Revised Chronostratigraphy and Biostratigraphy of the John Day Formation (Turtle Cove and Kimberly...

[The Journal of Geology, 2008, volume 116, p. 211–237] � 2008 by The University of Chicago.All rights reserved. 0022-1376/2008/11603-0001$15.00. DOI: 10.1086/587650

211

ARTICLES

Revised Chronostratigraphy and Biostratigraphy of the John DayFormation (Turtle Cove and Kimberly Members), Oregon, with

Implications for Updated Calibration of the ArikareeanNorth American Land Mammal Age

L. Barry Albright III, Michael O. Woodburne,1 Theodore J. Fremd,2

Carl C. Swisher III,3 Bruce J. MacFadden,4 and Gary R. Scott5

Department of Chemistry and Physics, University of North Florida,Jacksonville, Florida 32224, U.S.A.

(e-mail: [email protected])

A B S T R A C T

Although the Arikareean North American land mammal age was first typified in the Great Plains, the successionthere contains significant unconformities, a generally poor magnetic record, relatively sparse radioisotopic calibration,and a major faunal hiatus. In the John Day Valley of central Oregon, however, is a thick, remarkably completesequence of Oligocene through early Miocene strata (the John Day Formation) potentially amenable to addressingthese shortcomings and long known to harbor one of the richest records of mid-Tertiary mammals in North America.Since Prothero and Rensberger’s first magnetostratigraphic study of the John Day Formation in 1985, new advancesin geochronology, together with a more comprehensive suite of paleomagnetic sections keyed to new radioisotopicand biostratigraphic data, have greatly enhanced chronostratigraphic precision. In our attempt to refine John Daychronostratigraphy, we sampled nearly 300 sites for magnetostratigraphy over a 500-m-thick interval and used severalradioisotopically dated volcanic tuffs for our correlation with the geomagnetic polarity timescale. Many of the rocksanalyzed showed unusual magnetic behavior, possibly due to the known zeolitization in this region, thereby precludingan abundance of class 1 polarity determinations. Nevertheless, preliminary results indicate that the Turtle CoveMember stratigraphically upward through the lower Kimberly Member extends from late chron C12n through C7n.1r,or from about 30.6 to 24.1 Ma. Intensive radioisotopic and magnetostratigraphic characterization of these strataprovides a framework by which the associated biostratigraphy is assessed for biochronological significance relativeto fossiliferous successions of the Great Plains, in turn resulting in reassessment of Arikareean subbiochron (Ar1–Ar4) boundaries. We present a revision of those boundaries that differs from their traditional timing as a hypothesisfor testing in other locations.

Introduction

In and around the John Day Valley of centralOregon is a thick, remarkably complete sequence

Manuscript received November 16, 2007; accepted February7, 2008.

1 Department of Geology, Museum of Northern Arizona,Flagstaff, Arizona 86001, U.S.A.

2 John Day Fossil Beds National Monument, 32651 Highway19, Kimberly, Oregon 97848, U.S.A.

3 Department of Geological Sciences, Rutgers University,Piscataway, New Jersey 08854, U.S.A.

4 Florida Museum of Natural History, University of Florida,Gainesville, Florida 32611, U.S.A.

5 Berkeley Geochronology Center, 2455 Ridge Road, Berke-ley, California 94709, U.S.A.

of middle Eocene to late Miocene (approximately47–7 Ma) terrestrial, predominantly tuffaceousstrata long known to harbor one of the richest re-cords of Tertiary animals and plants in NorthAmerica. Of 11 Cenozoic North American landmammal ages (NALMAs) spanned by this intervalof time (Bridgerian through Hemphillian), mam-mals representing all but the Chadronian and Clar-endonian have been recovered, as recorded in strataof the middle Eocene Clarno Formation throughthe Oligocene to upper Miocene John Day, Mascall,and Rattlesnake formations (fig. 1). Thus, nearly 40m.yr. of North American mammalian evolution

212 L . B . A L B R I G H T I I I E T A L .

Figure 1. Stratigraphic relationship of units comprisingthe John Day sequence in the eastern facies, with cor-relation with North American land mammal ages. Strat-igraphic thickness not to scale. Numbers indicate agesin Ma.

and climate change are spanned by this sequence.More remarkable still is the ability to correlate un-ambiguously from outcrop to outcrop, with virtu-ally no stratigraphic gaps, because of the numerousand regionally widespread volcanic marker bedsthat punctuate the sequence, together with severalother lithologically distinctive units. The John Daysequence is therefore ideally suited for modernchronostratigraphic investigation because it in-cludes a thick, structurally noncomplicated strat-igraphic section lithologically amenable to paleo-magnetic analysis that is rich in well-preservedmammal fossils sandwiched between numerous ra-dioisotopically datable volcanic units.

Presented here are the results of a magnetostrat-igraphic study conducted from 1997 to 1999 on theJohn Day Formation, which ranges in age fromabout 39 to 18 Ma and includes the strata that restdisconformably on the upper Eocene Clarno For-mation and nonconformably below the middleMiocene Columbia River Basalts (fig. 1). The JohnDay Formation crops out in three geographicallydistinct regions, informally referred to as the east-ern, western, and southern facies (fig. 2), and con-sists of approximately 1000 m of primarily airfall

tuffs and fluvially worked and pedogenetically al-tered tuffaceous claystones (Robinson et al. 1984;Bestland and Retallack 1994; Retallack et al. 2000).

Specifically, our magnetostratigraphic fieldworkfocused on the predominantly Arikareean-agedTurtle Cove and Kimberly members in the classicTurtle Cove region of the eastern facies within thePicture Gorge area of John Day Fossil Beds NationalMonument (JDFBNM). The region includes thestrata exposed in Wheeler and Grant counties onboth sides of the John Day River Valley, from Pic-ture Gorge in the south to the town of Kimberlyin the north (fig. 3). Equivalent strata also crop outin the Painted Hills area of the JDFBNM (fig. 2).The Turtle Cove and Kimberly members were fo-cused on because (1) they span the greatest intervalof time in the sequence; (2) they preserve abundantand stratigraphically important mammalian assem-blages, including a record that spans several inter-vals determined to be missing in the Great Plainssequences from which the Arikareean land mam-mal age is typified; (3) they contain abundant tephraand sediments suitable for radioisotopic dating andmagnetostratigraphic analysis, with which to en-hance their regional and global correlation; and (4)they are the least complicated in terms of structureand stratigraphy. The underlying Big Basin Membercontains only limited occurrences of vertebrate fos-sils, and the overlying fossiliferous Haystack ValleyMember was recently revised by Hunt and Steple-ton (2004) and shown to be more complex thanpreviously considered (fig. 1). We comment on thatpart of the section as well.

Historically, study of mammalian evolution inNorth America over the Whitneyan, Arikareean,and Hemingfordian land mammal ages focused onthe Great Plains sequences where these mammalages were first typified (Wood et al. 1941; Tedfordet al. 1987, 1996, and references therein). However,the succession there contains significant uncon-formities, a generally poor magnetic record, and rel-atively sparse radioisotopic calibration (Hunt 1990;MacFadden and Hunt 1998; Tedford et al. 2004;Hayes 2007). Realizing that these shortcomingscould be mitigated by the stratigraphically morecomplete sequence of fossiliferous strata in theJohn Day Formation, Prothero and Rensberger(1985) undertook the first magnetostratigraphicstudy there in an attempt to more accurately cal-ibrate this interval and to correlate it with the ma-rine record. In the intervening 2 decades, however,considerable advances have occurred in geochro-nology. This study was conducted, therefore, (1) asan attempt to update and refine the earlier mag-netostratigraphic work of Prothero and Rensberger

Journal of Geology J O H N D A Y F O R M A T I O N S T R A T I G R A P H Y 213

Figure 2. Index map showing generalized outcrop pattern of the eastern, western, and southern facies of the JohnDay Formation. Note Picture Gorge and Painted Hills regions in eastern facies. Modified from Robinson et al. (1984).

(1985), (2) to present a modern appraisal of thebiochronology of the John Day succession based onthe significantly more detailed understanding of itslitho- and biostratigraphy, and (3) to provide for re-appraisal and refinement of the correlation betweenJohn Day faunas and the classic, mostly Arikareeanmammalian faunas of the northern Great Plains.We also propose a revision of the age boundaries ofthe Ar2–Ar4 Arikareean biochrons based on ournew correlation of the John Day succession andindicate how those revisions are reconciled withthe Great Plains sequence on which those bio-chrons were based.

Background

The geological and paleontological importance ofthe John Day succession was recognized as early asthe late mid-1800s, when the Reverend ThomasCondon first studied fossils from this region. Con-don, in turn, introduced E. D. Cope, O. C. Marsh,and J. Leidy to the paleontological treasures of “thecove.” Several studies followed during the earlypart of the twentieth century, most notably thoseof Merriam (1901), W. J. Sinclair, C. Stock, and R.Chaney, which were followed by more recent stud-ies (e.g., Hay 1963; Fisher 1967; Robinson et al.

1984; Prothero and Rensberger 1985; Fremd et al.1994).

Merriam (1901) divided the John Day Formationinto three divisions—a lower, a middle, and an up-per—primarily on the basis of color, whereas Hay’s(1963) lower, middle, and upper members werebased on the widespread presence of a rhyolitic ig-nimbrite (his middle member) near the middle ofthe formation. Today, this unit is commonly re-ferred to as the Picture Gorge Ignimbrite (PGI; fig.1). Fisher and Rensberger (1972) divided the se-quence into four members—in ascending order, theBig Basin, Turtle Cove, Kimberly, and HaystackValley—and following this stratigraphic scheme,Fremd et al. (1994) further subdivided the highlyfossiliferous Turtle Cove and Kimberly membersinto several informal lithostratigraphic subunits togain more precise biostratigraphic control. LabeledA–M and based on “general lithologic similaritiesand the presence of distinctive tuff beds, commonlyat the base of the units” (Bestland 1995, p. 1), it isthis stratigraphic model that we follow.

As noted above, it was Prothero and Rensberger(1985) who first used magnetostratigraphy in an at-tempt to calibrate the John Day faunal sequence tothe geomagnetic polarity timescale (GPTS). Impor-tant for its time, their study laid the groundworkfor future studies such as ours. More recent ad-


Figure 3. Location of main study area along the John Day River in Grant and Wheeler counties. “H & S, 2004”refers to Hunt and Stepleton (2004).

vances in geochronology, however, together with amore comprehensive suite of paleomagnetic sec-tions keyed to new radioisotopic and biostrati-graphic data, have greatly enhanced chronostrati-graphic precision. For example, the age of the PGIavailable to Prothero and Rensberger (1985) for cor-relation of their magnetostratigraphy with theGPTS was 25.3 Ma, based on conventional K-Aranalysis (Evernden et al. 1964). More recent 40Ar/39Ar single-crystal laser-fusion dating of sanidinefrom the PGI by C. C. Swisher III has indicated anage of 28.7 Ma—a significant difference. Althoughthe quality of our magnetostratigraphic data wasnot as desirable as anticipated, apparently as a re-sult of the widespread degree of zeolitization im-posed on, particularly, the Turtle Cove Member (asdetailed by Hay [1963] and discussed further below),we believe that we have ameliorated the bulk ofthese problems. Like Prothero and Rensberger’s ear-lier study, ours provides a new foundation fromwhich to continue work toward the overall goalsnoted above.

Methods and Caveats

As part of our reinvestigation of John Day Forma-tion magnetostratigraphy, we collected at leastthree fist-sized oriented samples (following thelong-accepted sampling procedure noted in, e.g.,

Lindsay et al. 1987) from 270 sites over approxi-mately 500 vertical meters of the Big Basin, TurtleCove, and Kimberly members in and around thePicture Gorge area of JDFBNM (fig. 3). Several sitesthat on analysis gave ambiguous polarity resultswere resampled in an attempt to refine the data,resulting in a total of 914 samples collected. Mostsamples consisted predominantly of tuffaceous silt-stone or mudstone. Bedding orientations weretaken at or near most sites for structural correc-tions, and dips typically measured !20�. Field sea-sons spanned 2–4-wk intervals primarily over thespring and fall months of 1997–1999. Eight strati-graphic sections were measured and sampled, eachof which at least partly overlapped to avoid am-biguities in local correlation. In stratigraphicallyascending order, the four main sections are BlueBasin (fig. 4A), Foree (fig. 5), Airport (fig. 5), andRoundup Flat (fig. 5). We also sampled a section(Richmond–Blue Banks) that overlaps the lowerpart of the Blue Basin section and two sections atSheep Rock (fig. 4B) that overlap parts of the Foreeand Airport sections. A section was also sampledat Bone Creek (PG-36 in Prothero and Rensberger1985), which includes the disconformable contactwith the overlying Rose Creek Member (part of therevised Haystack Valley Member in Hunt andStepleton 2004). Determination of overlap wasentirely unambiguous because of the presence of


Figure 4. A, Location of Blue Basin sections from whichsamples of lithostratigraphic unit A were obtained up tothe A-B Tuff. B, Location of Sheep Rock section fromwhich samples of unit F were obtained between the BlueBasin Tuff and the Picture Gorge Ignimbrite (PGI) andsamples of units G–J between the PGI and the DeepCreek Tuff. Picture Gorge East and West 7.5� topographicbase.

several diagnostic volcanic marker beds and dis-tinctive lithologic units, as documented by Fremdet al. (1994).

