Geology and Paleontology of the Upper John Day Beds, John Day River

Home , John Day River, Runningwater Formation

University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

Papers in the Earth and Atmospheric Sciences Earth and Atmospheric Sciences, Department of

2004 GEOLOGY AND PALEONTOLOGY OF THE UPPER JOHN DAY BEDS, JOHN DAY RIVER VALLEY, OREGON: LITHOSTRATIGRAPHIC AND BIOCHRONOLOGIC REVISION IN THE HAYSTACK VALLEY AND KIMBERLY AREAS (KIMBERLY AND MT. MISERY QUADRANGLES) Robert M. Hunt Jr. University of Nebraska-Lincoln, [email protected]

Ellen Stepleton University of Nebraska State Museum

Follow this and additional works at: https://digitalcommons.unl.edu/geosciencefacpub Part of the Earth Sciences Commons

Hunt, Robert M. Jr. and Stepleton, Ellen, "GEOLOGY AND PALEONTOLOGY OF THE UPPER JOHN DAY BEDS, JOHN DAY RIVER VALLEY, OREGON: LITHOSTRATIGRAPHIC AND BIOCHRONOLOGIC REVISION IN THE HAYSTACK VALLEY AND KIMBERLY AREAS (KIMBERLY AND MT. MISERY QUADRANGLES)" (2004). Papers in the Earth and Atmospheric Sciences. 544. https://digitalcommons.unl.edu/geosciencefacpub/544

This Article is brought to you for free and open access by the Earth and Atmospheric Sciences, Department of at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Papers in the Earth and Atmospheric Sciences by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. GEOLOGY AND PALEONTOLOGY OF THE UPPER JOHN DAY BEDS, JOHN DAY RIVER VALLEY, OREGON: LITHOSTRATIGRAPHIC AND BIOCHRONOLOGIC REVISION IN THE HAYSTACK VALLEY AND KIMBERLY AREAS (KIMBERLY AND MT. MISERY QUADRANGLES)

ROBERT M. HUNT, JR. Geological Sciences, University of Nebraska, Lincoln, NE 68588

ELLEN STEPLETON Division of Vertebrate Paleontology, University of Nebraska State Museum, Lincoln, NE 68588

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 282, 90 pp., 29 ®gures, 4 tables, 14 appendices Issued February 4, 2004

The Rose Creek Member at the Picture Gorge 36 locality (at center right, geologist stands on basal ¯uvial facies) yields the youngest mammalian fauna (ϳ18.2±18.8 Ma) in the John Day Formation's eastern subregion. Here the member is in angular unconformity above pinnacled gray Kimberly tuff and is overlain by Columbia ¯ood basalts. 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 3

CONTENTS Abstract ...... 3 Introduction ...... 4 Geology of the John Day Beds: Previous Work ...... 7 ``Upper'' John Day Beds: Earlier ConceptsÐShifting Perceptions ...... 10 Lithostratigraphy of the Upper John Day Formation ...... 12 Kimberly and Haystack Valley Members ...... 12 Balm Creek Syncline ...... 14 John Day Valley South of Kimberly ...... 28 Biochronology ...... 43 40Ar/39Ar Dating ...... 49 Regional and Local Structure ...... 51 Discussion and Conclusions ...... 59 Acknowledgments ...... 69 References ...... 69 Addendum ...... 72 Appendices 1-1 to 1-15 ...... 75

ABSTRACT

The John Day Formation of north-central Oregon preserves a succession of speciose, superposed Oligocene through early Miocene mammalian faunas that establish the sequence of mid-Cenozoic mammalian evolution within the Paci®c Northwest. Upper John Day rock units initially described by Merriam (1900, 1901) in the Kimberly and Haystack Valley areas were later divided into lower (Kimberly) and upper (Haystack Valley) members by Fisher and Rensberger (1972). We focused our study on the lithostratigraphic succession within the Hay- stack Valley Member. Rocks previously included in the Haystack Valley Member can be subdivided into four unconformity-bounded, genetic lithostratigraphic units that range in age from ϳ24 to ϳ18 Ma, three of the units incorporating age-diagnostic mammalian faunas. We have identi®ed two principal depositional units within the Haystack Valley Member of Fisher and Rensberger south of Kimberly: (1) Johnson Canyon MemberÐlate or latest Ari- kareean tuffaceous siltstones and ®ne sandstones (ϳ?19±22.6 Ma) with ¯uvial monomictic intraformational pebble gravels, well exposed along the west wall of the John Day valley; (2) Rose Creek MemberÐcoarse polymictic welded tuff-bearing gravels, debris ¯ows, coarse obsidian-shard tuffs, and ®ne-grained tuffaceous units, yielding early Hemingfordian mammals (ϳ18.2±18.8 Ma), deposited in angular unconformity on lower units of the John Day For- mation along the east wall of the John Day valley. At Balm Creek in the type area of the Haystack Valley Member, the southern limb of the Balm Creek syncline exhibits the most complete local section of upper John Day rocks, here comprising three members: (1) a revised Haystack Valley Member made up of early late Ari- kareean ribbed tuffs (ϳ23.5±23.8 Ma) with monomictic welded tuff conglomerate channels, overlain by a gray massive airfall marker tuff (GMAT); (2) Balm Creek MemberÐtuffaceous late Arikareean siltstones and ®ne sandstones interbedded with lacustrine tuffs, overlain by stacked ¯uvial ®ning-upward sequences and gray airfall tuffs; (3) Rose Creek MemberÐcoarse polymictic welded tuff-bearing gravels, debris ¯ows, lacustrine units, and ®ne-grained tuffaceous sediments, believed to correlate to the fossiliferous early Hemingfordian unit south of Kimberly. The complexity of upper John Day rocks (evidenced by marked lithofacies variation within multiple unconformity-bounded subunits, punctuated by numerous paleosols) suggests an early Miocene depositional regime with more varied local environments and pronounced episodic sedimentation relative to the more uniform Oligocene environments documented by lower John Day strata. Whereas the lower John Day Formation consists of ®ne-grained volcaniclastic sediments that were deposited in basins with minimal topographic relief, the upper John Day For- mation is characterized by a succession of increasingly coarse ¯uvial channel ®lls, as well as 4 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

massive airfall and coarse-shard tuffs, well-developed paleosols, and relict topography. Regional compression appears to have triggered ¯uvial incision and valley ®lling from ϳ24 Ma to at least ϳ19 Ma. Extension in latest John Day times resulted in the development of half-grabens or grabens both in Haystack Valley and south of Kimberly, ensuring the preservation of upper John Day sediments. Signi®cantly, a ®nal episode of normal faulting appears to have immediately preceded the earliest eruption of Picture Gorge Basalt, as evidenced by ¯ows of the Twickenham Member (PGBS) abutting against vertical fault scarps south of Kimberly. The faunas of the upper John Day units thus play a critical role in dating a complex sequence of tectonic events which preceded the onset of Columbia River Basalt Group ¯ooding in the early Miocene.

INTRODUCTION dents (Calkins, 1902) and in more recent petrographic analyses (Hay, 1963). In north-central Oregon a sequence of vol- The formation consists of andesitic to dac- canogenic rocks spanning the Eocene itic tuffaceous sediments, airfall tuffs, ash through late Miocene preserves a lengthy re- ¯ows, and lavas (Robinson et al., 1990; Rob- cord of Cenozoic biotas, environments, and inson and Brem, 1981; Hay, 1963). The areal tectonism. The late Eocene±early Miocene distribution of the formation was governed by John Day Formation is pivotal to that record: the location of its source vents to the west and its richly speciose succession of faunas es- by local topography. In early John Day times, tablishes the pattern of mid-Cenozoic mam- a topographic high was established along the malian evolution within the Paci®c North- axis of the Blue Mountain uplift (Robinson et west paleogeographic province. The John al., 1990); this high and the Ochoco Moun- Day beds, due to their occurrence along a tains now divide the formation among three major river system, proximity to early mili- tary routes, and their fossil content, were subregions: eastern, southern, and western among the ®rst in the far western United (®g. 1). Following deposition of the John Day States to be explored and described. During Formation in the eastern subregion, the area the 1860s and 1870s, Thomas Condon, Jo- was inundated by lava ¯ows of the Columbia seph Leidy, J.S. Newberry, O.C. Marsh, and River Basalt Group (CRBG). E.D. Cope focused attention on the John Day The eastern subregion today is traversed fossil ¯ora and fauna; the ®rst geological by the north, middle, and main forks of the publications formally describing the John John Day River. The river and its tributaries Day sequence appeared at the turn of the have deeply dissected the basalt plateau to century (Merriam, 1901; Calkins, 1902). reveal extensive exposures of the John Day In the earliest geologic assessments, the Formation along the course of the river be- John Day rocks were considered lacustrine neath steep cliffs of the Picture Gorge Basalt. deposits (Le Conte, 1874; Marsh, 1875). When traveling north (downstream) along This was in keeping with the predominant the John Day River from Picture Gorge to- theory of lacustrine origin of the Rocky wards Kimberly (®g. 2), one is viewing an Mountain basin ®lls in the western United exposed cross section of the John Day For- States, advocated in the latter half of the 19th mation: in stratigraphic sequence, the Big century by Clarence King, F.V. Hayden, N.H. Basin, Turtle Cove, and Kimberly Members, Darton, and others. Merriam's (1901) work and younger units, disconformably capped ®rst challenged this view in Oregon, identi- by basalts of the Columbia River Group. fying much of the John Day sequence as air- This is the classic section and ``type area'' fall tuffs and ¯uvially worked tuffaceous described by Merriam (1901), Fisher (1967), sediments deposited on broad plains of low and Fisher and Rensberger (1972: 7), among topographic relief. This view has been gen- others. The aggregate thickness of the John erally substantiated by more recent studies Day Formation in the eastern subregion ap- (Fisher and Rensberger, 1972; Fisher, 1967; proaches 750 m (Robinson et al., 1990). Hay, 1963). The important pyroclastic con- In the type area the John Day strata are tribution to the John Day Formation was ex- preserved in sculpted badland exposures that plicitly recognized by Merriam and his stu- have some lateral continuity; additionally, 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 5

