradesh India J VertPaleontol 14 260ndash277 (doi10108002724634199410011556)

60 Prasad GVR Verma O Gheerbrant E Goswami AKhosla A Parmar V Sahni A 2010 First mammalevidence from the Late Cretaceous of India for bioticdispersal between India and Africa at the KTtransition CR Palevol 9 63ndash71 (doi101016jcrpv200912003)

61 Goswami A et al 2011 A radiation of arboreal basaleutherian mammals beginning in the LateCretaceous of India Proc Natl Acad Sci USA 10816 333ndash16 338 (doi101073pnas1108723108)

62 Rougier GW Wible JR Beck RM Apesteguiacutea S 2012The Miocene mammal Necrolestes demonstrates thesurvival of a Mesozoic nontherian lineage into thelate Cenozoic of South America Proc Natl Acad SciUSA 109 20 053ndash20 058 (doi101073pnas1212997109)

63 Pascual R Ortiz-Jaureguizar E 2007 TheGondwanan and South American Episodes twomajor and unrelated moments in the history of theSouth American mammals J Mamm Evol 1475ndash137 (doi101007s10914-007-9039-5)

64 Pascual R Balarino ML Udrizar Sauthier DE 2001The KT and Tertiary-Pleistocene South Americanmammalian turnovers similar phenomenaAsoc Paleontol Argent Publ Especial 7 151ndash156

65 Goin FJ Woodburne MO Zimicz AN Martin GMChornogubsky L 2016 A brief history of SouthAmerican metatherians evolutionary contexts andintercontinental dispersals New York NYSpringer

66 Scotese CR 2016 PALEOMAP PaleoAtlas for GPlatesand the PaleoData Plotter Program PALEOMAPProject See httpwwwearthbyteorgpaleomap-paleoatlas-for-gplates (accessed 17 October 2017)

67 Iglesias A Artabe AE Morel EM 2011 The evolutionof Patagonian climate and vegetation from theMesozoic to the present Biol J Linn Soc 103

on July 15 2018httprsosroyalsocietypublishingorgDownloaded from

11

rsosroyalsocietypublishingorgRSocopensci5180482

409ndash422 (doi101111j1095-8312201101657x)

68 Hay WW Floegel S 2012 New thoughts about theCretaceous climate and oceans Earth Sci Rev 115262ndash272 (doi101016jearscirev201209008)

69 Murphy WJ et al 2001 Resolution of early placentalmammal radiation using Bayesian phylogeneticsScience 294 2348ndash2351 (doi101126science1067179)

70 Lucas SG Hunt AP 1989 Alamosaurus and thesauropod hiatus in the Cretaceous of the NorthAmerican Western Interior Geol Soc Spec Pap238 75ndash85

71 Woodburne MO Rich TH Springer MS 2003 Theevolution of tribospheny and the antiquity ofmammalian cladesMol Phylogenet Evol 28360ndash385 (doi101016S1055-7903(03)00113-1)

72 Nishihara H Maruyama S Okada N 2009Retroposon analysis and recent geological datasuggest near-simultaneous divergence of the threesuperorders of mammals Proc Natl Acad Sci USA106 5235ndash5240 (doi101073pnas0809297106)

73 Luo Z-X Yuan CX Meng QJ Ji Q 2011 A Jurassiceutherian mammal and divergence of marsupialsand placentals Nature 476 442ndash445 (doi101038nature10291)

on July 15 2018httprsosroyalsocietypublishingorgDownloaded from