Most samples were cut into 21-mm cubes andthen incrementally demagnetized and measured ona 2-G Enterprises cryogenic magnetometer locatedin a magnetically shielded room at the Universityof Florida Paleomagnetics Laboratory and at Berke-ley Geochronology Center. After the natural rem-anent magnetization (NRM) of all samples wasmeasured, each was subsequently demagnetized bystepwise heating in 10–15 steps from 120�–150�Cto at least 500�C. As a test, a limited number ofsamples were subjected to alternating-field (AF) de-magnetization, but this technique failed to removelow-temperature, high-coercivity secondary com-ponents likely attributable to goethite. Therefore,some samples were heated to about 160�C to re-move the goethite component before AF demag-netization. Although this technique produced bet-ter results than the previous one, the best resultswere acquired using thermal demagnetization.

Our analyses, as well as those of Prothero andRensberger (1985), noted a sharp drop in intensityby 200�C. In many cases, the intensity of magne-tization dropped to nearly 50% of NRM by 130�Cand to significantly less than 50% by 180�C. Inten-sity was typically less than 10% by 400�C and lessthan 2% by 450�C. However, in sharp contrast tothe analysis of Prothero and Rensberger (1985), whonoted that a reversed component was first apparentat temperatures between 500� and 550�C, our anal-ysis found reversed behavior at temperatures as lowas 180�–230�C and that a consistent direction ofmagnetization was only rarely apparent at temper-atures above about 360�–450�C (fig. 6). Data gen-erated at higher temperatures were often spurious,thus calling into question the results of Protheroand Rensberger (1985), who initially demagnetizedtheir samples at 550�C, although blanket demag-netization with only limited thermal steps wascommon procedure at the time of their study. Ourmultistep analysis indicates that a single 550�Cstep would be above the unblocking temperatureof the major magnetic carrier in most cases, thusresulting in values that may not have accuratelycharacterized the geomagnetic polarity of the JohnDay rocks.

Although most samples were at least somewhatamenable to principal components analysis and, inturn, Fisher statistical analysis (Kirschvink 1980),many were not because of problems likely attrib-utable to zeolitization and a possible complex mag-netic mineralogy. Also, like the Prothero and Rens-berger study, our study resulted in very few class

Figure 5. Location of Foree, Airport, and Roundup Flat sections. Mt. Misery 7.5� topographic base.


Figure 6. Orthogonal vector plots of selected thermal demagnetization results from Turtle Cove units A and H inBlue Basin and Airport sections, respectively, representative of reversed-polarity (A–C) and normal-polarity (D) sam-ples. Open circles represent projection on the vertical plane (inclination); filled circles represent projection on thehorizontal plane (declination).

1 sites (following Opdyke et al. 1977), in which allthree samples yield a statistically significant con-cordant direction ( and for threea ! 15� k 1 3095

samples; Butler 1992 [p. 115 of 2004 electronic ver-sion, http://geography.lancs.ac.uk/cemp/resources/Butler_book/contents.htm]). Class 2 sites are thosein which two samples have similar magnetic vec-tors (the third sample either missing or divergentfrom the other two; Johnson et al. 1982). In thisarticle, class 3 sites are those in which only onesample yields a characteristic component, and class4 sites are those from which no characteristic vec-tor can be isolated but from which a polarity de-termination is discernible on the basis of the be-havior of the sample during demagnetization (fig.7). Classes 1–3 allow determination of a virtual ge-

omagnetic pole latitude for each site, and these 208(of 270) are plotted in figure 8. Of the total, only66 sites could be considered class 1 (and many ofthese may be present-field normal overprints),whereas the 142 class 2 and class 3 sites make upthe majority. Class 4 sites are not plotted.

Our data indicate that, contrary to the findingsof Prothero and Rensberger (1985), the character-istic remanence of rocks in the Turtle Cove Mem-ber is not carried in hematite or magnetite but bya more complex magnetic mineralogy with a rel-atively low unblocking temperature. The sharpdrop in intensity at very low temperatures and thespurious directional behavior above relatively lowtemperatures support this. To accurately charac-terize the mineralogy and better understand the


Figure 7. Equal-area projection of a site indicative ofclass 4 status, in which polarity can be determined onthe basis of demagnetization behavior (in this case re-versed) but from which no characteristic vector can beisolated.

complex behavior of these rocks as they undergodemagnetization, rock magnetic studies are re-quired.

On the basis of our preliminary dating of asso-ciated tuffs, our magnetostratigraphy of the PictureGorge area revealed several intervals of unusuallylong normal polarity for this time period. For ex-ample, in the Foree section (fig. 9), reversed chronC10r should occur stratigraphically almost imme-diately below the PGI on the basis of our 40Ar/39Ardate of 28.7 Ma instead of the long normal mag-netozones in intervals E2, E3, and F. Additionally,the Blue Basin Tuff (BBT), found about 30 m belowthe PGI, with a date of about 28.8 Ma, should fallwithin the upper part of reversed chron C10r. How-ever, our analysis revealed nearly 50 m of normal-polarity strata below the PGI, including correlativestrata in unit F of the Sheep Rock section (labeled“?overprint?” in fig. 9).

In our attempt to address this issue, we note thatHay (1963, p. 216) determined that part of the JohnDay Formation in the Picture Gorge region under-went severe zeolitic diagenesis about 24 m.yr. ago(based on K/Ar dates on celadonite from the PGI).To test whether zeolitization may be contributingto the paleomagnetic problems encountered in thePicture Gorge area, we conducted a pilot study onnonzeolitized rocks west of the Picture Gorge areain the Painted Hills sequence (fig. 2) studied by Hay

(1963). Analysis of seven samples collected fromHay’s (1963) section 6 in the Painted Hills area,stratigraphically correlative to the zeolitized Foreesection of the Picture Gorge area, yielded unam-biguous paleomagnetic signatures that included asecular variate normal direction for the PGI andreversed polarity for the ∼30 m below. Furthermore,thermal demagnetization of samples from the BBT(28.8 Ma) collected in both the nonzeolitizedPainted Hills area and the zeolitized Picture Gorgearea indicates that the normal directions measuredfor the tuff in the Picture Gorge area are spurious,likely as a result of zeolitization. It is interestingto note that Prothero and Rensberger’s (1985, theirfig. 6) analysis, like ours, also revealed a long in-terval of normal polarity below the PGI (which theycorrelated with chron C8) over which a long inter-val of reversed-polarity strata should have been re-corded. This further strengthens our suggestionthat the rocks in units F, E3, and upper E2 belowthe PGI in the Foree section, as well as those inunit F in the Sheep Rock section, have been alteredand strongly overprinted.

Given the abundance of radioisotopically datedtuffs in the John Day Formation, one might ques-tion whether magnetostratigraphic characteriza-tion of the sequence was even necessary for attain-ing the goals of this project (i.e., the detailedbiostratigraphic calibration), particularly given theproblems noted above. However, in many cases,only magnetostratigraphy allows biostratigraphiccorrelation and/or comparisons between the JohnDay Formation and, for example, lower ArikareeGroup rocks of Nebraska and South Dakota. Datedtuffs certainly strengthen and provide confidencein correlation of a local magnetostratigraphy withthe GPTS, but once such a correlation is attained,more refined calibration of various events can re-sult, including a refined, high-resolution chrono-stratigraphic foundation. Additionally, high preci-sion of radioisotopic dates can provide tests as tothe accuracy of chron boundaries. This study pro-vides such tests and such a chronostratigraphicfoundation.

Magnetostratigraphy

The following discussion details the magnetic po-larity zonation of each of the sampled stratigraphicsections and provides a correlation with the GPTS(of Lourens et al. [2004] and Luterbacher et al.[2004]) based primarily on radioisotopically datedtuffs in each section. Keeping in mind the problemsand caveats noted above, figure 9 provides the bestestimate of our correlation and accompanies the

Figure 8. Magnetostratigraphy of all transects sampled, with magnetic polarity plotted as virtual geomagnetic pole(VGP) latitudes (positive indicates normal polarity; negative indicates reversed polarity). Filled circles represent class1 sites; open circles represent class 2 and/or class 3 sites (Opdyke et al. 1977). Sites with no indicated latitude areeither class 4 sites or sites from which no magnetic direction could be determined. Sites are not connected by a solidline if not stratigraphically consecutive. Scale bar indicates stratigraphic separation between sites in meters.


Figure 9. Magnetostratigraphy of all transects sampled, with correlation to geomagnetic polarity timescale of Lourenset al. (2004 and references within) and Luterbacher et al. (2004 and references within). Boundaries of North Americanland mammal ages follow Luterbacher et al. (2004 and references within). A–M denote informal lithostratigraphicsubdivisions of Fremd et al. (1994). ABT p A-B Tuff; ATR p Across-the-River Tuff; BBT p Blue Basin Tuff; BT pBiotite Tuff; DCT p Deep Creek Tuff; PGI p Picture Gorge Ignimbrite; TRT p Tin Roof Tuff. See text for discussionof correlations.

following discussion. Although the Foree sectionpresented the most difficulties, the impact of zeo-litization and/or overprinting in all other sectionsappears to have been less significant with respectto any ambiguity of our correlation.

Relative to the results of Prothero and Rensberger(1985), our work places John Day strata exposed inthe Turtle Cove region earlier in time. Using a con-ventional K-Ar date for the PGI of about 26 Maresulted in Prothero and Rensberger’s correlationof the section from chron C8 through chronC6Cn.3n (Prothero and Rensberger 1985, their fig.6). The more recent 40Ar/39Ar date for the PGI ofabout 28.7 Ma, together with 40Ar/39Ar dates forseveral additional tuffs, results in a new correlationof this sequence that is more in accord with theage indicated by the mammalian biochronology.Our correlation indicates that the section spans aninterval that begins in upper chron C12n and endsin chron C7r or possibly C7n.1r (fig. 9).

Richmond–Blue Banks Section. The Richmond–Blue Banks section is located in the western halfof section 22, T10S, R22E, Toney Butte 7.5� USGSquadrangle. This section includes the major tran-sition in paleoenvironments that is marked by thered-hued clayed paleosols of the upper Big BasinMember below and the greenish and brownishhighly tuffaceous beds of the lower Turtle CoveMember above (Bestland and Retallack 1994). Thestratigraphically lowest sample was collected froma light buff-colored tuff in the upper Big Basin Mem-ber about 4 m above the local base of the section(fig. 9). Five additional samples were collected frommaroon and brown claystones and siltstones up toanother prominent white tuff that marks the localboundary with the Turtle Cove Member. Unfor-tunately, the lithology of neither tuff was amenableto 40Ar/39Ar age analysis. Eight more samples werecollected in Turtle Cove unit A up to the A-B Tuff(ABT) in order to provide overlap with the BlueBasin section discussed below. Fourteen sites weresampled over a stratigraphic interval of about 82m, which, because of the poorly indurated natureof several beds throughout, resulted in this beingthe only section where an average of 2 m betweensites was not obtainable.

The uppermost Big Basin Member and the lower∼14 m of the Turtle Cove Member in this sectionappear to be of normal polarity. Approximately 22m of reversed-polarity strata overlie the basal nor-mal interval, which, in turn, is overlain by the be-ginning of another normal interval in which theABT is located. Although the ABT in the Blue Basinsection, where it is best exposed, has not yielded areliable radioisotopic date, another sample (91CS-JD4) that was collected from what is believed to bethe ABT in the Painted Hills area of JDFBNMyielded an 40Ar/39Ar date of Ma. This29.75 � 0.02date places the ABT in chron C11n and thereforesupports correlation of the reversed 22 m in theRichmond–Blue Banks section with C11r.