Fig. 1. The principal exposures of the John Day Formation of north-central Oregon occur along the course of the John Day River (eastern facies), Crooked River (southern facies), and Deschutes River (western facies). The eastern and southern facies have produced abundant fossil mammals that form the basis for an Oligocene and early Miocene mammalian biochronology in the Paci®c Northwest. This report examines the upper John Day Formation of the eastern facies, outcropping along the John Day River, geographically situated between the Blue Mountain anticlinal uplift and the Ochoco-Aldrich Mountains axis. conspicuous marker units such as the Blue episodes of stream incision and ¯uvial and Basin Tuff, Picture Gorge Ignimbrite, and lacustrine deposition during the last stages of Deep Creek Tuff correlate outcrops that are John Day sedimentation prior to the out- not conjoined (those on the eastern and west- pouring of Columbia River basalts. The fau- ern ¯anks of the river valley, for example). nas from these rocks provide biochronologic The lower and middle units of the formation control for dating those events. are lithologically fairly uniform, comprised Recent work in North America, Africa, and mainly of ®ne-grained tuffaceous sediments South America has demonstrated the impor- and airfall tuff in conformable sequence; by tance of ®ne-grained pyroclastic depositional contrast, the upper units exhibit much greater environments relative to the Cenozoic mam- textural and lithologic variety, as well as un- malian record (Pascual et al., 1985; Pickford, conformable stratigraphic relationships. The 1986; Hunt, 1990), particularly during the Ol- uppermost rocks register tectonic events and igocene and Miocene. Fine-grained pyroclas- 6 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 2. Map of the John Day River valley with geographic localities discussed in this report. Our ®eld study of upper John Day rocks was focused in the Haystack Creek-Balm Creek area in Haystack Valley east of the town of Spray; in the vicinity of Kimberly post of®ce and for ϳ20 mi to the south of Kimberly along the John Day valley; and at Sutton Mountain east of Bridge Creek. The geographic area within the circle includes the classic John Day exposures studied by Merriam (1901), Hay (1963), and Fisher and Rensberger (1972). tics blanketing large geographic areas have graphic province. The rift basins of East Af- preserved mammalian faunas distributed rica, the Patagonian sequence east of the An- across a variety of subenvironments. When des, and the White River±Arikaree volcani- regional volcanism is temporally prolonged, clastics of the North American midcontinent these volcaniclastic prisms often retain a fau- include thick terrestrial pyroclastic sequences, nal succession documenting the pattern of commonly reworked by ¯uvial, eolian, and la- mammalian evolution within a major geo- custrine processes, that contain abundant 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 7 mammal remains. The John Day Formation and slope wash in some areas. Merriam also represents an additional example of an Oligo- recognized an ``upper division'' of light- Miocene faunal succession preserved in ®ne- toned or buff-colored tuffaceous sediments grained pyroclastic sediments, in this instance with a signi®cant ¯uviatile component, in- in association with a convergent continental cluding coarse sands and gravels. margin. R.L. Hay (1963) called attention to the Our purpose in this investigation was to role of diagenesis, and especially zeolitiza- examine the sediments of the upper part of tion, in modifying the ®ne-grained tuffaceous the John Day Formation, which have never sediments of the formation. Air-fall sediment been studied in detail, in order to interpret had accumulated incrementally, for the most the lithostratigraphy, depositional environ- part, and weathered to montmorillonite at the ments, and biochronology of their fossil land surface before burial (Hay, 1962, 1963; mammals, and to compare these ®ndings Fisher, 1964, 1966a, 1968). A subtropical to with our earlier results from the North Amer- warm temperate climate supplied suf®cient ican midcontinent (Hunt, 1985, 1990). rainfall for subsurface ¯uid migration through the ash-rich deposits. Subsequently, GEOLOGY OF THE JOHN DAY BEDS: ground water in®ltrating the sediments at PREVIOUS WORK moderate depth promoted the precipitation of the zeolite clinoptilolite as a pseudomorph in As described in earlier studies (Calkins, cavities from which glass shards had been 1902; Hay, 1963; Fisher and Rensberger, dissolved. In both Merriam's ``lower'' and 1972, Robinson et al., 1990), the John Day ``middle'' divisions, glass was almost entire- Formation is a widely distributed sequence ly replaced by zeolite and/or clay; an evident of pyroclastic sediments and ¯ows. In most distinction between these two divisions was areas, it is lithologically distinct from the an- the occurrence of kaolinite and ferric iron ox- desitic and basaltic rocks of the underlying ides in the lower unit, lending a distinct red- Eocene Clarno Formation. The John Day se- dish-maroon hue. The predominantly gray to quence owes its existence to andesitic-dacitic yellow-gray tuffaceous sediments of Merri- and rhyolitic eruptions from vents in the Cas- am's ``upper division'' lacked a zeolitic sig- cade range and its eastern outliers (Robinson nature and remained in a relatively fresh, un- et al., 1990). The formation is lithologically altered state (Hay, 1963). and texturally very uniform in its lower and At Sutton Mountain north of the town of middle parts; the higher units much less so. Mitchell (®g. 2), Hay determined that the The task of segregating the formation into boundary between the zeolitized and nonzeo- mappable subunits was ®rst attempted by litized units was diachronous. By employing J.C. Merriam (1900, 1901) who advocated a a geographically widespread rhyolitic ash threefold division of the John Day ``series'' ¯ow (Picture Gorge Ignimbrite) as a syn- within its eastern subregion (®g. 3). At the chronous horizon, he observed that the zeo- base of the series, overlying deeply weath- litization boundary variably occurred above, ered and eroded ¯ows, tuffs, lahars, and below, and within the ignimbrite. Preferring breccias of the Clarno Formation, are dark to differentiate rock units by a strict litho- red tuffaceous claystones which comprise chronologic scheme, Hay (1963) divided the Merriam's lowest unit. These are conform- formation into three members, with the low- ably overlain by drab to bluish-green tuffa- est including all rocks that predated the ig- ceous claystones and siltstones: a ``middle nimbrite, the second comprising the ignim- division''. The ``lower'' and ``middle'' divi- brite itself, and the third including all rocks sions are chie¯y ®ne-grained tuffaceous air- that postdated the ignimbrite (®g. 3). fall sediment that weathered on land surfac- During the 1960s, geologist R.V. Fisher es; numerous paleosol horizons are indicated and paleontologist J.M. Rensberger initiated by rhizoliths, burrows, and other invertebrate lithostratigraphic and biostratigraphic studies traces. These rocks show almost no evidence of the John Day Formation that included of ¯uvial incision and lack deposits of coarse Merriam's type area of the formation from detritus, but were reworked by sheet ¯oods Picture Gorge north to Kimberly along the 8 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 3. Stratigraphic nomenclature applied in previous studies to Oligocene and early Miocene rocks of the John Day Formation in the eastern subregion. The ignimbrite of Hay (1963) occurs within the Turtle Cove Member of Fisher and Rensberger (1972), who recognized its importance as a regional marker unit within the John Day Formation.

John Day River (Fisher and Rensberger, heavily on the identi®cation of a nonzeoliti- 1972: 7), continuing northwest to Haystack zed coarse-shard tuff on Rudio Creek as the Valley. The abundant fossil mammals, varied correlative of the zeolitized Deep Creek Tuff lithofacies (particularly in the upper units), south of Kimberly, supported by coeval ro- and numerous marker tuffs in stratigraphic dent assemblages found in proximity to both sequence presented suitable conditions for tuffs (Fisher and Rensberger, 1972: 14±15; extracting a detailed geologic and faunal his- Rensberger, 1983: 50±51, ®g. 20). Also, they tory from these rocks. noted that while zeolitization was usually re- Fisher and Rensberger (1972) believed stricted to the lower units, in one instance that the diagenetic boundary between zeoli- 200 ft (61 m) of upper John Day rocks in the tized and unzeolitized rocks in the type area Balm Creek valley were zeolitized (secs. 33± of the formation was a diachronous strati- 34, T8S, R25E). graphic horizon, as Hay had demonstrated in Thus, one had to be cognizant of the con- the Mitchell area. Their conclusion rested trol that topographic position of strata had ex- 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 9 erted over diagenesis, and to be alert to and the capping Columbia River Group ba- whether deformation (if any) had preceded or salts. The Balm Creek drainage southeast of postdated diagenesis. Zeolitization boundaries Haystack Valley is the implied type locality were apt to be diachronous where folding or (Fisher and Rensberger, 1972: 17). Here the faulting preceded chemical alteration. Con- contact of the Haystack Valley Member with versely, such boundaries could approximate underlying rocks was not clearly indicated; synchronous horizons in local areas where the memberÐas de®ned by Fisher and Rens- diagenesis preceded deformation of essential- bergerÐbegins with zeolitized well-indurat- ly horizontally strati®ed beds. Rensberger ed pale green to buff tuffs that include nu- (1971: 9) emphasized that, in the formation's merous local channel ®lls with lenses of ¯u- type area, fossils suggested that the boundary vial sandstones and welded tuff conglomer- between his middle and upper divisions ate, and continues upward with a sequence ``more closely approximates a time plane'' of light gray or buff, nonzeolitized, massive than in the Mitchell area studied by Hay air-fall tuffs and ¯uvial tuffaceous gray to (1963). greenish and pale yellow-gray sandstones Working with the classic John Day se- and siltstones with some lacustrine units. quence between Middle Mountain and Kim- The Haystack Valley Member was also berly, Fisher and Rensberger opted to retain recognized along the east wall of the John Merriam's lower and middle ``divisions''. Day valley south of Kimberly, where ``the Named the Big Basin and Turtle Cove Mem- distinctive welded-tuff-bearing conglomerate bers, respectively, these units are lithologi- clearly overlaps all earlier members'', and is cally similar but readily distinguishableÐin accompanied by ¯uvial tuffaceous sediments, this areaÐin terms of color and surface ex- air-fall tuffs, and paleosols similar to those pression (®g. 3). in Haystack Valley (Fisher and Rensberger, The zeolitization boundary de®ned the 1972: 18). Signi®cantly, Fisher and Rensber- base of Fisher and Rensberger's next younger noted that the conglomerates were found gest unit, the Kimberly Member. Hence, `àlmost exclusively in the lower third of the there is a transitional and ``stratigraphically member'', and concluded that the welded tuff variable'' contact between it and the Turtle conglomerate in the lower part of the Balm Cove Member below. The Kimberly Member Creek section and the conglomerate at the is a thick sequence (in aggregate, reported as base of the member south of Kimberly along 900 ft by Fisher but locally, south of Kim- the east wall of the John Day valley were berly, 200±300 ft) of unzeolitized, gray vol- correlatives, and approximately synchronous. caniclastic sediments conformably overlying We have learned in our investigation that the green or gray-green, zeolitized tuffs of such an assumption was mistaken; a clue the Turtle Cove Member. This unit is rec- hinting at this was Fisher and Rensberger's ognizable in outcrops distributed along the observation that the lower ϳ200 ft of the John Day River from approximately 4 mi member in the Balm Creek drainage was south of Middle Mountain northward to zeolitized, but that the member south of Kimberly, then northwest to the Richmond Kimberly was not. In fact, we shall show that Fault south of Balm Creek; it is also well- the zeolitized and nonzeolitized parts of the exposed along the North Fork of the John ``Haystack Valley Member'' prove to be of Day River east of Kimberly, and in the valley markedly different chronologic ages. of Rudio Creek. The Kimberly Member is, Fisher and Rensberger (1972: 21) also re- in some places, unconformably overlain by ported the existence in the Kimberly Quad- younger John Day sediments, and elsewhere rangle of rare outcrops (Յ20 ft thick) of a by basalts of the Columbia River Group. post-John Day rock unit beneath the Colum- The youngest unit formally named by bia basalts. The unit was more common in Fisher and Rensberger (1972) is the Hay- the adjacent Monument Quadrangle outside stack Valley Member. In most areas (except- their map area, where it comprised ¯uvial ing the basin margin, for example) the mem- sandstones, pebble gravel, and lacustrine ber included all sediments sandwiched be- beds. The sandstone is made up of basaltic tween the Kimberly Member (where present) grains and lapilli and overlies the gray tuffs 10 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 of the Kimberly Member. These sediments eastward to Kimberly and the town of Mon- were interpreted as epiclastic debris derived ument, as well as south of Kimberly (Turtle from the earliest basalt ¯ows entering the re- Cove of Merriam, 1901: pl. 6) where impor- gion, admixed with tuffaceous sediments tant outcrops occur along the valley margins eroded from subjacent upper John Day units. beneath the thick series of Columbia basalt Organic material was noted in lacustrine ¯ows. These exposures in most cases are vis- beds, and a ¯ora of Miocene age (J.A. Wolfe ible and accessible from highways paralleling in Fisher and Rensberger, 1972: 21±22) in- the river and have ®gured signi®cantly in a cluded oak, maple, birch, elm, Zelkova, and number of later studies (Fisher and Wilcox, Sophora (pagoda tree). The chronologic re- 1960; Wilcox and Fisher, 1966; Fisher and lationship of this ``post-John Day'' unit to Rensberger, 1972). Additional outcrops attri- the Haystack Valley Member was not inves- buted to the upper division by Merriam were tigated during our ®eld study. found to the north at Fossil and Lone Rock, to the east along the Middle Fork of the John THE `ÙPPER'' JOHN DAY BEDS: Day River, and along Bridge Creek at Sutton EARLIER CONCEPTSÐ Mountain (Merriam, 1901: 298). SHIFTING PERCEPTIONS The upper beds of the John Day Formation are dif®cult to study for reasons explicitly Recognition of an upper part or ``division'' mentioned by Merriam (1901): (1) the John of the John Day ``Series'' ®rst appears in Day Formation in the eastern subregion was Merriam's (1900, 1901) pioneering work. structurally disturbed, folded, and faulted pri- These upper beds included, in Merriam's or to deposition of the Columbia ¯ood basalts; words, ``typical products of erosion'' relative (2) following tectonic deformation, the John to the predominantly airfall origin of the tuff- Day beds were subjected to extensive erosion. aceous lower units, and his remarks document These erosional episodes, superimposed on his awareness of ¯uvially reworked pyroclas- the previously folded and faulted strata, pro- tics together with conglomerate lenses in the duced broad topographic lows that later ®lled upper beds. According to Merriam (1901: with the Columbia basaltic lavas. Consequent- 291), in Haystack Valley, and in the vicinity ly, (3) beds of the upper division may be sep- of Spray, and south of Kimberly (i.e., ``the arated by considerable distances from adja- lower end of Turtle Cove''), ``good exposures cent correlative strata by the intervening ba- show the highest portion of the series to be salts, complicating the lateral tracing and lith- composed of one or two hundred feet of hard, ostratigraphic correlation of outcrops. To blocky tuff, below which is about the same determine age equivalence of geographically thickness of sand and gravel. The gravels are disjunct outcrops of the upper John Day beds, in some places quite coarse, containing peb- biocorrelation using fossil mammals has been bles four or ®ve inches in diameter. The sands and remains an important tool, and it offers sometimes show cross-bedding. Included in considerable potential for the development of these deposits are worn fragments of bones a re®ned regional biochronology. and teeth ....''Infact he stated that the upper Whereas Hay's (1963) investigation of division contained the `ònly typical sands and low-temperature diagenesis of the John Day gravels in the John Day'' and produced many tuffs identi®ed zeolitization as a key element fossils (Merriam, 1900). His only photograph in the early postdepositional history of these of beds assigned to the upper division shows rocks, it left open the question of the litho- conglomerate lenses and sandstones in Hay- stratigraphic succession in the upper part of stack Valley (Merriam, 1901: pl. 8, ®g.1); the formation which, in the Sutton Mountain these conglomerate bodies are made up of the area, included 1300±1700 ft of tuffaceous welded tuff clasts employed by Fisher and sediments above the Picture Gorge marker Rensberger (1972) in de®ning their Haystack ignimbrite. A signi®cant amount of time is Valley Member. represented by John Day rocks above the ig- Exposures of the upper John Day ``divi- nimbrite, demonstrated by the great thickness sion'' were reported by Merriam (1901) along of ®ne-grained tuffaceous sediments, by fos- the John Day River from Haystack Valley sil mammals, and by an impressive paleosol 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 11 succession found throughout the upper part ley, and are characterized by numerous re- of the formation at Sutton Mountain. In ad- mains of [the camel] . . . Paratylopus.'' dition, Hay's (1963) study, focused in the The lithostratigraphy for the upper John Mitchell-Sutton Mountain area, did not dis- Day beds developed by R.V. Fisher (Fisher, cuss the lithostratigraphy of Merriam's `ùp- 1967; Fisher and Rensberger, 1972), in con- per John Day division'' in the type area south junction with Rensberger's biostratigraphy, of Kimberly, or in Haystack Valley. identi®ed two members in the type area of In contrast to Hay's concept, Merriam's Merriam, cropping out from Haystack Valley (1901) descriptions of upper John Day rocks eastward to Kimberly and Rudio Creek, and in the type area did not include all sediments south along the John Day River to Picture above the Picture Gorge ignimbrite in his Gorge. The Kimberly and Haystack Valley `ùpper division''. Merriam emphasized that Members of Fisher and Rensberger (1972) the upper part of the formation was charac- were mapped together as a single unit in the terized by sands, gravels, and by buff tuffa- Kimberly and Picture Gorge quadrangles ceous ®ne-grained beds several hundred feet where they focused their study, but were de- thick, but he did not recognize a precise low- ®ned and described separately in their text er boundary. Merriam and Sinclair (1907) (Fisher and Rensberger, 1972: 15±19). The later provided a more explicit statement: ``No two units did not exactly correspond to Mer- sharp stratigraphic line can be drawn be- riam's concept of the `ùpper division''. Mer- tween the Middle and Upper John Day, the riam and Sinclair (1907), who placed the up- color of the beds usually serving for their per boundary of the ``middle division'' at discrimination in the ®eld. Faunally, the di- ϳ100 ft above the Deep Creek Tuff, included viding line between them may be ®xed by in their `ùpper division'' the uppermost part the downward range of Promerycochoerus of the Turtle Cove Member of Fisher and which is not known to occur in the Middle Rensberger. John Day. This limit is for the present re- In our assessment of the rocks and faunas of the upper John Day, we focused our study garded as lying about 100 feet above a prom- on the classic exposures of Merriam in Hay- inent stratum of coarse gray tuff [Deep Creek stack Valley and in the vicinity of Kimberly Tuff of Fisher, 1962] exposed quite generally (®g. 2) that also served as the lithologic basis near the top of the Middle John Day in Turtle for the members of Fisher and Rensberger. Cove, where, by differential weathering, it The detailed geologic and faunal relation- usually forms a terrace.'' Fisher and Rens- ships within this geographic area have an im- berger (1972: ®g. 6) later reported that the portant bearing on the concept of `ùpper range of the oreodont Promerycochoerus ex- John Day'' in the region. We have examined tended downward to the Deep Creek Tuff. these rocks in somewhat greater detail than Merriam (1901) did not identify lithostrat- evidenced in previous investigations, and igraphic subdivisions within his `ùpper John provide measured sections of the key expo- Day series''. However, he recognized its sures that de®ne the members. composite nature and was aware that the up- In our study of the upper John Day beds permost rocks of the `ùpper division'' might we ®rst discuss the lithostratigraphic succes- contain a unique and distinctive mammal as- sion at speci®c localities in the ``type area'' semblage. The discovery of a tapir (Sinclair, where Merriam, Fisher, and Rensberger estab- 1901), the rodent Mylagaulodon (Sinclair, lished a local lithostratigraphy. We next re- 1903), and small camels in the `ùppermost view the earlier fossil collections from these beds of the Upper John Day'' in the Johnson units, and supplement them with newly dis- Canyon area prompted Merriam and Sinclair covered specimens from our ®eldwork in (1907: 184) to suggest that ``the sands, grav- 1992±2001. This enabled us to establish els and tuffs at the top of the Upper John biochronologic relationships between isolated Day may represent a third faunal subdivi- sections. Mammalian fossils from the upper sion.'' These rocks were ``typically exposed John Day units recognized in this report pro- on Johnson and Bologna Creeks, on Bridge vide the ®rst reliable biochronological frame- Creek, and in the upper end of Haystack Val- work for the upper part of the formation. 12 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