Blue Basin Section. The Blue Basin section is lo-cated in the southern half of section 20, T11S,R26E, Picture Gorge East and West 7.5� USGS quad-rangles, approximately 5 km north of the JDFBNMheadquarters (fig. 4A). The section includes thelowermost part of the Turtle Cove Member, al-though the base is not exposed. Twenty-two sitesspanning approximately 50 m were originally sam-pled, with 10 additional sites collected to refinefirst-run analyses. Most of Turtle Cove unit A andall of unit B up to the unit B-C boundary were in-cluded in the sampled section, and the presence ofthe ABT allows correlation with the Richmond–Blue Banks section.

The lower ∼25 m of the section is of reversedpolarity, although there is a single site of apparentnormal polarity in unit A that may be spurious.The ABT lies near the base of an overlying normal,resembling the pattern seen in the Richmond–BlueBanks section, although in the latter, the ABT oc-curs about 5 m into this normal interval. Perhapsan unconformity immediately below the ABT inthe Blue Basin section accounts for the absence ofnormal-polarity strata immediately below it. Onthe basis of the thickness of reversed strata, the dateof the ABT, and the mammals from unit A, wecorrelate the reversed interval in the Blue Basinsection with C11r. A single-site reversal recordedin unit B, if not spurious, may represent C11n.1r.It follows, therefore, that the overlying normal in


which the unit B-C boundary occurs representsC11n.1n.

Foree (and Overlapping Sheep Rock Unit F) Section.The Foree section is located in the SE quarter ofsection 31, T10S, R26E, Mt. Misery 7.5� USGSquadrangle (fig. 5). The base of the section slightlyoverlaps the top of the Blue Basin section and in-cludes the B-C boundary. The Foree section alsoincludes Turtle Cove units D, E1–E3, and F, and itis capped by the PGI. Unit F, which is bounded bythe BBT below and the PGI above, was resampledin the Sheep Rock region of the JDFBNM (fig. 4B)because of the poor lithology and poor quality ofthe magnetic data generated from this unit in theForee section. Forty-two sites over an interval ofabout 100 m were originally sampled at Foree, with11 added later to address ambiguous first-run re-sults. Fifteen sites were sampled in unit F of theSheep Rock section over an interval of about 35 m.The Sheep Rock section is located in the NE quar-ter/SW quarter of section 5, T12S, R26E of the Pic-ture Gorge East and West 7.5� USGS quadrangles.

Consistent with the Blue Basin section, the unitB-C boundary in the Foree section, as well as all ofunit C, was determined to be of normal polarityand, as above, is correlated with C11n and likelyC11n.1n (fig. 9). A short interval of reversed polar-ity was recorded at site 7 in unit D, as was a similarsingle-site reversal at the unit D-E1 boundary.These may be spurious, or they may represent re-versals that are real but were not captured in themuch more slowly accumulating seafloor recordfrom which the GPTS was developed. They mayalso represent glimpses of basal chron C10r if thenormal sites of unit D and lower E1 are secondarynormal overprints. The correlation shown in figure9, with the base of the reversal in E1 correlated withthe base of C10r, considers the short reversal withinunit D to be spurious or not captured in the GPTS.

Stratigraphically higher is a relatively well-char-acterized interval of reversed polarity ∼19 m thickthat spans the E1-E2 boundary. On the basis of ourcorrelation up to this point, we consider this re-versal representative of C10r. However, an over-lying very long interval of normal polarity resultsin several correlation problems based on our pre-liminary dating of associated tuffs.

As noted in “Methods and Caveats,” reversedchron C10r should occur stratigraphically almostimmediately below the PGI on the basis of our 40Ar/39Ar date of Ma (Luterbacher et al.28.7 � 0.072004). Yet, our magnetostratigraphy revealed nearly50 m of normal-polarity strata between the top ofthe E1-E2 reversed interval and the base of the PGIin the Foree section. Furthermore, the BBT occurs

well within this long normal interval (about 30 mbelow the PGI), although it would be expected tofall within the upper part of C10r on the basis ofits 28.8 Ma date.

On the other hand, there are about 8 m ofreversed-polarity strata immediately below the PGIin unit F of the Sheep Rock section (fig. 9) that,accordingly, should correlate with C10r. Even inthis section, however, the BBT still does not fallwithin a reversed interval; it occurs more than 25m below the reversed interval in another long nor-mal interval that we consider spurious (labeled“?overprint?” in fig. 9). On the basis of the radio-isotopic dates of the PGI and BBT, the reversed in-terval immediately below the PGI in unit F of theSheep Rock section is likely a glimpse of uppermostC10r that was not as affected by zeolitization/nor-mal overprinting as the tens of meters below it.

Our correlation of the Foree section with theGPTS is therefore based more on radioisotopicdates and biochronological considerations than onmagnetostratigraphy. These considerations supportcorrelation of the reversal that straddles the E1-E2boundary in the Foree section with C10r, as wellas the overlying 50 m of normal-polarity strata upto the PGI, together with the normal- and reversed-polarity strata immediately below the PGI in unitF of the Sheep Rock section (fig. 9). This charac-terization of the Foree section also is compatiblewith our investigation of the same interval in thePainted Hills area (see “Methods and Caveats”).

Airport (and Overlapping Sheep Rock) Section. TheAirport section is located west of the John DayRiver and west of the Longview Ranch airstrip nearthe center of section 36, T10S, R25E, Mt. Misery7.5� USGS quadrangle (fig. 5). It is about 80 m thickand extends from immediately above the PGI toslightly above the Deep Creek Tuff (DCT). Twenty-seven sites were sampled in Turtle Cove units G–J between these two massive marker beds, one sitewas sampled from within the DCT, and anothertwo were sampled in the lowermost part of unitK1, just above the DCT, to provide overlap withthe Roundup Flat section discussed below. Thirty-one sites were also sampled over a stratigraphicallyequivalent section at Sheep Rock.

In the Airport section, the PGI appears to be ofnormal polarity, as are the overlying 12–14 m thatcomprise unit G. An interval of reversed polaritybegins about 3 m above the unit G-H boundary andcontinues upsection for about 18 m. The unit H-Iboundary falls within the lower ∼3 m of an over-lying long (26 m) normal interval that continuesthrough unit J, up to and including the DCT.The DCT, with an 40Ar/39Ar age (plagioclase) of


Ma, is about 7.5 m thick in this sec-27.89 � 0.57tion. Although the new 2004 GPTS (Lourens et al.2004; Luterbacher et al. 2004) indicates that theDCT should occur in upper C9r, its normal polaritysuggests correlation with C9n, with the reversedinterval below thus correlating with C9r. A verylong interval of normal polarity above the DCTfurther supports this correlation, which in turn ad-ditionally supports the previously suggested cor-relation of the normal interval in which occur thePGI and the overlying unit G to chron C10n.

Two sites were sampled in this section above theDCT at the base of unit K1. The first, collectedabout 2 m above the top of the tuff in a yellow- tobuff-colored interval about 4 m thick, appears to beof reversed polarity; the second, collected 2 mabove the first in green sediments, is of normalpolarity. However, at the same stratigraphic levelin the Roundup Flat section, no yellowish-coloredsediments were encountered, nor was a short in-terval of reversed polarity detected.

Results from equivalent strata sampled at SheepRock were similar, with the exception that the re-versed interval confined to unit H in the Airportsection extended into and included most of unit Gin the former. Note from figure 8 that there are noclass 1 sites extending through this interval in ei-ther of these sections. This discrepancy does not,however, affect our correlation with the GPTS.

Roundup Flat Section. Like the Airport section,the Roundup Flat section is also located on Long-view Ranch but on the east side of the John DayRiver in the NE quarter of section 19, T10S, R26E,Mt. Misery 7.5� USGS quadrangle (fig. 5). Ninetysites were sampled in this 170-m-thick sectionfrom the base of unit K1 just above the DCT, up-ward through K2 and L, and into unit M. In additionto the DCT, several other prominent radioisotopi-cally dated tuffs occur in this section, including theBiotite Tuff (BT) and the Tin Roof Tuff (TRT).

The lower 70 m of this section, which includesall of unit K1 and the lower part of K2, is of normalpolarity. The -Ma age of the DCT at27.89 � 0.57the base of this section, together with the

-Ma age of the BT at the ∼40-m level27.18 � 0.13(of the Roundup Flat section), supports an unam-biguous correlation of this long normal to C9n.Therefore, the 13 m of reversed-polarity strataoverlying this long normal and in the central partof K2 (unit K2 is bounded by the BT below and theTRT above) likely represents C8r. It follows thatanother long (53-m) normal in the upper part of K2should correlate with C8n.2n; this is corroboratedby the -Ma date of the TRT found in that25.9 � 0.3interval. Stratigraphically above this normal is a

∼12-m section of reversed polarity in the centralpart of unit L, a ∼14-m section of normal polaritythat spans the unit L-M boundary, and another ∼10m of reversed-polarity strata in unit M capping thesection. This reversed-normal-reversed sequence inthe upper part of the section is correlated withchrons C7Ar, C7An, and C7r, respectively.

Whereas the overall correlation of the RoundupFlat section appears correct in general terms, therelative thickness of the reversed strata correlatedwith C8r is substantially thinner than expected incomparison to its bounding normal intervals on thebasis of their relative lengths in the GPTS. Thissuggests either that accumulation rates were muchhigher in the normally magnetized parts of K1 andK2 (unlikely) or that some of the normally mag-netized samples immediately above and below thereversed part of K2 are overprinted.

Bone Creek Section. The Bone Creek section isequivalent to Prothero and Rensberger’s (1985) PG-36 section and is the northernmost and stratigraph-ically highest section sampled in our study (fig. 3).It is located approximately 2 km east of the JohnDay River in the central part of section 17, T10S,R26E, Mt. Misery and Miller Flat USGS 7.5� to-pographic maps. Twenty-five sites were sampled inthis section, in unit M, up to the major discon-formity that separates the Turtle Cove/Kimberlymembers of the John Day Formation from what waspreviously considered the Haystack Valley Member(in Fisher and Rensberger 1972). Hunt and Steple-ton (2004) recently revised these upper John Dayunits stratigraphically above the disconformity,and this is discussed further below.

Although rocks sampled over this intervalyielded relatively poor paleomagnetic signatures,the section appears to carry a predominantly re-versed polarity, which likely correlates with thereversed interval at the top of the Roundup Flatsection (fig. 9). However, Bone Creek is the onlysection sampled that lacks a distinctive lithologicmarker to provide unambiguous correlation toRoundup Flat. Within the lower 15 m of this ∼50-m-thick interval are two short intervals of normalpolarity, perhaps spurious. On the basis of the cor-relation with the GPTS of all sections previouslydiscussed and the assumption that this reversal cor-relates with that at the top of Roundup Flat, thelower ∼35 m of the Bone Creek section is thus con-sidered representative of chron C7r. The 10 moverlying this reversed interval is poorly charac-terized magnetostratigraphically but appears to beof normal polarity. If not a secondary, unremovablepresent-field overprint, this interval should cor-respondingly correlate with chron C7n.2n. The

Figu

re10

.


Figure 10. Biostratigraphy of selected taxa from the John Day Formation. Abbreviations of volcanic tuffs as in figure9. Boundaries of Arikareean subbiozones based on our revised interpretation. Note that the lettered lithostratigraphicunits of the Turtle Cove and Kimberly members are placed relative to the volcanic tuffs that bound them, notnecessarily to the magnetochrons with which they were found to correlate in this study. For example, units I and Jlie below the Deep Creek Tuff, as indicated, but they are of normal polarity and correlate with chron C9n. Note thatalthough several listed taxa occur in pre-Arikareean-age units in the Great Plains (i.e., those representative of theWhite River Chronofauna), chronostratigraphic ranges with down-pointing arrows refer only to taxa that have a pre–Turtle Cove Member first occurrence in the John Day Formation. Dotted lines indicate stratigraphic intervals fromwhich the noted taxon is not known.

overlying uppermost ∼6 m of this section (belowthe disconformity) is of reversed polarity, whichwe, in turn, correlate with chron C7n.1r.

Geochronology

On the basis of our geochronologic correlation,Turtle Cove Member unit A lies predominantlywithin C11r, although the base of the Richmond–Blue Banks section begins in C12n and the upper-most few meters may lie within C11n.2n (fig. 9).Unit A ranges from a little greater than 30.6 toabout 29.8 Ma (capped by the ABT at 29.75 �

) and therefore spans the interval of time oc-0.02cupied by the late Whitneyan NALMA, as currentlycorrelated in stratigraphic sections elsewhere inNorth America (Tedford et al. 2004). The mam-malian component of unit A supports this.