LITHOSTRATIGRAPHY OF THE UPPER However, if the member is traced northward JOHN DAY FORMATION along the west wall of the valley, a prominent calcareous paleosol appears at the top The work of Fisher and Rensberger (1972) of the unit (well exposed in SE ,NW , provides the most thorough assessment of the ¼ ¼ SW ,NE , sec. 12, T10S, R25E) that prob- upper beds and their mammalian faunas ¼ ¼ ably represents an uneroded remnant of the (Rensberger, 1971, 1973, 1983). Fisher's work terminal Kimberly Member surface. (1967) on regional structure, together with the At other locations south of Kimberly, 1972 study, formed a basis for our initial geo- higher stratigraphic units ®ll an irregular to- logic work in the Haystack Valley and Kim- pography developed on the surface of the berly areas. Fisher and Rensberger clari®ed Kimberly Member (Fisher, 1967). These are and re®ned Merriam's (1901) earlier obser- the sands, gravels, and tuffs of Merriam. vations, recognizing two lithostratigraphic Fisher mapped these higher beds together units in the upper part of the formation, the with the rocks of the Kimberly Member as a Kimberly and Haystack Valley Members. single map unit (Fisher and Rensberger, 1972: Maps 2, 3, Tjkh), but in his text he KIMBERLY AND HAYSTACK VALLEY MEMBERS recognized the higher beds as a distinct KIMBERLY MEMBER: The Kimberly Mem- stratigraphic unit, the Haystack Valley Mem- ber was de®ned by Fisher and Rensberger ber (Fisher and Rensberger, 1972: 17). This (1972: 15) as gray, yellow-gray, pale green- re¯ected Fisher's assumption that these sands ish-gray, and buff tuffs well exposed near the and gravels correlated with coarse ¯uvial Kimberly post of®ce. Reported to be ϳ900 sediments in the Balm Creek area of Hay- ft in composite maximum thickness, its lower stack Valley, as described below. boundary was determined by zeolitization Our examination of rocks referred by Fish- and hence was considered by them to be er to the Kimberly Member was centered on diachronous on a regional scale. However, in the principal exposures of the unit: the type the type area of the member, they noted that area near Kimberly, the northern Rudio the zeolitization boundary and the regional Creek drainage, and along the John Day river dip of the contact between the Kimberly and from Haystack Valley to Kimberly. Rocks of Turtle Cove members is nearly coincident so the Kimberly Member in the type area are that locally the boundary may be approxi- uniformly gray tuffaceous sediments, pre- mately synchronous. Our examination of the dominantly siltstones, sandstones, and less contact in the type area of the member south frequent intraformational conglomerate, with of Kimberly demonstrated that the lithology a distinctive obsidian-shard tuff near the base of the tuffaceous claystones above and below of the Rudio Creek section, considered by the boundary is identical with the exception Fisher and Rensberger (1972: 14±15) as a of zeolitic alteration. Nor is there any evi- possible equivalent of the Deep Creek Tuff. dence of unconformity or other pause in de- These tuffaceous sediments display primary position at this horizon. On the basis of these sedimentary structures and incised channel lithologic observations it would be reason- ®lls indicating local ¯uvial reworking togeth- able to predict faunal continuity across the er with interbedded airfall units. Burrows and boundary: Rensberger's (1971, 1973, 1983) rhizoliths are common at many levels, dem- rodent zonation for these sediments suggests onstrating a succession of paleosols sugges- this is in fact the case. tive of a gradual, episodic pattern of sedi- The upper boundary of the Kimberly ment accumulation through periodic ashfall Member is an erosional surface at many lo- events. The sediments within channels sug- calities. The upper part of the formation was gest both rapid deposition of volcaniclastic removed by downcutting accompanying tec- debris in intense ¯ood events as well as more tonic deformation during late John Day time. prolonged periods of channel ®lling and win- Along the west wall of the John Day valley nowing that yielded uniformly well-sorted south of Kimberly (secs. 14, 23, 26, T10S, sediments. In the Rudio Creek area, a ¯uvial R25E) the eroded surface of the Kimberly channel within the Kimberly Member Member is overlain by the Columbia basalts. (NW¼,NE¼,NE¼, sec. 27, T9S, R26E) is 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 13 incised into tuffaceous gray siltstones and is we regard as much younger in age. In appar- ®lled with poorly sorted, locally derived, ent agreement with Rensberger (1983: 29), we tuff-pebble conglomerate interbedded with recognize these gray tuffs in the Haystack ¯uvial tuffaceous sandstones and diatomite. Valley-Balm Creek area as part of the Hay- These sediments were rapidly deposited with stack Valley Member, as revised in this study. little winnowing. Fossils in these channels Fisher and Rensberger (1972: 17) did sug- include larger unabraded skeletal elements of gest that rocks age-equivalent to the Kim- mammals, indicating minimal transport and berly Member (along Rudio Creek) are rep- rapid burial. On the other hand, incised chan- resented by a zeolitized section of green beds nel ®lls situated within the member in its in Haystack Valley at UCMP locality Hay- type area south of Kimberly contain well- stack 1. This age-equivalence was said to be sorted, clast-supported, tuff-pebble conglom- demonstrated by the common occurrence of erates ®lling the base of channel scours. The similar species of entoptychine rodents good clast sorting and rounding and the pres- (Rensberger, 1983: 29).1 ence of only well-abraded fragments of HAYSTACK VALLEY MEMBER: This member mammal bone within these conglomerates was named by Fisher and Rensberger (1972: demonstrate more prolonged ¯uvial process- 17) for mostly unzeolitized ¯uvial and lacus- ing of the channel sediments. trine tuffaceous sediments, including airfall We did not ®nd the Kimberly Member to tuffs, interbedded with coarse sandstone and be particularly fossiliferous, although occa- conglomerate in the Haystack and Balm sional specimens occur in many outcrops. Creek valleys, and along the east wall of the Much of the bone from coarse sandstones John Day valley south of Kimberly. In actual and conglomerate bodies is fragmented and practice, Fisher's concept placed all John Day strongly abraded, indicating ¯uvial process- rocks that incorporated welded tuff-bearing ing. Mammal remains in the siltstones and conglomerates in the Haystack Valley Mem- sandstones tend to range from partially artic- ber. Such rocks are typically separated from ulated limb segments to isolated teeth, mandibles, and limb elements. Complete skele- 1 The exact geographic location of outcrops of gray Kimberly Member tuffs in the Haystack Valley was not tons are uncommon and we encountered speci®ed by Fisher and Rensberger (1972), resulting in none during our work. The condition of the some confusion as to their situation. Rensberger (1973: skeletal material indicates that carcasses de- 91) reported gray unzeolitized rocks of the Kimberly cayed, were scavenged, and in some instanc- Member overlying greenish zeolitized claystones of the Turtle Cove Member at UCMP locality Haystack 1, but es the remnants were buried by ashfalls or the thickness of the Kimberly Member was not given: by ¯uvial reworking of tuffaceous sediments, however, the thickness (ϳ140 ft) can be determined possibly during rainfall events following vol- from their section 11, which we believe includes the canic eruptions. exposures at Haystack 1. Later, Rensberger (1983: 29) Gray unzeolitized tuffaceous sediments of reconsidered the assignment of these gray tuffs to the Kimberly Member, regarding them as part of their Hay- the Kimberly Member cropping out in the stack Valley Member. We agree and include them in our type area south of Kimberly can be traced revised Haystack Valley Member. eastward into the Rudio Creek drainage and Fisher and Rensberger (1972: ®g. 3), however, show northwestward along the John Day valley ϳ200 ft of Kimberly Member in their section 10 at Balm Creek (NE¼, sec. 33, T8S, R25E). Their placement of from Kimberly nearly to Haystack Valley. this thick 200-ft interval in the Kimberly Member in sec- Typical gray Kimberly Member tuffs when tion 10 seemingly con¯icts with an earlier remark that the traced northwest from Kimberly terminate at member was quite thin in the Haystack Valley area in the projected trace of the Richmond Fault secs. 33±34, T8S, R25E where section 10 is located (Fish- along the east margin of the Balm Creek val- er and Rensberger, 1972: 17). The 200 ft of Kimberly beds shown in section 10 apparently refer to green zeo- ley (Fisher and Rensberger, 1972: Map 3, E½, litized beds along Balm Creek (E½, sec. 33, T8S, R25E) sec. 34, T8S, R25E). Kimberly Member gray that correspond to the lowest ϳ160 ft of our Asher Sec- tuffs occur in the upper part of outcrops tion (®g. 4), and which elsewhere in this same paper are southeast of the fault on the upthrown block considered to be the lower part of the Haystack Valley Member by Fisher and Rensberger (1972: 17). Thus, the in sec. 34; sections on the downthrown block 200 ft of Kimberly Member shown in section 10 appears north of the fault in Balm Creek valley con- to be a lapsus calami, and in fact belong to their Haystack tain unzeolitized gray cliff-forming tuffs that Valley Member, an assignment with which we concur. 14 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 the Kimberly Member by erosional unconfor- glomerates in more detail in the two principal mity south of Kimberly along the east wall of geographic areas mapped by Fisher and Rens- the John Day valley and include ¯uvially re- berger (1972): (1) Haystack Valley, and (2) worked tuffaceous sediments of all textural south of Kimberly along the John Day valley. grades from clay to gravel. We discovered that, based on the contained Fisher and Rensberger (1972: 17) frequently mammal faunas and lithostratigraphic rela- refer to ``welded-tuff-bearing conglomerate at tions, several stratigraphic units of different the base of the Haystack Valley Member ...''. ages were included within the Haystack Val- Their generalized regional stratigraphic section ley Member, and could be discriminated on depicts the conglomerate as the basal unit superpositional relations, bounding unconfor- within the member (Fisher and Rensberger, mities, lithofacies, and faunal content. In ad- 1972: ®g.1), separated by erosional unconfor- dition, the conglomerates in these units dif- mity from the Kimberly Member below. In- fered in both clast composition and size in a deed, two of their measured sections (Fisher predictable manner (table 1). Three regionally and Rensberger, 1972: ®g. 3, nos. 5 and 8) signi®cant unconformities of early Miocene south of Kimberly show a welded tuff con- age (EMU) are recognized in the study, and glomerate as the basal unit of the Haystack are indicated in our measured sections, from