Units B–D and lower E1 are correlated withC11n.1n, and the single-site reversal in unit B mayrepresent C11n.1r (fig. 9). Problems noted abovewith the Foree section preclude an unambiguouscorrelation of units E and F with the GPTS, al-though the 28.7- and 28.8-Ma dates of the PGI andBBT, respectively, imply that much more of thislong interval (units E3 and F) than is indicated bythe magnetostratigraphy should be of reversed po-larity. If the single-site reversal in unit D representsthe base of C10r, then units D–F could encompassnearly all of C10r, or from about 29.4 to 28.75 Ma.This interpretation is borne out, in part, by themagnetostratigraphy of correlative nonzeolitizedsediments in the Painted Hills area, as discussedabove.

Between units F and G is the PGI. Its 28.7 �-Ma date, together with its normal magnetic0.07

polarity, places it at the base of C10n.2n, althoughthis date falls within uppermost C10r, according toLuterbacher et al.’s (2004) GPTS. Combining themagnetostratigraphy of the Airport section and thatfrom Sheep Rock between the PGI and the DCT

indicates that most of units G and H are of reversedpolarity, which we correlate with C9r (fig. 9).

Units I, J, K1, and the lower part of K2 appearto be entirely within chron C9n. Units I and Joccur below the DCT, radioisotopically dated at

Ma. On the basis of Luterbacher et27.89 � 0.57al.’s (2004) GPTS, this date places the DCT in up-permost C9r. However, the normal polarity of theDCT supports correlation with lower C9n, and the�0.57-Ma error of the date does not preclude this.This results in our correlation of normal-polarityunits I and J with chron C9n as well. Unit K1 liesbetween the DCT and the BT, and unit K2 is brack-eted by the BT below and the TRT above, dated at

and Ma, respectively, re-27.18 � 0.13 25.9 � 0.3sulting in an unambiguous correlation with C9n,C8r, and C8n.2n.

Between the TRT and the prominent erosionalunconformity above which rests what was previ-ously considered the Haystack Valley Member (seeHunt and Stepleton 2004) are units L and M, whichtogether comprise the Kimberly Member of theJohn Day Formation. Unit L extends from the mid-dle of C8n.2n into either C8n.1n or C7An, or fromabout 25.9 to about 25 Ma if the upper normal in-terval of unit L is correlated with the latter chron.The full temporal span of unit M is undeterminableas a result of the variable nature of the unconfor-mity between it and the overlying upper John Dayunits (Hunt and Stepleton 2004). However, our cor-relation of the upper part of the Bone Creek sectionbelow the unconformity (i.e., unit M in that sec-tion) with chrons C7r and C7n (about 24.9–24.1 Ma)is supported by the following.

First, in the Bone Creek section, resting discon-formably above unit M is the stratigraphicallyhighest of Hunt and Stepleton’s (2004) four newlyrecognized units, the Rose Creek Member. In de-scending order, these new units include the RoseCreek, Johnson Canyon, Balm Creek, and HaystackValley (revised) members. They are bounded aboveby the Columbia River Basalts and crop out mostlynorth of our study area, primarily between the Bone


Creek section and the town of Kimberly (figs. 1, 3,10). Before Hunt and Stepleton’s (2004) revision,and as noted above, the unit disconformablyoverlying unit M in the Bone Creek section wasconsidered the Haystack Valley Member (Fisherand Rensberger 1972). Hunt and Stepleton’s (2004)revision indicates, therefore, that the unconformityin the Bone Creek section represents a major in-terval of missing section including all of the John-son Canyon, Balm Creek, and Haystack Valley (re-vised) members, as well as an unknown thicknessof the upper Kimberly Member.

Second, the age of the Rose Creek Member wasconsidered by Hunt and Stepleton (2004) to beabout 18.2–18.8 Ma on the basis of an early Hem-ingfordian fauna recovered from the basal part ofthe member at the Bone Creek locality. We slightlyrevise the age of this fauna on the basis of prelim-inary magnetostratigraphic data of L. B. Albrightthat indicate that these rocks are reversely polar-ized, thus supporting correlation with either C5Eror C5Dr, i.e., from about 18.7–18.5 Ma or about18.1–17.6 Ma, respectively. In the lower part of theunderlying Johnson Canyon Member (missing inthe Bone Creek section as a result of the uncon-formity) is the Across-the-River Tuff (ATR) radio-isotopically dated by C. C. Swisher III to 22.6 �

Ma (Hunt and Stepleton 2004, pp. 37, 65), and0.13in the lowest of Hunt and Stepleton’s (2004, p. 64)new units, the revised Haystack Valley Member(also missing at the Bone Creek locality) is anothertuff (JD-BC-3) that dates to about 23.7 Ma.

Therefore, our correlation of unit M in the upperpart of the Bone Creek section with C7r and C7n(approximately 24.9–24.1 Ma) seems highly reason-able in consideration of the presence of early Hem-ingfordian mammals in beds immediately abovethe unconformity, in concert with strata dated to23.7 Ma in the revised Haystack Valley Memberabove unit M but missing in this section. The hi-atus spanned by the unconformity in the BoneCreek section therefore represents about 6 m.yr.(fig. 10).

Biostratigraphy and Biochronology

Until recently, the stratigraphic position of the fos-sil mammals of the John Day Formation has beenonly generally indicated (Merriam and Sinclair1907; Janis et al. 1998, app. I). Whereas relativelyrecent taxonomic appraisals have been provided formany groups of North American large mammals(Prothero and Emry 1996; Janis et al. 1998), thetreatments have been necessarily rather general,with only a few detailed revisions of John Day

groups, such as those on canids by Wang and col-leagues (1994, 1999; also see Bryant 1996; Wang andTedford 1996; Lander 1998; MacFadden 1998; Mar-tin 1998; Coombs et al. 2001). In order to facilitatea modern appraisal of this succession, Fremd et al.(1994), as noted above, proposed an informal divi-sion, based on lithostratigraphy, that divided thefossiliferous Turtle Cove and Kimberly membersinto several subunits labeled A–M (figs. 9–11). Thissubunit scheme has been recently and successfullyemployed as a basis for constructing the first de-tailed biostratigraphy of both large and small mam-mals of the John Day Formation.

Here we give preliminary biostratigraphic com-mentary for the key lithic (lettered) subdivisions ofFremd et al. (1994). We emphasize that the subdi-visions of the Arikareean (Ar1, Ar2, etc.) are basedon biostratigraphic characterizations of these in-tervals as developed primarily in the Great Plainsand proposed by Tedford et al. (1987, 2004), which,in turn, were calibrated on the basis of associatedradioisotopic and paleomagnetic correlations. Inthis context, the following evaluation of elementsof the Picture Gorge fossil mammal succession re-veals general compliance with the early Arikareeanbiozonations (Ar1, Ar2) developed in the GreatPlains (Tedford et al. 1987, 1996, 2004). Impor-tantly, however, our work significantly modifiesthe later Arikareean biozonations (Ar3, Ar4), andthis is discussed in detail below and shown in fig-ures 10 and 11. This modification, and consequentrecalibration, is a direct result of the fundamentalcompleteness of the John Day litho- and biostra-tigraphic succession relative to the unconformity-bounded succession of the Great Plains. Recentwork by T. J. Fremd has resulted in the discoveryin the John Day Formation of several new taxa andnew specimens of taxa previously only poorlyknown, new localities rich in both mammalian andpaleobotanical remains, and a far better under-standing of the lithostratigraphy and biostratigra-phy as a result of detailed studies of key sectionsthat are here precisely tied to a robust, high-reso-lution, verifiable geochronologic framework. Be-low, we therefore necessarily update, in part, themost recent overview discussion of John Day bio-stratigraphy and biochronology by Tedford et al.(2004). It is important to note, however, that thetaxonomic referrals on which our biostratigraphicassumptions are based (i.e., the alpha-level taxon-omy) are, in some cases, still tentative and in needof additional study (e.g., oreodonts). Figure 10 ac-companies the following discussion.

Units A and B. Within Turtle Cove unit A areseveral mammalian taxa typical of the late Whit-

Figu

re11

.R

evis

ion

ofA

rik

aree

ansu

bbio

zon

ebo

un

dari

es,a

sde

term

ined

from

Joh

nD

ayFo

rmat

ion

bioc

hro

nol

ogy,

and

aco

mpa

riso

nw

ith

trad

itio

nal

bou

nda

ries

ases

tabl

ish

edin

the

Gre

atP

lain

s.A

Ap

Aga

teA

sh;

EC

Ap

Eag

leC

rag

Ash

;R

RA

pR

oun

dhou

seR

ock

Ash

;SC

Tp

Shee

pC

reek

Tu

ff;

SRA

pSp

lit

Roc

kA

sh.


neyan and earliest Arikareean biochrons. Taxa ofpotential biochronological importance makingtheir first appearance in unit A include the mar-supial Herpetotherium merriami, eomyid and eu-typomyid rodents, the nimravids Hoplophoneusstrigidens and Nimravus brachyops, the canids En-hydrocyon prolatus and Phlaocyon latidens, the ur-sid Allocyon loganensis, the amphicyonid Tem-nocyon altigenus, the equids Mesohippus andMiohippus, the rhinocerotid Diceratherium (truefirst occurrence in the John Day Formation is inthe upper part of the underlying Big Basin Member),the entelodont Archaeotheirum caninus, thehypertragulid Hypertragulus hesperius, the agri-ochoerids Agriochoerus macrocephalus and Agri-ochoerus guyotianus, the oreodont Eporeodon oc-cidentalis, and the peccaries Perchoerus probus andThinohyus lentus. (Note: recent studies by D.Prothero indicate that only Thinohyus occurs inthe John Day Formation and that Perchoerus ap-pears restricted to the White River Group [D. Proth-ero, pers. comm., 2007]. T. J. Fremd, on the otherhand, maintains that both genera are present in theJohn Day Formation. It should also be noted thatthe lowest stratigraphic datum given above for taxain the lowermost Turtle Cove Member may notnecessarily represent their first-appearance datumin the John Day Formation. Rather, this may be ataphonomical artifact imposed by what appears tobe poor preservational potential of the underlyingBig Basin Member.)

Several taxa characteristic of the White RiverChronofauna occur in unit A, such as Hoplopho-neus, Nimravus, Archaeotherium, E. occidentalis,A. guyotianus, Perchoerus, and Mesohippus (Bryant1996; Wang and Tedford 1996; Janis et al. 1998;Martin 1998; Prothero and Emry 2004; Tedford etal. 2004); E. occidentalis and A. guyotianus areknown only from Whitneyan faunas in the GreatPlains.

Combining lithostratigraphic, biostratigraphic,and magnetostratigraphic data from Nebraska andSouth Dakota, Tedford et al. (1996, p. 329, 2004;also see Emry et al. 1987) placed the Whitneyan-Arikareean boundary at about 30 Ma on the basisof a distinct enrichment of the White River Chro-nofauna that began in late C11r and early C11n andcontinued until a “true faunal turnover with ex-tinction and origination components” occurred atabout 28 Ma “effectively end[ing] the White RiverChronofauna through extinction of many of itscharacteristic elements” (i.e., beginning of Ar2).Tedford et al. (1996, 2004) also suggested that, forthe Great Plains at least, the co-occurrence of sur-viving Whitneyan forms and the new “enrich-

ment” taxa could serve to characterize the earliestpart of the Arikareean (Ar1) but that the rodentPlesiosminthus, being the lone immigrant at thistime, should be the defining taxon. Plesiosminthusdoes not occur in unit A (or B) in the John DayFormation. Conspicuously absent, in fact, are sev-eral other taxa typically considered characteristicof the earliest Arikareean, such as Palaeolagus, Al-lomys, Alwoodia, Palaeocastor, Kirkomys, Leidy-mys, Ocajila, and Nanotragulus (Tedford et al.1996, 2004; Prothero and Emry 2004; Hayes 2007).Allocyon loganensis and P. latidens are known onlyfrom the John Day Formation (Hunt 1998b; Wanget al. 1999), thereby precluding biochronologicalsignificance with respect to regional correlation.Although Agriochoerus, Hypertragulus, and Per-choerus last occur in Ar1 of the Great Plains, theabsence of taxa definitive and/or characteristic ofthe early Arikareean, in concert with an abundanceof White River forms, supports a late Whitneyanage, compatible with the radioisotopic and paleo-magnetic framework (correlation to C11r) as well(figs. 9–11).

Units C and D. Unit C records the first appear-ance of the canid Rhizocyon oregonensis, whichWang et al. (1999, fig. 141) indicated is not knownbefore the Arikareean. This occurrence, togetherwith our correlation of unit C with upper chronC11n.1n based on the 29.75-Ma date of the ABT(fig. 9), suggests that age for the Whitneyan-Arikareean boundary in the John Day Formation—only slightly later than the same boundary in theGreat Plains. The first appearance of R. oregonensiscould be used to define the beginning of Ar1 in thePacific Northwest.

Additional taxa from these units also support cor-relation with Ar1, including Cynarctoides lemurand Paraenhydrocyon josephi. Although the ear-liest occurrence of C. lemur is in the WhitneyanPoleside Member of the Brule Formation (Wang etal. 1999), it also occurs in the Ar1 Sharps Forma-tion, South Dakota, slightly above the ∼30-MaRockyford Ash. This occurrence is just slightly ear-lier than its first John Day appearance in unit C,which, like unit B, we correlate with C11n.1n.Paraenhydrocyon josephi ranges from the Whit-neyan of South Dakota through most of theArikareean (Wang 1994), and the nimravid Eus-milus cerebralis, also known from the Whitneyanof South Dakota, first appears in and is confined tounit D in the John Day Formation at approximately29.5 Ma. Tedford et al. (2004) noted that this taxonlast occurs in the Great Plains in Ar1. Hunt (1998a,p. 205) also noted the presence of three species ofDaphoenus from the “Whitneyan–early Arika-


reean” of the John Day Formation. Known primar-ily from the Duchesnean through Whitneyan of theGreat Plains, the stratigraphic position of two ofthe John Day species (including Daphoenus trans-versus) was not accurately known until the veryrecent work of R. M. Hunt and T. J. Fremd (unpub.data) shed new light on their John Day occurrences.In addition to confirming the presence of a verylarge new species of Daphoenus from unit C, theyhave also determined that the genus extends upinto unit K2.

The only taxon in this interval characteristic ofAr1 in the Great Plains is the aplodontid rodentAllomys (Tedford et al. 1987, p. 184), which firstappears in unit D. Bailey (2004) reported Allomysfrom the earliest Arikareean Ridgeview local fauna,Nebraska, which, like the Wagner Quarry localfauna from the Pine Ridge area (Hayes 2007), wasrecovered from basil Arikaree Group strata abovethe Brown Siltstone beds.

Units E and F. Unit E, the most fossiliferous unitbetween the ABT and the PGI, yields taxa indica-tive of Ar1, as well as several members of the WhiteRiver Chronofauna found in lower units. The nim-ravids Dinictis, Pogonodon, and Dinaelurus andthe hesperocyonine canid Osbornodon sesnoi, allof which are known from older rocks in the GreatPlains and two of which (Dinictis and Osbornodon)mark the end of the Whitneyan in the midconti-nent (Prothero and Emry 1996), first appear in unitE. Additional first occurrences in unit E include therodents Haplomys liolophus and Palaeocastor pen-insulatus; the canids Cormocyon copei, Mesocyoncoryphaeus, Mesocyon brachyops, and Leptocyonmollis; the oreodont Eporeodon bullatus; and whatappears to be the oldest record of the hypertragulidNanotragulus (Nanotragulus planiceps). In theGreat Plains, Palaeocastor last occurs in Ar1. An-other first occurrence in unit E is the leporid Ar-chaeolagus ennisianus, which is not known untilAr2 (C9n) in the Great Plains (Tedford et al. 2004),although Bailey (2004) notes that Archaeolagusfirst appears in Ar3 in Nebraska. Thus, the JohnDay appearance of this taxon in C10r represents aconsiderable downward extension of its range.

Continuing from lower John Day strata are R.oregonensis, C. lemur, P. latidens, and P. josephi;the oreodont E. occidentalis; the Whitneyanholdovers A. guyotianus, Hypertragulus, and Ar-chaeotherium; and the perissodactyls Miohippusand Diceratherium. Although Palaeocastor andNanotragulus are characteristic of the earliestArikareean in the Great Plains, still missing fromthe John Day sequence are Ar1 contemporariessuch as Palaeolagus, Alwoodia, Plesiosminthus,

and Leidymys. In summary, the faunas from unitsC–F continue to characterize Ar1 (figs. 9–11).

Units G–J. Taxa from units G through J, strati-graphically between the PGI ( Ma) be-28.7 � 0.07low and the DCT ( Ma) above, con-27.89 � 0.57tinue to typify the early Arikareean. Making theirfirst appearance in this interval are the rodents Al-woodia magna, Meniscomys uhtoffi, Capicikalagradatus, Florentiamys lulli, and Pleurolicus sul-cifrons; the primate Ekgmowechashala; the canidPhilotrox condoni; and the large oreodonts Mery-cochoerus minor and Oreodontoides oregonensis.Taxa that continue from lower units include Al-lomys, R. oregonensis, M. brachyops, Enhydrocyon(as Enhydrocyon stenocephalus), C. copei, N. plan-iceps, E. occidentalis, Miohippus, and Dicerathe-rium. Nimravus brachyops and H. hesperius lastoccur in this interval (although Hypertragulus cal-caratus continues upsection).

Tedford et al. (2004, p. 211) noted that the firstappearance of A. magna in the John Day Formationhelped to characterize Ar1 but that in the GreatPlains, the first appearance of Alwoodia (Alwoodiaharkseni) was characteristic of the “second phase”of the early Arikareean (Ar2). Tedford et al. (1996,2004) concluded, on the basis of a significant faunalturnover event that effectively ended the WhiteRiver Chronofauna, that this second phase beganat about 28 Ma. Hayes (2007, pp. 18–19), however,recently noted a record of Alwoodia from the baseof the Brown Siltstone member of the Brule For-mation in Nebraska (∼C11n.2n), which places thegenus in the Great Plains approximately 1 m.yr.earlier than in the John Day region (∼C10n.1n).Hayes (2007) also noted Alwoodia cf. A. magnafrom the Wagner local fauna, Nebraska, which hecorrelated with C10n, essentially contemporane-ous with John Day unit G (C10n). Thus, the firstappearance of Alwoodia in both regions appears tocharacterize Ar1, although A. harkseni is appar-ently indicative of Ar2 in the Great Plains (Tedfordet al. 2004, p. 211).

Tedford et al. (2004) also included the first oc-currence of Florentiamys, Pleurolicus, and O. or-egonensis as characteristic of Ar2 in the GreatPlains, as well as that of A. ennisianus, which, asnoted above, occurs earlier in the John Day region(in unit E, C10r). Like Hayes’s (2007) work notedabove, Wahlert (1983) and Bailey (2004) also re-corded Florentiamys in faunas of Ar1 age in Wyo-ming and Nebraska. Thus, the first appearance ofFlorentiamys, like that of Alwoodia, appears to beindicative of Ar1 rather than Ar2. Two other taxaconsidered characteristic of Ar1 in the Great Plainsthat first appear in unit G of the John Day succes-


sion, C. gradatus and Ekgmowechashala, appear atnearly coincident times in both regions (i.e., C9r;regarding the latter, see Albright 2005). On theother hand, the unit G–J first appearance of Pleu-rolicus and Oreodontoides, both of which charac-terize Ar2 in the Great Plains, is nearly 1 m.yr.earlier in the John Day Formation (C10-C9r in theJohn Day region vs. C9n in the Great Plains). Alsointeresting is the much earlier appearance of Ar-chaeolagus in John Day unit E (∼29 Ma) than inthe Great Plains (post–28 Ma, where it character-izes Ar2) and its absence in units G–J.

In summary, although most taxa recorded in JohnDay units G–J still suggest an Ar1 age for this in-terval, the first appearance of Pleurolicus andOreodontoides, in concert with the last record inthe John Day Formation of Nimravus, heralds theupcoming transition from Ar1 to Ar2. Given themargins of error for the radioisotopic dates of theDCT in the John Day Formation ( Ma)27.89 � 0.57and for the Roundhouse Rock Ash near the top ofthe Gering Formation in Nebraska (28.11 � 0.18Ma), it appears that Ar2 begins at nearly the sametime in the two regions, i.e., about 28 Ma (figs. 10,11).

Unit K. Unit K includes nearly 120 m of stratabounded by the DCT below and the TRT (25.9 �

Ma) above (fig. 9). The BT ( Ma)0.3 27.18 � 0.13separates the lower one-third of the section, unitK1, from the upper two-thirds that comprise unitK2. Unit K continues to record the persistence ofpredominantly early Arikareean taxa. Those ap-pearing for the first time include the rabbitPalaeolagus haydeni and the rodents Prosciurus,Protospermophilus, Paradjidaumo, Entoptychus(Entoptychus basilaris and Entoptychus wheeler-ensis), Leidymys, and Paciculus; the canid Paraen-hydrocyon wallovianus; the amphicyonid Para-daphoenus cuspigerus; the insectivores Ocajila,Proscalops, and Domninoides; the camels Gentil-icamelus and Paratylopus; the hypertraglid H. cal-caratus; and the oreodonts Hypsiops, Merycocho-erus superbus, “Promerycochoerus” macrostegus,Merycoides, and Paroreodon parvus. We note herethat although the nimravid Pogonodon platycopisis confined to units E and F, what appears to be anew species of this genus is now known from unitK2 (H. Bryant and T. J. Fremd, unpub. data).

Continuing from lower strata are Capacikala,Florentiamys, Meniscomys, and Allomys; the ca-nids M. brachyops and C. copei; the amphicyonidTemnocyon (recorded from A–D but absent in F–J);the entelodont A. caninus; Hypertragulus (as H.calcaratus vs. H. hesperius in lower strata); the or-

eodonts Eporeodon and Oreodontoides; and the pe-rissodactyls Miohippus and Diceratherium.

Of the above taxa, those that last occur in unitK include Capacikala, Paradjidaumo, Paciculus,Protospermophilus, Paraenhydrocyon, Mesocyon,Cormocyon, Paradaphoenus, Temnocyon, Ocajila,Archaeotherium, Eporeodon, Oreodontoides, andDiceratherium (although regarding the latter taxon,see discussion in “Overlying Units”). In the GreatPlains, Palaeolagus, Proscalops, and Archaeothe-rium last occur in Ar2 (Tedford et al. 2004, p. 211).

Taxa from unit K that first appear in Ar2 in theGreat Plains (keeping in mind the range extensionsdown into Ar1 for Alwoodia and Florentiamysnoted above) include Paciculus, Entoptychus, andOreodontoides, the latter of which occurs belowunit K (i.e., below the DCT) in the John Day For-mation. Species of Entoptychus considered some-what more primitive than those from John Day For-mation occur in early Ar2 beds in eastern Idaho(Tedford et al. 2004, p. 187), suggesting that theentoptychines, so characteristic of and apparentlylargely restricted to the John Day region after about28 Ma, may have a slightly earlier trans–RockyMountains source to the east.

Other taxa that first appear in unit K, such asOcajila and Leidymys, are known from Ar1 faunasin the Great Plains. Ocajila, for example, last oc-curs in (late) Ar1 faunas in the Great Plains, havingbeen found at the base of the Gering Formation andin the Ridgeview local fauna, Nebraska, and in theupper Sharps Formation, South Dakota (C10n.1n–C9r), whereas Leidymys is known from the earlyAr1 Brown Siltstone member of the Brule Forma-tion (C10n; Tedford et al. 1987, 2004; Bailey 2004;Hayes 2007). The unit K presence of Leidymys pre-dates the unit L occurrence noted by Tedford et al.(2004), but this taxon still occurs much later in theJohn Day Formation (C9n) than in the Great Plains.

Paraenhydrocyon wallovianus is known in theGreat Plains from the late Arikareean (Ar3), ac-cording to Wang (1994), and Hypsiops is consideredlimited to the late Arikareean (Tedford et al. 1987,p. 185). In contrast, however, is the unit K appear-ance of H. calcaratus, which occurs substantiallyearlier (Chadronian through Ar1) in the GreatPlains. Overall, faunas from unit K appear to cor-relate with those of late early Arikareean (Ar2) age,especially as shown by the first appearance ofPaciculus and Entoptychus and the last occurrenceof Archaeotherium.

Units L and M. Above the TRT is unit L, which,together with overlying unit M, comprises the Kim-berly Member. The fauna from the Kimberly Mem-


ber is derived predominantly from unit M but alsoincludes taxa from unit L, although the two unitsare in some places difficult to distinguish becauseof the facies context of the Kimberly and TurtleCove members (Hay 1963). Taxa making theirfirst appearance in this interval include the lago-morph Desmatolagus; the rodents Campestral-lomys, Schizodontomys, Proheteromys, Protosciu-rus, and several species of Entoptychus; theinsectivores Scalopoides and Domnina; the mus-telid Oligobunis; the amphicyonid Daphoenodon;the artiodactyls Agriochoerus bullatus, Agriocho-erus trifrons, Poebrotherium, Daeodon, Leptome-ryx; and the perissodactyls Moropus, Anchither-ium, and Kalobatippus (from unit L). Of these,Desmatolagus, Campestrallomys, Proheteromys,Protosciurus, Oligobunis, Scalopoides, Domnina,Poebrotherium, Leptomeryx, and Anchitheriumare currently known only from the KimberlyMember.