Valley Member. However, their principal mea- older to younger, as EMU1, EMU2, and EMU3. sured section for Haystack Valley (Fisher and We describe these units ®rst in the Hay- Rensberger, 1972: ®g. 3, no. 10), sited at Balm stack Valley-Balm Creek area, then from ex- Creek, shows the welded tuff conglomerate as posures along the east and west walls of the a single horizon ϳ100 ft above the base of the John Day valley south of Kimberly. unit. This placement also appears in the composite lithostratigraphic column for the Kim- BALM CREEK SYNCLINE berly and Picture Gorge quadrangles (Fisher and Rensberger, 1972: ®g. 3). We became in- The thickest section of Fisher and Rens- terested in the stratigraphic occurrence of berger's Haystack Valley Member occurs welded tuff conglomerates when we encoun- along Balm Creek (®g. 2), a small drainage tered them at multiple levels at Balm Creek, that enters the John Day River about 1.5 mi including one prominent bed quite high in the (2.4 km) southeast of Haystack Creek. This local section. This contrasted with the member is the locale for their section 10, ϳ730 ft south of Kimberly along the east wall of the thick, which was measured in the NE¼, sec. John Day valley where, as indicated by Fisher 33 and SE¼, sec. 28, T8S, R25E, Kimberly 2 and Rensberger, the welded tuff conglomerate 7.5-min. quadrangle. The Balm Creek sec- nearly always occurs at a single stratigraphic tion is represented by badland and cliff ex- horizon at the base of the unit. posures along the southern limb of a syncline Comparing the mammalian faunas from the that forms a topographic high between Hay- two areas (Balm Creek and the east wall of stack Creek and Balm Creek. We have des- the John Day valley south of Kimberly) clar- ignated this the Balm Creek syncline, noting i®ed the situation. The Haystack Valley Mem- that it was recognized and mapped by Fisher ber rocks at Balm Creek carried a different as a northeast-to-east-trending fold (Fisher and much older fauna (appendices 1-1 to 1- and Rensberger, 1972: Map 3, secs. 26±28, 7) than that found in the conglomerate unit T8S, R25E). south of Kimberly (appendices 1-12 to 1-15). Although both stratigraphic units had been in- 2 Stratigraphic sections in this report show thickness of cluded in the Haystack Valley Member, they rock units in both English and metric systems. Elevations are expressed in feet so that future investigators can more proved to be of signi®cantly different ages. easily and accurately locate particular stratigraphic levels We thus became aware that welded tuff con- on 1:24,000 7.5-minute topographic quadrangle maps of glomerates occurred over a considerable time the U. S. Geological Survey. Earlier geologic and faunal interval in the upper John Day beds. studies of the John Day Formation are severely limited in their usefulness by the lack of detailed measured sections These observations led us to study the lith- tied to geographic locations and topographic elevations, ostratigraphy, the distribution of fossil mam- hence the inability of later workers to relocate faunal sites mals, and the clast composition of the con- and marker beds in local stratigraphic sections. 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 15

TABLE 1 Intermediate Diameter (I.D., in centimeters), 10 Largest Clasts, from Conglomerates of the Upper John Day Formation, Wheeler and Grant Counties, Oregon 16 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

We measured ®ve sections to demonstrate and siltstones with interbedded channel sand- the lithostratigraphic succession of the south stone lenses comprise the lowest ϳ160 ft of limb of the syncline (®gs. 4, 5, also Map A: our measured section. This interval lacks measured sections 1±5); these are correlated welded tuff conglomerate, with the exception by a marker bed that can be traced through- of a few weathered welded tuff pebbles out the southern limb, a prominently exposed found in a single lens of cross-strati®ed ¯u- gray massive airfall tuff (GMAT). vial sandstone. The lowest 40±50 feet of the Particularly important to our interpretation 160-ft section contain the only occurrences of the Haystack Valley Member is a mea- of entoptychine rodents in the Asher Section, sured section situated 0.4 mi (0.64 km) here represented by the most advanced spe- northeast of Fisher and Rensberger's section cies, Entoptychus individens (Rensberger, 10 in the W½,SW¼, sec. 27, T8S, R25E. 1971). The stratigraphically highest entop- Named the Beardog Section (®gs. 5±7, also tychine that we encountered was found 5 feet Map A: measured sections 5A, 5B)3, it pre- above a prominent green siltstone (``celadon- serves higher units missing from section 10. ite bench'', Asher Section, ®g. 4). Fisher and The upper 150 ft of the Beardog Section (lost Rensberger (1972: 17±18, 25, ®g. 6) regard- to erosion at section 10) includes a promi- ed the greenish zeolitized rocks of Unit 1 as nent, coarse welded tuff conglomerate, piv- the lower part of their Haystack Valley Mem- otal in assessing the distribution and signi®- ber, and also noted that the beds included the cance of conglomerates within the upper terminal John Day entoptychine, E. indivi- John Day. We ultimately determined that this dens. Associated with E. individens at stratigraphically high conglomerate of the UCMP locality Haystack 19 is the rare aplo- Beardog Section correlated with coarse weld- dontid rodent Allomys tessellatus; thus, the ed tuff conglomerate along the east wall of lowest part of the Haystack Valley Member the John Day valley south of Kimberly, and along Balm Creek includes an E. individens- that these polymictic coarse conglomerates A. tessellatus biozone (Rensberger, 1983: 90, and associated lithofacies de®ned a terminal table 16).4 rock unit (Rose Creek Member of this report, Figure 4 shows that no basal contact of see below) in angular unconformity with all Unit 1 with a subjacent stratigraphic unit is subjacent John Day members. By contrast, evident in the Balm Creek drainage. How- the welded tuff conglomerates in the lower ever, older John Day rocks are prominently part of the Haystack Valley Member along exposed to the west within Haystack Valley Balm Creek are local in occurrence and have itself along Haystack Creek, but their strati- no apparent correlative elsewhere in the graphic relationship to Unit 1 is dif®cult to study area. Thus, the Beardog Section and discern because of faulting and vegetative Fisher and Rensberger's section 10 (the Ash- cover along both the east and west walls of er Section of this study) are central to a re- Haystack Valley (see addendum). appraisal of the Haystack Valley Member In addition to geomyoid rodents, a repre- and are described here: sentative mammal fauna occurs in Unit 1 UNIT 1 (®g. 4, Asher Section, lower part, 0toϳ160 ft; appendix 1±4): Greenish-gray 4 Rensberger (1971: 155) at ®rst did not apply the to olive-gray tuffaceous zeolitized claystones term Haystack Valley Member to the greenish zeolitized rocks containing E. individens since the member had not 3 The name given the measured section refers to a been named at that time; he described the stratigraphic temnocyonine amphicyonid cranium excavated at this occurrence of E. individens as follows: ``The highest ob- locality. The skull and associated mandibles (Northwest served stratigraphic position for Entoptychus is slightly Museum NM 280/61, Portland State University, Port- above the lowest beds of coarse-grained, water-laid clas- land, OR) are the most complete and best preserved am- tics in Haystack Valley ...''. Once the member had phicyonid cranial remains found in the upper John Day been named and de®ned, Rensberger (1973: 86, ®g. 2) Formation. The fossil carnivore was discovered by M. indicated that the range of entoptychines extended into Fingerhut and G. Pierson on April 2, 1986, during ex- the base of the Haystack Valley Member (also Fisher ploration of the John Day beds along Balm Creek, and and Rensberger, 1972: ®gs. 6, 8). Thereafter, Rensberger was donated to Portland State University, where it is (1983: 29) explicitly noted that E. individens occurs in conserved by Dr. David Taylor. Taylor and Fingerhut ``greenish, zeolitized deposits at the base of the [Hay- generously permitted our study of the skull. stack Valley] member (interval G).'' 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 17 18 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 4. Stratigraphic pro®le of the upper John Day Formation from the mouth of Haystack Creek 1 mi (1.65 km) eastward to the Balm Creek drainage. Four measured sections are correlated using a gray massive airfall tuff (GMAT) and local monomictic welded-tuff conglomerate bodies (see p. 17 for lithologic key). Rocks below the GMAT marker bed are particularly fossiliferous and are the source of the mammal fauna from the Haystack Valley Member (revised). Entoptychine rodents occur no higher in the revised member than the 40±50-ft level of the Asher Section (lower part). Unit thicknesses in lowermost 160 ft are estimated within Ϯ5%. 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 19

Fig. 4. Continued.

(appendices 1-4, 1-7). These taxa are com- UNIT 2A (®g. 4, Asher Section, upper bined here with those from Unit 2A to form part, ϳ160 to 382 ft): From ϳ160 to 382 ft a composite assemblage (table 4; appendices in the Asher Section, yellow-gray to olive- 1-1 to 1-7, from the Haystack Valley Mem- gray tuffaceous sediments continue to resem- ber, revised). ble the rocks in the lower 160 ft of the sec- 20 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 21 tion but appear not as strongly zeolitized. Haystack Valley by a gray massive airfall The ®eld term ``ribbed tuff'' was applied to tuff (GMAT), a marker bed that ranges from these sediments because weathered outcrops ϳ5 to 35 ft in thickness. It forms a vertical are typi®ed by alternating, close-spaced, cliff in the Balm Creek area and can be more and less resistant tuff levels, giving the traced from the S½, sec. 27, to SE¼, sec. 28, outcrop a horizontally ribbed appearance. and into N½, sec. 33, T8S, R25E. The rock Multiple lenses of welded tuff conglomerate matrix includes predominantly glass shards characterize the middle part of the interval, with scattered volcanogenic euhedral crystals occurring at three or more stratigraphic levels and abundant pumice fragments. The upper over an interval of at least 130 vertical ft in 10±15 ft of this unit are overprinted by a sections at the mouth of Haystack Creek and well-developed paleosol with siliceous rhi- along Balm Creek. These are monomictic zoliths, small burrows, petrosilicic laminae, conglomerates comprising only welded tuff and soil breccia. Directly beneath this airfall pebbles with mean intermediate diameter tuff at one locality (®g. 8, HS-1 section; Map (I.D.) less than 5 cm (table 1, Haystack Val- A: measured section 6, SE¼,SE¼, sec. 27, ley Member, revised). These are the con- T8S, R25E) are 10±30 ft of dark gray, cross- glomerates recognized in Haystack Valley strati®ed ¯uvial sandstones, incorporating within the Haystack Valley Member by Fish- polymictic (made up of welded tuff, other er and Rensberger (1972: 17±18, ®g. 1). rhyolitic/dacitic volcanic, and intraforma- Their measured section 10 shows conglom- tional tuff clasts) pebble conglomerate, that erate present in the lower part of the member, incise the uppermost beds of Unit 2A. These appearing ϳ100 ft above the base (Fisher and ¯uvial sediments appear to grade upward Rensberger, 1972: ®g. 3). However, the des- into the GMAT, hence these sediments and ignation of a particular conglomerate level in the GMAT are considered a single deposi- their plate 7 as ``the [authors' italic] marker tional unit postdating Unit 2A by an un- bed at the base of Haystack Valley Member'' known amount of time (no fossil mammals contradicts section 10. In fact, section 10 cor- have been found as yet in Unit 2B to gauge rectly represents the stratigraphy. Although this). However, the GMAT unit is most often multiple levels of these welded tuff conglom- underlain not by coarse ¯uvial sediments but erates are present in the 160 to 382-ft interval by silty tuffaceous buff claystones that ap- in the Balm Creek measured sections (®g. 4: pear to be a lateral facies of the dark gray sections 1, 2, 4), no marker conglomerate channel sands and pebble conglomerate seen bed identi®es the base of the unit. in HS-1. It may be advisable at a later date Fossil mammals are present as scattered to segregate Unit 2B as a distinct lithostrat- occurrences throughout Unit 2A, and are igraphic unit should an erosional unconfor- combined with those of Unit 1 to form a mity identi®ed by incision of ¯uvial Unit 2B composite assemblage (table 4; appendices channels into Unit 2A (®g. 8) prove to be 1-1 to 1-7). recognizable throughout the Haystack Val- UNIT 2B (®g. 4, Asher Section, upper ley-Balm Creek area. part, ϳ382 to 435 ft): Unit 2A is succeeded Although description of the Balm Creek in the Balm Creek drainage and throughout stratigraphy could be continued here using

← Fig. 5. Beardog Section: situated 0.3 mi (0.5 km) northeast of the Asher Section in the Balm Creek drainage in SW¼, sec. 27, T8S, R25E. This measured section transects the south limb and core of the Balm Creek syncline. Unit thicknesses (in feet and meters) include reference topographic elevations because the attitude of strata varies markedly upward in the section as the core of the syncline is approached; the lowermost 350 ft of section dip steeply northwest, with progressively decreasing apparent dip in the upper cliff outcrops. Unit thicknesses in the lowermost 350 ft are estimated within Ϯ5%. Roman numerals (I±X) indicate ¯uvial ®ning-upward sequences within the Balm Creek Member. The GMAT unit in the Beardog Section (Lower) also occurs in ®gure 4, correlating the Asher and Beardog Sections. 22 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 23 the less complete Asher Section, we em- Two major paleosols are present, a lower one ployed the GMAT as a marker bed that al- developed on a gray airfall tuff (425±440 ft), lowed us to trace the local stratigraphic sec- and a terminal paleosol 12 ft thick at the top tion 0.4 mi (0.64 km) to the northeast in or- of the sequence (508±520 ft). der to describe the more complete upper part No fossil mammals have been found in of the Balm Creek sequence at the Beardog this unit. Section. UNIT 5 (®g. 5, Beardog Section: Upper, UNIT 3 (®g. 5, Beardog Section: Lower ϳ517±544 ft, measured section 5A; however, and Middle, ϳ175 to 350 ft): Above the Unit 5 is incised to the 469 ft level 0.2 mi GMAT marker bed we measured ϳ175 ft of (0.3 km) to the northwest in ®g. 6, measured ¯uvial tuffaceous siltstones and ®ne sand- section 5B): Here the terminal Unit 4 paleo- stones overlain by and interbedded with ®ne- sol is deeply incised by a Unit 5 channel to ly laminated lacustrine claystones and thin a depth of ϳ50 feet (®g. 6, Beardog Section: airfall tuffs. Unit 3 can be divided into a low- Upper, 0.2 mi northwest, channel incision er 65±85 ft-thick interval of yellow tuffa- from 518 ft to ϳ469 ft level). The base of ceous claystones, with occasional lenses of this channel is ®lled with clast-supported po- ¯uvially reworked siltstone and ®ne sand- lymictic welded tuff-bearing conglomerate stone, that weathers to a smooth slope that includes cobbles as large as 14 cm I.D. marked by infrequent thin resistant beds. This is a much coarser conglomerate than the Above this are ϳ90 ft of white to light gray conglomerate lenses of Unit 2A; mean I.D. lacustrine tuff and a few thin interbedded for the various conglomerate bodies in Unit dark gray airfall tuffs. These ®nely laminated 5 range from 7.5 to 10.0 cm (table 1, Balm lacustrine sediments and airfall tuffs were Creek syncline). Unit 5 conglomerates are deposited in a shallow con®ned body of wa- found only at the highest elevations beneath ter. From 272 to 312 ft in the Beardog Sec- the capping basalts of the Balm Creek drain- tion, the yellow tuffaceous claystones are in- age and are polymictic (varied clasts types, terbedded with the lacustrine tuffs, demon- most commonly, welded tuff derived from strating that the lower and upper parts of the Picture Gorge ignimbrite, and indurated Unit 3 represent a single depositional se- intraformational vitric tuff) in contrast to the quence. monomictic (a single clast type of welded Only a single fossil mammal, a partial tuff) conglomerates of Unit 2A. The coarsest skull of a peccary (cf. Hesperhys) in a pri- conglomerate forms the basal lithofacies of vate collection, has been found in Unit 3. It Unit 5 in the Beardog Section but polymictic occurred a few feet above the base of Unit 3 welded tuff-bearing conglomerate lenses in the Beardog Section (appendix 1-6). The with slightly smaller average clast diameter lacustrine beds in the upper part of the unit occur at higher levels within the unit (®gs. have not produced fossil mammals. 5, 6, Beardog Section: Upper, at ϳ520 ft). UNIT 4 (®g. 5, Beardog Section: Middle Unit 5 appears to be angularly unconform- and Upper, ϳ350 to 520 ft): Above Unit 3 able on lower units in the Balm Creek valley, are ϳ170 ft of stacked ®ning-upward se- and hence it postdates the compressive fold- quences composed of ¯uvial tuffaceous sand- ing and associated faulting that created the stone, siltstone, and rare lenses of intrafor- Balm Creek syncline. The tuffaceous sedi- mational pebble conglomerate (I.D. Ͻ2cm) ments and polymictic conglomerates of Unit lacking welded tuff pebbles. Coarse obsidian 5 are not mentioned by Fisher and Rensber- shards are evident in some ¯uvial sandstones. ger (1972) because the unit was removed by