Taxa that persist into the Kimberly Member fromearlier units include the marsupial H. merriami;the lagomorph Palaeolagus; the rodents Allomys,Prosciurus, Florentiamys, E. wheelerensis, and Lei-dymys; the insectivores Proscalops, Domninoides,and Micropternodus; the artiodactyls Agriocho-erus, Gentilicamelus, Hypertragulus, Eporeodon,Hypsiops, Merycochoerus, Merycoides, O. orego-nensis, P. parvus; and the perissodactyl Miohippus.Of these, those that last appear in the John DayFormation are Herpetotherium, Palaeolagus, Al-lomys, Florentiamys, Leidymys, Proscalops, Dom-ninoides, the Agriochoeridae, Gentilicamelus, Hy-pertragulus, Eporeodon, Promerycochoerus, andOreodontoides. Although Tedford et al. (2004, p.179) noted that Diceratherium is not known abovethe TRT and it is apparently absent in units L andM, Hunt and Stepleton (2004, p. 49) recorded itsreappearance in upper John Day units; we discussthis further below.

Of the taxa noted above from the Kimberly Mem-ber, Moropus is considered by Tedford et al. (2004)as definitive of the early late Arikareean, Ar3, andthey additionally include the earliest occurrence ofDaphoenodon and Kalobatippus as characteristicof that biochron. According to Bailey (2004),Schizodontomys also first appears during Ar3 inNebraska. Tedford et al. (2004) noted that the flor-entiamyids, Hypertragulus, Eporeodon, Promery-cochoerus, Hypsiops, and Oreodontoides last occurin Ar3, as does Temnocyon, which last occurs inJohn Day unit K. No taxa indicative of Ar4, as char-acterized in the Great Plains, occur in the KimberlyMember.

By these criteria, the faunas occurring strati-graphically above the TRT in the Kimberly Mem-ber, particularly the occurrence of Kalobatippus inunit L, mark the beginning of Ar3 in the John DayFormation; i.e., the late Arikareean there begins atabout 25.9 Ma. This is in marked contrast to theGreat Plains, where the beginning of Ar3 is placedat about 23 Ma (fig. 11) and remains so even if weplace this boundary in unit M at about 24.9 Ma onthe basis of the first appearance of Moropus. How-ever, the 23-Ma beginning of Ar3 in the GreatPlains is an artifact of geology and is not based ona turnover event mapped biostratigraphically in thelocal sequence. Rather, it is due to the well-knownstratigraphic hiatus between the Monroe Creek andHarrison formations within which the transitionfrom Ar2 to Ar3 has long been hypothesized to haveoccurred (Tedford et al. 1996, 2004; MacFadden andHunt 1998; Albright 1999). Moreover, it also ap-pears to be due in part to the mistaken concept thatthe Agate Ash ( Ma) occurs near the22.9 � 0.08base of the Harrison Formation, which harbors thefauna that characterizes Ar3 (e.g., Tedford et al.2004, p. 220, fig. 6.2), when, in fact, this ash occurscloser to the top of the formation (Hunt 1990;MacFadden and Hunt 1998; R. M. Hunt, pers.comm., 2007), as we show in figure 11. This, inturn, has the effect of pushing the hiatus betweenthe Harrison and Monroe Creek formations furtherback in time. Regarding, in part, this interval in theJohn Day sequence, Tedford et al. (2004, p. 179)noted that “[t]he occurrence and succession of taxain the John Day Formation above the DCT repre-sent a span of time that has not been clearly delin-eated biochronologically in the Great Plains.” Thenearly continuous, robust, and well-calibrated bio-stratigraphic record in the John Day Formation fi-nally fills this interval missing from the GreatPlains sequence and thus provides what we con-sider to be a more accurate date for the beginningof the late Arikareean (Ar3) at about 25.9 Ma (fig.11).

Overlying Units. Overlying the Kimberly Mem-ber is a complex sequence of additional fossiliferousstrata that, before recent work, were collectivelyreferred to as the Haystack Valley Member (Fisherand Rensberger 1972). In most places within thestudy area, the contact between the KimberlyMember and these overlying strata is recognized asan erosional unconformity typically marked bybasal welded tuff conglomerates (Fisher 1967; Ste-pleton and Hunt 1994; Hunt and Stepleton 2004).A major effort to more clearly understand thesestrata was recently completed by Hunt and Steple-


ton (2004), with the result that at least four distinct,mappable geologic units are represented. As notedabove, these new units, in descending order, com-prise the Rose Creek, Johnson Canyon, Balm Creek,and Haystack Valley (revised) members (figs. 3, 10).Although our magnetostratigraphic study was pri-marily confined to the Turtle Cove and Kimberlymembers, we include a discussion of these over-lying units because of the bearing they have on thelater part of the Arikareean NALMA.

The fauna from the Haystack Valley Member wasoriginally referred to the Mylagaulodon concur-rent-range zone of Fisher and Rensberger (1972).With greater understanding of these supra-Kim-berly units, both structurally and biostratigraphi-cally, we recommend abandonment of this termi-nology. Hunt and Stepleton (2004, p. 64, table 4)determined that the revised Haystack Valley Mem-ber (missing at the Bone Creek locality, the north-ernmost and stratigraphically highest unit of ourstudy; fig. 3) yields a “reasonably abundant earlylate Arikareean [Ar3] mammal fauna,” includingthe following taxa: an advanced species of Schi-zodontomys or early species of Tenudomys, Entop-tychus individens, Allomys tessallatus, Paroreo-don cf. marshi, cf. Hypsiops, cf. Nanotragulus,Paratylopus, cf. Hesperhys, Cynorca sociale, Mio-hippus, Diceratherium, and Nexuotapirus robus-tus. Additionally, a tuff (JD-BC-3; fig. 10) associatedwith the lower part of the unit produced an 40Ar/39Ar age of about 23.8 Ma (Hunt and Stepleton 2004,p. 64). Although numerically the 23.8-Ma date oc-curs within upper Ar2 as established in the GreatPlains, our revision of the Ar2-Ar3 boundary, asdiscussed above, brings this Ar3 fauna into confor-mity with our new hypothesis, i.e., that Ar3 beginsat about 25.9 Ma (fig. 11).

Unconformably overlying the revised HaystackValley Member is the Balm Creek Member (fig. 10),from which, thus far, only cf. Hesperhys has beennoted (Hunt and Stepleton 2004, table 4). An Ar3age for this unit is compatible with that taxon.

Hunt and Stepleton’s (2004) Johnson CanyonMember rests unconformably on the KimberlyMember, but temporally it spans an interval oc-cupied by the unconformity between the BalmCreek and overlying Rose Creek members (figs. 10,11). The lower part of the Johnson Canyon Membercontains the -Ma ATR and a mamma-22.6 � 0.13lian fauna younger than that noted above from therevised Haystack Valley Member (of Ar3 age). Huntand Stepleton (2004, pp. 37, 45, 65) noted that thefauna occurs both above and below this tuff, andthey considered it of “late to latest Arikareean [Ar4]

age.” Mammalian taxa from the Johnson CanyonMember include the rodents Schizodontomysgreeni and Mylagaulodon angulatus (in addition toan advanced species of Schizodontomys discussedbelow); the canids Cynarctoides cf. luskensis andDesmocyon thomsoni; the oreodonts Paroreodon(“advanced species”) and cf. Promerycochoerus; thecamelid “Paratylopus” cameloides; the moschidPseudoblastomeryx cf. advena; Hesperhys; and theperissodactyls Kalobatippus, Archaeohippus, Mor-opus oregonensis, Diceratherium, cf. Menoceras,and Miotapirus harrisonensis. The advanced spe-cies of Schizodontomys noted above compares fa-vorably with Schizodontomys cf. harkseni from theUva fauna of Platte County, Wyoming, which Huntand Stepleton (2004, p. 47) note “is one of the youn-gest Arikareean assemblages yet discovered in thetype area of the Arikareean NALMA.” In addition,the first appearance of a dromomerycid and themoschid deer P. cf. advena in the Johnson CanyonMember, as well as an advanced species of the or-eodont Paroreodon and the camel “P.” cameloides,all support a post–Haystack Valley Member age.Conspicuously absent from the Johnson CanyonMember are taxa from the underlying HaystackValley Member that would be indicative of an Ar3age, such as Miohippus, entoptychine or allomyinerodents (Entoptychus and Allomys), and hypertra-gulids. Similarly absent are taxa suggestive of anearly Hemingfordian age, such as parahippineequids, which are found in the stratigraphicallyhigher Rose Creek Member. These criteria firmlyestablish the fauna from the Johnson Canyon Mem-ber as Ar4 in age.

On the basis of the presence of a fauna indicativeof Ar4 associated with a tuff in the lower part ofthe Johnson Canyon Member dated to about 22.6Ma, together with a fauna indicative of Ar3 that isassociated with a tuff in the upper part of the Har-rison Formation dated to about 22.9 Ma, we revisethe Ar3-Ar4 boundary to a time between thesedates, or to approximately 22.7–22.8 Ma (fig. 11).Thus, the faunas that extend from units L and Mto the Johnson Canyon Member, together with thedates of tuffs within, allow a revised estimate forthe range of the early late Arikareean (Ar3) in theJohn Day Formation from about 25.9 to about 22.7–22.8 Ma.

Further supporting our revised Ar3-Ar4 boundaryis the following. Note that the end of Ar3 in theGreat Plains (base of Ar4) is placed at about 19.5Ma by Tedford et al. (2004). This is not based onthe fauna from the Harrison Formation of westernNebraska that characterizes Ar3 extending up to


that point in time. Rather, it reflects the 19.2 �-Ma age of the Eagle Crag Ash associated with0.5

taxa that characterize Ar4 found near the base ofthe unconformably overlying Anderson Ranch For-mation (Hunt 2002; Tedford et al. 2004). In fact,Hunt (2002, fig. 28) shows the Ar3-Ar4 boundarywithin a hiatus between the Harrison and AndersonRanch formations. Moreover, there appears to beonly a single, uniform fauna from the Harrison For-mation (R. M. Hunt, pers. comm., 2007), much ofwhich is derived from below the -Ma22.9 � 0.08Agate Ash, which, as emphasized above, occurs inthe upper part of the formation, as we show in fig-ure 11, not near the base, where it had previouslybeen noted. In figure 11, we revise the correlationof the magnetostratigraphy of Arikaree rocks byMacFadden and Hunt (1998) on the basis of Lourenset al.’s (2004) Neogene GPTS. This places the EagleCrag Ash within chron C6n and the Agate Ashwithin C6Cn.2n. Thus, another substantial inter-val of time may be missing in the above-noted hi-atus between the Harrison and Anderson Ranch for-mations, in addition to the hiatus between theMonroe Creek and Harrison formations.

Germane to this discussion is the recent mag-netostratigraphic work on Arikaree Group rocks ofwestern Nebraska by F. G. Hayes (unpub. data) whonoted, in contrast to MacFadden and Hunt (1998),that “[t]he unconformity represented by the ter-minal Harrison [Formation] paleosol … may in-clude a significant amount of missing time.” Ad-ditionally, Hayes noted the following with respectto the magnetostratigraphy of the Harrison For-mation: “Unfortunately, magnetostratigraphy isnot currently an applicable correlation tool for theHarrison Fm. This formation, at least in the Mon-roe Creek Canyon section, is too altered throughbioturbation, pedogenesis, and diagenesis to pre-serve coherent depositional remanent magnetic sig-natures,” thus calling into question the magneto-stratigraphic results for the Harrison Formation ofMacFadden and Hunt (1998). On the other hand,Hayes noted that “magnetostratigraphic signaturesare found to be reliable in calcareous sedimentsfrom the formations of the lower Arikaree Group.”Importantly, the data discussed here underwriteour proposal to revise and to bring into conformitythe intervals of time included within the Ar2–Ar4biochrons in both the John Day and the GreatPlains regions.

The uppermost of Hunt and Stepleton’s (2004)revised units and the stratigraphically highest unitof the John Day Formation is the Rose Creek Mem-ber. It is this unit that rests unconformably on unit

M at the Bone Creek section (PG-36), the northernlimit of our study area (fig. 9), which we reem-phasize was considered the Haystack Valley Mem-ber before the work of Hunt and Stepleton (2004).The mammals from this locality were designatedthe Picture Gorge 36 local fauna by Hunt and Ste-pleton (2004, p. 47) and include the rodents My-lagaulodon, cf. Hystricops, and Sewellelodon cf.Sewellelodon predontia; the canid Desmocyonthomsoni; cf. Amphicyon and Daphoenodon n. sp.(R. Hunt, pers. comm., 2008); the oreodont Mery-cochoerus magnus; the moschid Pseudoblastome-ryx schultzi; the camelids “Paratylopus” and cf.Protolabis; the dromomerycid Barbouromeryx; theequids Parahippus pawniensis, cf. Kalobatippus,and Archaeohippus; the chalicothere M. oregonen-sis; and the rhinoceros Diceratherium. This faunais clearly of early Hemingfordian age and compareswith the Northeast of Agate local fauna from Ne-braska (MacFadden and Hunt 1998, p. 163), whichlies stratigraphically above the 19.2-Ma Eagle CragAsh.