← Map A: Topographic map of the Haystack Creek-Balm Creek area, John Day valley, Oregon, Kim- berly 7.5-minute quadrangle, indicating the Balm Creek syncline and Richmond fault in relation to measured sections of this report: 1, Crazy Woman; 2, Paratylopus Hill; 3, Hopper; 4, Asher; 5, Beardog; 6, HS-1. Triangles indicate approximate locations of UCMP Haystack 21 and 22 where Entoptychus individens occurs in the lower part of the Haystack Valley Member (revised). 24 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 erosion from their section 10, prior to the ther placed in doubt when, in section 10, an outpouring of the Columbia basalts. The additional ϳ100 feet of the member is shown principal outcrop of Unit 5 in the Beardog below the welded tuff conglomerate level. Section is exposed (®g. 7) over a very small Fisher and Rensberger considered all rocks area within the axis of the Balm Creek syn- of their section 10 from the welded tuff con- cline (NW¼,NE¼,SE¼,NW¼,SW¼, sec. glomerates up to the capping basalts as the 27, T8S, R25E) and could be easily missed Haystack Valley Member. We suggest a more in a regional reconnaissance study. circumscribed stratigraphic de®nition of the Despite diligent searching, no fossil mam- Haystack Valley Member (revised) incorpo- mals have been found in Unit 5. rating only our Units 1 and 2, thus limiting REVISION OF THE HAYSTACK VALLEY MEM- the member to the lower 435 feet of the Ash- BER5: Fisher and Rensberger did not describe er Section along Balm Creek (®g. 4). The the rocks of the Haystack Valley Member in upper boundary of the unit is demarcated by the Balm Creek valley in detail; however, the prominent siliceous paleosol developed their principal stratigraphic section (Fisher on the gray marker tuff (GMAT, ®g. 4) of and Rensberger, 1972: ®g. 3, section 10) was Unit 2B. Rocks below this marker bed in measured in the Balm Creek drainage at the Units 1 and 2 are of uniform lithology and same location as our Asher Section. Because surface-weathering characteristics; the alter- section 10 had lost Unit 5 and the upper part nating resistant ledges and softer interbeds of Unit 4 to erosion, Fisher and Rensberger's give a horizontally ribbed appearance to concept of the Haystack Valley Member did these outcrops along Balm Creek and were not include these uppermost John Day rocks. termed ``ribbed tuffs'' during our ®eld study. Fisher and Rensberger (1972: pls. 7, 8) se- Units 1 and 2A are zeolitized, more strongly lected the welded tuff-bearing conglomerate toward the base. Importantly, monomictic at the 300 ft level in our Asher section as the welded tuff conglomerates are present at base of the Haystack Valley Member. The multiple levels in Unit 2A, although more totality of exposures in the Balm Creek concentrated from 100 to 170 feet below the drainage (®gs. 4±8) demonstrates the dif®- gray marker tuff (GMAT) in the middle part culty in using the appearance of a particular of the section. In Unit 2B, only a single lens welded tuff-bearing conglomerate to identify of polymictic conglomerate with welded tuff the base of the Haystack Valley Member pebbles was observed, occurring at the base since there are multiple levels with conglom- of the unit in a channel cut into Unit 2A (®g. erate of this type in the Haystack Valley- 8, ϳ2632 ft). Above Unit 2, only rare, minor Balm Creek area, and the conglomerate body welded tuff conglomerate lenses occur in the of their plate 7 is not the stratigraphically section below the coarse basal polymictic lowest such lens. In fact, lithologies above conglomerate of Unit 5 (at Section HS-1, and below this bed are similar and do not thin granule to pebble conglomerate lenses at warrant stratigraphic separation. The location two levels within the lower part of Unit 3 of the lower boundary of the member is fur- contained welded tuff clasts). Fossil mammals in the Balm Creek drainage are entirely 5 The North American Stratigraphic Code (1983: 855) restricted to the revised member and do not de®nes revision of a geologic unit as ``either minor changes in the de®nition of one or both boundaries of a occur above it, except for one specimen, a unit, or in the unit's rank.'' Our revised Haystack Valley peccary skull, found at the base of Unit 3 Member alters only the upper boundary of the unit, re- (appendix 1-6). taining in the unit the lower 435 ft of section that in- The revised member is principally exposed cludes the primary lithologic content attributed to the along Balm Creek in sections 33±34, T8S, member by Fisher and Rensberger (1972). We remove the uppermost ϳ100 ft of their measured section 10 and R25E, but does not occur south of the Rich- assign it to the Balm Creek Member (new). This action mond Fault in section 34 (Map A). It also on our part conforms to the recommendation of the Code crops out in cliffs along the west and east (1983: article 19a) which states that ``Revision is justi- walls of Haystack Valley where the member ®able if a minor change in boundary or content will make a unit more natural and useful. If revision modi®es is in fault contact with the Turtle Cove Mem- only a minor part of the content of a previously estab- ber which occupies the central part of the lished unit, the original name may be retained.'' valley. 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 25

Fig. 6. Uppermost part of the Beardog Section: a basal ¯uvial channel of the Rose Creek Member is incised ϳ50 ft into the Balm Creek Member. Debris ¯ow and ®nely laminated lacustrine tuffs comprise much of the channel ®ll overprinted by numerous siliceous paleosols. The Rose Creek Member was not identi®ed by Fisher and Rensberger (1972: section 10) whose measured section terminated below this stratigraphic level.

It may prove advisable after future ®eld A. tessellatus were considered part of the study to emphasize a lithologic and bioch- Haystack Valley Member by Rensberger ronologic distinction between Units 1 and (1983: 90, table 16). We concur, thinking it 2A. At this level in the Asher section there appropriate at present to include these beds in is an apparent division between green zeoli- the member because (1) the zeolitized coarse tized soft-weathering claystones and silt- sandstones of Unit 1 at one of the localities stones of Unit 1 carrying the Entoptychus in- yielding E. individens (SW¼,NE¼,NE¼, dividens±Allomys tessellatus rodent assem- SE¼, sec. 33, T8S, R25E: probably UCMP blage, and overlying pale gray-green to yel- locality Haystack 22) included a few highly low-gray well-indurated tuffaceous siltstones weathered welded tuff pebbles, suggesting of Unit 2A lacking this assemblage. that these coarse ¯uvial lenses mark the ®rst The rocks of Unit 1 with E. individens± access of local streams to exposures of the 26 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 7. The valley of Balm Creek (in foreground), looking north, with the Beardog Section (center of photograph) exposed along the east valley wall of a small tributary trending northwest from Balm Creek. This locality is in the W½,SW¼, T8S, R25E, Kimberly 7.5-minute quadrangle. a, Picture Gorge Group Basalt; b, Rose Creek Member; c, Balm Creek Member; d, GMAT marker tuff; e, Haystack Valley Member (revised).