MacFadden and Hunt (1998) calibrated theNortheast of Agate local fauna paleomagneticallyto chron C5En, which, according to the GPTS ofLourens et al. (2004), spans about 18.1–18.5 Ma. Incontrast, preliminary magnetostratigraphic analy-sis of the Rose Creek Member at the Bone Creeklocality by L. B. Albright indicates that the PG 36local fauna rests in rocks of reversed polarity (fig.9). Combined with the early Hemingfordian fauna,this reversed polarity supports correlation with ei-ther uppermost C5Er (∼18.5–18.8 Ma) or C5Dr(∼17.5–18.1 Ma), based on the correlation of theAnderson Ranch Formation, with its Ar4 fauna,with most of C5Er by MacFadden and Hunt (1998).Thus, the Ar4-He1 boundary is slightly revised toabout 18.5 Ma.

In summary, the detailed biostratigraphic datapresented above for the Turtle Cove, Kimberly, andoverlying Upper John Day members, in concertwith our revised chronostratigraphy (and that pro-vided by Hunt and Stepleton [2004]), result in areassessment of Arikareean subbiochron bound-aries, primarily because many of these data fill ingaps known to occur in the Great Plains record.Whereas the Whitneyan-Arikareean and Ar1-Ar2boundaries appear to be nearly coincident in boththe John Day and Great Plains regions, the lateearly Arikareean, Ar2, is here recalibrated to spanfrom 28 to 25.9 Ma; the early late Arikareean, Ar3,spans the interval between 25.9 and approximately22.7 Ma; and the late late Arikareean, Ar4, extendsfrom about 22.7 to 18.5 Ma.


Implications for Regional Dispersal

The ultimate goal of this project was to providegreater age precision for biotic events recorded inthe John Day sequence in order to test the syn-chroneity of such events that define mammal ageboundaries in geographically widespread and/ordisjunct regions, particularly those in the GreatPlains where the Arikareean NALMA was typified.In addition to the reassessment and recalibrationof Arikareean subbiozone boundaries, as notedabove, our biotic correlations also allow appraisalof faunal, or at least individual taxon, dispersal be-tween the John Day and Great Plains regions. Ourwork to date recognizes several intervals of previ-ously underappreciated faunal exchange betweenthese regions during the span of about 30.5–18.5Ma. Below we address these dispersals across thewestern United States by examining, primarily, therecord of taxonomic first appearances throughouteach of the John Day lithostratigraphic units.

From about 30.5 to 29.5 Ma (C11r–C11n; unitsA–D), Allomys, Nimravus, Hoplophoneus, Eu-smilus cerebralis, Paraenhydrocyon josephi, Me-socyon coryphaeus, Cynarctoides lemur, Da-phoenus, Perchoerus probus, Archaeotherium,Eporeodon occidentalis, Agriochoerus guyotianus,Mesohippus, Miohippus, and Diceratherium arefound in both the John Day and Great Plainsregions. This suggests a late Whitneyan to very ear-liest Ar1 interregional distribution and, essentially,homogeneity of the White River Chronofaunaacross what is now the northern and westernUnited States.

In units E and F, which we correlate with C10r(approximately 29.5–28.5 Ma), a new wave of im-migrants appears in the John Day Formation. Likethose in lower strata, however, many are knownfrom older rocks in the Great Plains, thus suggest-ing emplacement of a filtering mechanism that de-layed arrival of these taxa into the Pacific North-west until this time. They include Dinictis,Pogonodon platycopis, Dinaelurus, Osbornodonsesnoi, Mesocyon brachyops, A. guyotianus, andNanotragulus. The presence of Osbornodon andDinictis in unit E, both of which last occur at theend of the Whitneyan in the Great Plains, togetherwith P. platycopis, which last occurs in the GreatPlains in the early Arikareean (Bryant 1996), sug-gests dispersal between the two regions during the29–30-Ma interval. This is further supported by thepresence of Palaeocastor peninsulatus, which, ac-cording to Xu (1996), occurs in the Sharps Forma-tion, South Dakota, in association with theRockyford Ash (∼30 Ma). Another product of this

interregional dispersal may be M. brachyops (seeWang 1994, p. 43), as it appears to occur at nearlythe same time (Ar1) in Oregon, California, and theGreat Plains (Tedford et al. 1987). On the otherhand, Archaeolagus ennisianus appears to occurearlier in the John Day than in the Great Plains,again suggesting somewhat filtered dispersal, inboth directions, between the two regions.

Between the PGI and the DCT in units G–J (ap-proximately 28.7–27.8 Ma; mainly C9r), a few newtaxa appear in the John Day Formation coincidentwith their first appearance in the Great Plains, suchas Capacikala gradatus and Ekgmowechashala.Like Archaeolagus noted above, however, othertaxa from units G–J, including Pleurolicus and Ore-odontoides, appear earlier in the John Day region,by nearly 1 m.yr., than in the Great Plains.

In contrast, the unit K first appearances of Oca-jila, Leidymys, and Hypertragulus above the DCTare later in the John Day than in the Great Plains—slightly so for Ocajila but substantially so for Lei-dymys (C11 in the Great Plains vs. C9n in the JohnDay) and Hypertragulus (Chadronian-Ar1)—yet thefirst occurrence in unit K of Paraenhydrocyon wal-lovianus and Hypsiops predates their Ar3-Ar4 firstappearance in the Great Planes (Tedford et al. 1987,p. 185; Wang 1994). Thus, the unit K records ofthese taxa continue to infer two-way yet apparentlyfiltered dispersal between the two regions through-out late Ar1 into Ar2.

Above the TRT, in units L and M, are the firstrecords of several members characteristic of a fullydeveloped Runningwater Chronofauna, such as Da-phoenodon, Daeodon, Moropus, and Kalobatippus,among others noted previously. These occurrences,beginning at about 26 Ma, appear to indicate a sub-stantially earlier presence in the John Day regionthan in the Great Plains, where these taxa areknown from about 23 Ma. As discussed above, how-ever, this apparent diachroneity is much morelikely due to the significant hiatus recorded in theGreat Plains sequence between the Monroe Creekand Harrison formations. Dispersal between thetwo regions by 23 Ma is obvious, but faunal ho-mogeneity may have been emplaced much earlier,provided that barriers to dispersal were limited af-ter about 26 Ma.

The upper John Day Beds studied recently byHunt and Stepleton (2004) appear to document sim-ilar, essentially open dispersal between the tworegions, although the appearance of diachroneitystill exists because of the aforementioned hiatusesin the Great Plains sequence, in addition to signif-icant unconformities documented throughout andbetween the Haystack Valley (revised), Balm Creek,


Johnson Canyon, and Rose Creek members. For ex-ample, before the work of Hunt and Stepleton(2004), Miotapirus harrisonensis was known onlyfrom the Anderson Ranch Formation (previouslyUpper Harrison Beds) of Nebraska, which is datedon the basis of the Eagle Crag Ash at Ma19.2 � 0.5(Hunt 2002). But during their study, Hunt and Ste-pleton recovered a molar of this tapir from thelower part of the Johnson Canyon Member about1.5 m above the -Ma ATR (see Albright22.6 � 0.131998, pp. 203–204). This would seem to indicate alate arrival for M. harrisonensis in the Great Plainsrelative to its earlier presence in the John Day re-gion, but its first arrival in Nebraska more likelyoccurred during the interval of time missing in theunconformity between the Harrison and AndersonRanch formations.

In summary, an early interval of faunal exchangeappears to have transpired between about 30 and27 Ma and may have consisted of episodes at 30,29, and 27 Ma. Another episode, whereby elementsof the Runningwater Chronofauna typically foundin early late Arikareean faunas of the Great Plainsat about 23 Ma, appears to have transpired earlierin the John Day region but more likely is due tothe incomplete biostratigraphic record in the mid-continent. Another strong episode of faunal ex-change is recorded in the early Hemingfordian, atabout 18.5 Ma. Whether the apparent periods ofisolation (27–26, 24–23, and 20–19 Ma) actually re-flect that situation is open to further testing withnew biostratigraphic information.

Conclusions

This study has resulted in a new hypothesisregarding the calibration of the biochronologicalsubdivisions of the Arikareean NALMAs based onrecent lithostratigraphic, biostratigraphic, magne-tostratigraphic, and geochronologic data recoveredfrom the John Day Formation, Oregon. We inde-pendently appraised the chronological significanceof the taxa found in the Turtle Cove, Kimberly, and,to some extent, overlying units to ascertain the ex-tent to which they were compatible with, primar-ily, the Arikareean biochronological framework asdeveloped in the northern Great Plains. Whereasthe two frameworks were found to be more or lessconsistent, we recognize that hiatuses in the north-ern Great Plains succession have played an impor-tant role in constraining biochron ages in that clas-sic sequence. In fact, this study was designed, inpart, to address a problem noted by MacFadden andHunt (1998, p. 163) that “the late Oligocene andearly Miocene interval between 30 and 20 m.yr. is

poorly characterized for its [magnetic] polarity sig-nature.” Their study on the Arikaree Group ofnorthwestern Nebraska contributed to the resolu-tion of the problem, but it also served to reem-phasize the significant hiatuses within the group.The John Day Formation, therefore, was consideredthe ideal unit in which these gaps could be filledbecause of the thick, well-exposed, mammal-rich,structurally unambiguous, and essentially contin-uous sequence it provides. In turn, these data haveallowed more accurate assessment of Arikareeanbiozone limits.

For example, data from the John Day Formationconfirm that Ar1 spans an interval from about 30to 28 Ma. But the data also indicate that the bound-aries of the remaining Arikareean biozones can bemore accurately tuned because of the robust andmore temporally continuous biostratigraphy thatessentially fills in known temporal gaps in theGreat Plains sequence. The John Day record sug-gests that biochron Ar3 begins at about 25.9 Ma,in contrast with an age of about 23 Ma in the GreatPlains. Similarly, Ar4 in the John Day Formationbegins at about 23 Ma relative to an approximatebasal age of about 19.5 Ma in Nebraska. Althoughwe are aware that this proposal cannot be thor-oughly tested in the classic Arikareean exposuresof Nebraska and South Dakota for reasons statedabove, it is our hope that other sequences in themidcontinent and elsewhere can be used to test ourhypothesis and evaluate the applicability of ourresults.

A C K N O W L E D G M E N T S

Gratitude is extended to the National Park Serviceand the University of Florida for funding and lo-gistical support that made this project possible.Funding was also provided by a Research and Ex-ploration grant from the National Geographic So-ciety (grant 6185-98). S. Foss, K. Cahill, A. Pajak,S. Rickabaugh, and D. Sawatzky are thanked fortheir assistance over the course of the project. J.Channel, N. Opdyke, and K. Huang are thanked forallowing use of the paleomagnetics laboratory atthe University of Florida, as is P. Renne for the useof similar facilities at Berkeley GeochronologyCenter. We additionally thank N. Opdyke and S.Lucas for helpful reviews of an earlier version ofthis article, D. Prothero and R. Tedford for reviewsof the final version, R. M. Hunt for helpful discus-sions, and G. Hayes for allowing us to cite infor-mation from his as-yet unpublished work on themagnetostratigraphy of Monroe Creek Canyon,Nebraska.


R E F E R E N C E S C I T E D

Albright, L. B. 1998. New genus of tapir (Mammalia: Tap-iridae) from the Arikareean (earliest Miocene) of theTexas Coastal Plain. J. Vertebr. Paleontol. 18:200–217.

———. 1999. Ungulates of the Toledo Bend local fauna(Late Arikareean, Early Miocene), Texas Coastal Plain.Bull. Fla. Mus. Nat. Hist. 42:1–80.

———. 2005. Ekgmowechashala (Mammalia, ?Primates)from the Gulf Coastal Plain. In Hulbert, R. C.; Mor-gan, G. S.; and Baskin, J. A., eds. Cenozoic vertebratesof the Americas: papers to honor S. David Webb. Bull.Fla. Mus. Nat. Hist. 45:355–361.

Bailey, B. E. 2004. Biostratigraphy and biochronology ofearly Arikareean through late Hemingfordian smallmammal faunas from the Nebraska panhandle and ad-jacent areas. Paludicola 4:81–113.

Bestland, E. A. 1995. Stratigraphy of the Turtle CoveMember of the John Day Formation at Sheep Rock,John Day Fossil Beds National Monument, Oregon.Unpublished Rep. John Day Fossil Beds Natl. Monu-ment:1–14.

Bestland, E. A., and Retallack, G. J. 1994. Geology andpaleoenvironments of the Painted Hills Unit, JohnDay Fossil Beds National Monument, Oregon. Na-tional Parks Service Final Report, contract CX-9000-1-10009.

Bryant, H. N. 1996. Nimravidae. In Prothero, D. R., andEmry, R. J., eds. The terrestrial Eocene-Oligocene tran-sition in North America. New York, Cambridge Uni-versity Press, p. 453–475.

Butler, R. F. 1992. Paleomagnetism: magnetic domainsto geologic terranes. Boston, Blackwell Scientific, 319p.

Coombs, M. C.; Hunt, R. M.; Stepleton, E.; Albright, L.B.; and Fremd, T. 2001. Stratigraphy, chronology, bio-geography, and taxonomy of early Miocene smallchalicotheres in North America. J. Vertebr. Paleontol.21:607–620.

Emry, R. J.; Bjork, P. R.; and Russell, L. S. 1987. TheChadronian, Orellan, and Whitneyan North Americanland mammal ages. In Woodburne, M. O., ed. Ceno-zoic mammals of North America: geochronology andbiostratigraphy. Berkeley, University of CaliforniaPress, p. 118–152.

Evernden, J. F.; Savage, D. E.; Curtiss, G. H.; and James,G. T. 1964. Potassium-argon dates and the Cenozoicmammalian chronology of North America. Am. J. Sci.262:145–198.

Fisher, R. V. 1967. Early Tertiary deformation in north-central Oregon. Am. Assoc. Petrol. Geol. Bull. 51:111–123.

Fisher, R. V., and Rensberger, J. M. 1972. Physical stra-tigraphy of the John Day Formation, central Oregon.Univ. Calif. Publ. Geol. Sci. 101:1–33.

Fremd, T. J.; Bestland, E. A.; and Retallack, G. J. 1994.John Day Basin paleontology field trip guide and roadlog. In Society of Vertebrate Paleontology, 1994 An-

nual Meeting. Seattle, Northwest Interpretive Asso-ciation, 80 p.

Hay, R. L. 1963. Stratigraphy and zeolitic diagenesis ofthe John Day Formation of Oregon. Univ. Calif. Publ.Geol. Sci. 42:199–262.

Hayes, F. G. 2007. Magnetostratigraphy and paleontologyof Wagner Quarry (Late Oligocene, Early Arikareean)basal Arikareean Group of the Pine Ridge region,Dawes County, Nebraska. Bull. Fla. Mus. Nat. Hist.47:1–48.

Hunt, R. M. 1990. Taphonomy and sedimentology ofArikaree (lower Miocene) fluvial, eolian, and lacus-trine paleoenvironments, Nebraska and Wyoming: apaleobiota entombed in fine-grained volcaniclasticrocks. In Lockley, M. G., and Rice, A., eds. Volcanismand fossil biotas. Geol. Soc. Am. Spec. Pap. 244:69–112.

———. 1998a. Amphicyonidae. In Janis, C. M.; Scott, K.M.; and Jacobs, L. L., eds. Evolution of Tertiary mam-mals of North America. Vol. 1. Terrestrial carnivores,ungulates, and ungulatelike mammals. Cambridge,Cambridge University Press, p. 196–227.

———. 1998b. Ursidae. In Janis, C. M.; Scott, K. M.; andJacobs, L. L., eds. Evolution of Tertiary mammals ofNorth America. Vol. 1. Terrestrial carnivores, ungu-lates, and ungulatelike mammals. Cambridge, Cam-bridge University Press, p. 174–195.

———. 2002. New amphicyonid carnivorans (Mammalia,Daphoeninae) from the Early Miocene of southeasternWyoming. Am. Mus. Novit. 3385:1–41.

Hunt, R. M., and Stepleton, E. 2004. Geology and pale-ontology of the upper John Day beds, John Day RiverValley, Oregon: lithostratigraphic and biochronologicrevision in the Haystack Valley and Kimberly areas(Kimberly and Mt. Misery quadrangles). Bull. Am.Mus. Nat. Hist. 282:1–90.

Janis, C. M.; Scott, K. M.; and Jacobs, L. L., eds. 1998.Evolution of Tertiary mammals of North America.Vol. 1. Terrestrial carnivores, ungulates, and ungu-latelike mammals. Cambridge, Cambridge UniversityPress, 691 p.

Johnson, N. M.; Opdyke, N. D.; Woodward, G. D.; Lind-say, E. H.; and Tahirkheli, R. A. K. 1982. Magneticpolarity stratigraphy and ages of Siwalik Group rocksof the Potwar Olateau, Pakistan. Palaeogeogr. Palaeo-climatol. Palaeoecol. 37:17–42.

Kirschvink, J. L. 1980. The least-squares line and planeand the analysis of paleomagnetic data. Geophys. J. R.Astr. Soc. 62:699–718.

Lander, B. 1998. Oreodontoidea. In Janis, C. M.; Scott,K. M.; and Jacobs, L. L., eds. Evolution of Tertiarymammals of North America. Vol. 1. Terrestrial car-nivores, ungulates, and ungulatelike mammals. Cam-bridge, Cambridge University Press, p. 402–425.

Lindsay, E. H.; Johnson, N. M.; Opdyke, N. D.; and Butler,R. F. 1987. Mammalian chronology and the magnetic


polarity time scale. In Woodburne, M. O., ed. Ceno-zoic mammals of North America: geochronology andbiostratigraphy. Berkeley, University of CaliforniaPress, p. 267–284.

Lourens, L.; Hilgen, F.; Shackleton, N. J.; Laskar, J.; andWilson, D. 2004. The Neogene period. In Gradstein,F. M.; Ogg, J. G.; and Smith, A. G., eds. A geologictime scale 2004. Cambridge, Cambridge UniversityPress, p. 409–440.

Luterbacher, H. P.; Ali, J. R.; Brinkhuis, H.; Gradstein, F.M.; Hooker, J. J.; Monechi, S.; Ogg, J. G.; et al. 2004.The Paleogene period. In Gradstein, F. M.; Ogg, J. G.;and Smith, A. G., eds. A geologic time scale 2004.Cambridge, Cambridge University Press, p. 384–408.

MacFadden, B. J. 1998. Equidae. In Janis, C. M.; Scott, K.M.; and Jacobs, L. L., eds. Evolution of Tertiary mam-mals of North America. Vol. 1. Terrestrial carnivores,ungulates, and ungulatelike mammals. Cambridge,Cambridge University Press, p. 537–559.

MacFadden, B. J., and Hunt, R. M. 1998. Magnetic po-larity stratigraphy and correlation of the ArikareeGroup, Arikareean (late Oligocene–early Miocene) ofnorthwestern Nebraska. In Terry, D. O.; LaGarry, H.E.; and Hunt, R. M., eds. Depositional environments,lithostratigraphy, and biostratigraphy of the WhiteRiver and Arikaree Groups (Late Eocene to Early Mio-cene, North America). Geo. Soc. Am. Spec. Pap. 325:143–166.

Martin, L. D. 1998. Nimravidae. In Janis, C. M.; Scott,K. M.; and Jacobs, L. L., eds. Evolution of Tertiarymammals of North America. Vol. 1. terrestrial car-nivores, ungulates, and ungulatelike mammals. Cam-bridge, Cambridge University Press, p. 228–235.

Merriam, J. C. 1901. A contribution to the geology of theJohn Day Basin. Univ. Calif. Publ. Geol. Sci. 2:269–314.

Merriam, J. C., and Sinclair, W. J. 1907. Tertiary faunasof the John Day region. Univ. Calif. Publ. Geol. Sci.5:171–205.

Opdyke, N. D.; Lindsay, E. H.; Johnson, N. M.; andDowns, T. 1977. The paleomagnetism and magneticpolarity stratigraphy of the mammal-bearing sectionof Anza Borrego State Park, California. Quat. Res. 7:316–329.

Prothero, D. R., and Emry, R. J. 1996. Summary. In Proth-ero, D. R., and Emry, R. J., eds. The terrestrial Eocene-Oligocene transition in North America. New York,Cambridge University Press, p. 664–683.

———. 2004. The Chadronian, Orellan, and WhitneyanNorth American land mammal ages. In Woodburne,M. O., ed. Late Cretaceous and Cenozoic mammalsof North America: biostratigraphy and geochronology.New York, Columbia University Press, p. 156–168.

Prothero, D. R., and Rensberger, J. M. 1985. Preliminarymagnetostratigraphy of the John Day Formation,Oregon, and the North American Oligocene-Mioceneboundary. Newsl. Stratigr. 15:59–70.

Retallack, G. J.; Bestland, E. A.; and Fremd, T. J. 2000.

Eocene and Oligocene paleosols of central Oregon.Geol. Soc. Am. Spec. Pap. 344:1–192.

Robinson, P. T.; Brem, G. F.; and McKee, E. H. 1984. JohnDay Formation of Oregon: a distal record of early Cas-cade volcanism. Geology 12:229–232.

Stepleton, E., and Hunt, R. M. 1994. Geology and pale-ontology of the upper John Day beds, John Day RiverValley, Oregon: lithostratigraphic and biochronologicrevision in the Haystack Valley and Kimberly areas(Kimberly and Mt. Misery quadrangles). Society ofVertebrate Paleontology Field Conference, John DayBasin, Field Guide.

Tedford, R. H.; Albright, L. B.; Barnosky, A. D.; Ferrus-quia-Villafranca, I.; Hunt, R. M.; Storer, J. E.; Swisher,C. C.; Voorhies, M. R.; Webb, S. D.; and Whistler, D.P. 2004. Mammalian biochronology of the Arikareeanthrough Hemphillian interval (Late Oligocene throughEarly Pliocene epochs). In Woodburne, M. O., ed. LateCretaceous and Cenozoic mammals of North Amer-ica: biostratigraphy and geochronology. New York,Columbia University Press, p. 169–231.

Tedford, R. H.; Skinner, M. F.; Fields, R. W.; Rensberger,R. M.; Whistler, D. P.; Galusha, T.; Taylor, B. E.; Mac-donald, J. R.; and Webb, S. D. 1987. Faunal successionand biochronology of the Arikareean through Hem-phillian interval (Late Oligocene through earliest Pli-ocene epochs) in North America. In Woodburne, M.O., ed. Cenozoic mammals of North America: geo-chronology and biostratigraphy. Berkeley, Universityof California Press, p. 153–210.

Tedford, R. H.; Swinehart, J. B.; Swisher, C. C.; Prothero,D. R.; King, S. A.; and Tierney, T. E. 1996. TheWhitneyan-Arikareean transition in the High Plains.In Prothero, D. R., and Emry, R. J., eds. The terrestrialEocene-Oligocene transition in North America. NewYork, Cambridge University Press, p. 312–336.

Wahlert, J. H. 1983. Relationships of the Florentiamyidae(Rodentia, Geomyoidea) based on cranial and dentalmorphology. Am. Mus. Novit. 2769:1–23.

Wang, X. 1994. Phylogenetic systematics of the Hesper-ocyoninae (Carnivora: Canidae). Bull. Am. Mus. Nat.Hist. 221:1–207.

Wang, X., and Tedford, R. H. 1996. Canidae. In Prothero,D. R., and Emry, R. J., eds. The terrestrial Eocene-Oligocene transition in North America. New York,Cambridge University Press, p. 433–452.

Wang, X.; Tedford, R. H.; and Taylor, B. E. 1999. Phylo-genetic systematics of the Borophaginae (Carnivora:Canidae). Bull. Am. Mus. Nat. Hist. 243:1–391.

Wood, H. E.; Chaney, R. W.; Clark, J.; Colbert, E. H.;Jepsen, G. L.; Reeside, J. B.; and Stock, C. 1941. No-menclature and correlation of the North Americancontinental Tertiary. Bull. Geo. Soc. Am. 52:1–48.

Xu, X.-F. 1996. Castoridae. In Prothero, D. R., and Emry,R. J., eds. The terrestrial Eocene-Oligocene transitionin North America. New York, Cambridge UniversityPress, p. 417–432.

Date post:	22-Nov-2023
Category:	Documents
Upload:	ucriverside
View:	0 times
Download:	0 times

Revised Chronostratigraphy and Biostratigraphy of the John Day Formation (Turtle Cove and Kimberly...

Documents