Picture Gorge ignimbrite, the source of the claystone that in its lower part intertongues welded tuff pebbles in these beds; and (2) a with the yellow-gray tuffs of the lower part general similarity in the ``ribbed tuff'' li- of Unit 3. Occasional thin detrital volcani- thologies of Units 1 and 2A. clastic sands and airfall tuffs occur within the BALM CREEK MEMBER: Fisher and Rens- upper 25 feet of the lacustrine beds. The berger included all rocks of section 10 from presence of quiet-water lacustrine sediments the welded tuff conglomerates up to the cap- probably re¯ects the damming of local drain- ping basalts in their Haystack Valley Mem- ages in response to syntectonic compression ber. We think it useful to recognize a marked that modi®ed local topography during the de- change in the style of sedimentation within velopment of the Balm Creek syncline. this stratigraphic sequence that occurs at the Units 3 and 4 are lithically distinguished boundary between Units 2 and 3. Unit 3 in from the units below by their lack of zeoli- its lower part is comprised of yellow-gray tization. Unit 3 weathers to smooth slopes tuffaceous siltstones, claystones, and very with a soft powdery surface suggestive of ®ne sandstones with thin interbeds of low- montmorillonitic clay; Unit 4 is distinguished angle, cross-strati®ed gray ®ne-to-medium by fresh glass, including in some beds black sandstones. The yellow-gray beds appear to obsidian shards. be airfall tuffs occasionally reworked by Unit 4 is characterized by lower (beds I± lower-¯ow regime ¯uvial processes, punctu- IV) and upper (beds V±X) sets of stacked ated by slightly coarser detrital gray sands ®ning upward sequences (®g. 5, Beardog and gravel representing intermittent sheet- Section: Middle and Upper) of tuffaceous ¯ooding. The upper part of Unit 3 is made ¯uvial sandstones (some with coarse obsidi- up of thinly laminated lacustrine tuff and an shards), separated by a massive 15-ft thick 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 27

Fig. 8. Section HS-1, situated 0.5 mi (0.8 km) east of the Beardog Section, occupies the southern margin of the south limb of the Balm Creek syncline. Here steeply dipping strata of the south limb become horizontal and abut directly against the Richmond Fault of Fisher. The section from 2360 to 2660 ft (Units 2A, 2B) represents the Haystack Valley Member (revised), which is fossilferous except for Unit 2B. Unit 2B (2630±2660 ft) includes a ¯uvial channel complex at the base overlain by the GMAT marker bed; GMAT correlates Section HS-1 with measured sections of ®gures 4 and 5. The

Rose Creek Member is unconformable on older John Day rocks (EMU2), indicating removal of the upper part of the Balm Creek Member. 28 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 gray airfall tuff. This tuff is overprinted by a rocks were much younger than previously prominent paleosol (®g. 5, 425±440 ft). believed. Although no fossil mammals were The upper limit of Unit 4 is marked by a found in Unit 5 at Balm Creek, a represen- 12-ft thick terminal paleosol (®g. 5, 508±520 tative early Hemingfordian fauna (table 4; ft) developed on the uppermost ®ning-up- appendices 1-12 to 1-14, fauna of the Rose ward sequence (®g. 5, bed X). This paleosol Creek Member, see below) occurred in the is removed where incised by the ¯uvial chan- unit south of Kimberly. We eventually real- nels of Unit 5 (®g. 6). Unit 4 lithologies are ized that these particularly coarse polymictic clearly distinct from the zeolitized ribbed conglomeratic beds had no close temporal re- tuffs and conglomerates of our revised Hay- lationship with the monomictic welded tuff stack Valley Member. conglomerates of the revised Haystack Val- We designate Units 3 and 4 the Balm ley Member at Balm Creek and in Haystack Creek Member, noting that these units are Valley where 40Ar/39Ar dating of Unit 2A (ta- well exposed along the axis of the Balm ble 2, and ®g. 5, Beardog Section: Lower, Creek syncline (Map A: W½,SW¼, sec. 27, 25±45-ft level) revealed that the revised T8S, R25E). Because no fossil mammals Haystack Valley Member is a much older have been found in Unit 4 and only a single rock unit. fossil mammal (a peccary skull) has been found in Unit 3 at its base, age is conjectural. JOHN DAY VALLEY SOUTH OF KIMBERLY However, the peccary (table 4, cf. Hesper- hys) suggests a late or latest Arikareean age South of Kimberly along the east and west for Unit 3, which is supported by the strati- margins of the John Day valley, Fisher iden- graphic placement of Units 3±4 between ti®ed the Haystack Valley Member but did Units 2A (early late Arikareean) and 5 (early not map it as a discrete body of rock, having Hemingfordian, see below). included it with the Kimberly Member in an IMPORTANCE OF UNIT 5 IN REGIONAL COR- undifferentiated upper John Day unit (Fisher RELATION: As mentioned previously, Fisher and Rensberger, 1972: Map 2, map symbol and Rensberger's section 10 omitted Unit 5 Tjkh). We identi®ed the Haystack Valley because at that location post-John Day ero- Member of Fisher and Rensberger south of sion removed the uppermost rocks of the for- Kimberly using lithology and stratigraphic mation, and the resulting topographic low position above the Kimberly Member tuffs. was later ®lled by basalt ¯ows. Section 10 On both the east and west walls of the John was measured 0.43 mi (0.7 km) southwest of Day valley these rocks are separated from the the more complete Beardog Section where subjacent Kimberly tuffs by erosional uncon- Unit 5 escaped post-John Day erosion. The formity. Signi®cantly, only on the east wall existence of Unit 5 has important implica- does the member include polymictic, welded tions for regional lithostratigraphy and tec- tuff-bearing, pebble-to-cobble conglomerate, tonics in the Haystack Valley-Kimberly area. whereas along the west wall only intrafor- Unit 5 includes coarse polymictic welded mational monomictic vitric tuff-pebble con- tuff-bearing conglomerates and stream-de- glomerates occur. It is not possible to walk posited sands, debris ¯ows, lacustrine clay- out rocks of the Haystack Valley Member be- stones, and yellow-gray tuffaceous ®ne- tween Haystack Valley and Kimberly be- grained rocks with siliceous paleosols that cause post-John Day erosion has removed correspond to a similar lithofacies associa- much of the upper part of the formation, re- tion found in the highest stratigraphic unit of sulting in topographic lows ®lled by Colum- the upper John Day Formation along the east bia basalts. Along Highway 19 that follows wall of the John Day valley south of Kim- the John Day River between Haystack Valley berly. It became evident that the polymictic and Kimberly, one observes only isolated welded tuff-bearing cobble conglomerates of outcrops of the John Day Formation, mostly Unit 5 correlated to the equally coarse po- Kimberly gray tuff, separated by basalt-®lled lymictic welded tuff conglomerates of this valleys. high unit along the east wall of the valley ROSE CREEK MEMBER (EAST WALL OF THE south of Kimberly (table 1), and that these JOHN DAY VALLEY FROM KIMBERLY TO 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 29

TABLE 2 40Ar/39Ar analytical data for the Haystack Valley Member (Revised), Balm Creek, Oregon. 30 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

MCCARTY CREEK): The outcrops used by Creek (®gs. 6, 10). Both units show erosional Fisher and Rensberger to identify the Hay- and angularly unconformable relations with stack Valley Member south of Kimberly the John Day rocks below; both have basal were considered by them to be ``particularly polymictic conglomerates of similar clast well developed along the eastern margin of type and mean intermediate diameter for the the John Day River valley (secs. 7 and 17, 10 largest clasts (table 1, Ϫ6toϪ8ù, cobble T10S, R26E), where [this unit] lies in well- gravel of the Wentworth-Udden scale); both exposed contact with the underlying Kim- include similar lithologies and lithofacies berly Member.'' At the base of the unit they above the basal conglomerate; and both rep- recognized a distinctive welded tuff-bearing resent the stratigraphically highest unit in conglomerate (Fisher and Rensberger, 1972: their local areas beneath the Columbia ba- 17±18, ®g. 3, measured section 8). The unit salts. We select the name Rose Creek Mem- appears as a series of topographically high ber for this unit and designate as its type area exposures not exceeding 70±100 feet in the narrow belt of nearly continuous expo- thickness along the eastern wall of the John sures along the east wall of the John Day Day valley for about 7 miles south of Kim- valley south of Kimberly (®gs. 9, 10), ex- berly post of®ce (®g. 2). It lies with erosional tending from the NW¼, sec. 8, T10S, R26E and angular unconformity (EMU2) across on the north, through the west half of sec- both Kimberly and Turtle Cove Members, as tions 17, 20, 29, T10S, R26E, to the SW¼, Fisher noted. The polymictic welded tuff- sec. 32, T8S, R26E, Mt. Misery 7.5-minute bearing conglomerate, which can be traced at quadrangle (Maps C, D, sections 7±14). We the base of the unit from Spring Creek south also apply this member designation to Unit to McCarty Creek (®gs. 9±11, Maps C, D), 5 of the Balm Creek syncline based on sim- is strikingly similar in clast size and com- ilar lithologic, lithofacies, and positional re- position to the polymictic conglomerate at lations. the base of Unit 5 in the Beardog Section at This interpretation differs from that of Balm Creek (®g. 6, Map A). Fisher and Rensberger (1972), who suggest- A distinctive lithofacies association char- ed that the high unit south of Kimberly along acterizes the unit south of Kimberly: polym- the east wall of the John Day valley corre- ictic welded tuff-bearing conglomerate, de- lated with our Unit 2 and higher beds below bris ¯ows, coarse-shard tuffs (commonly the Columbia basalts in the Balm Creek sec- with vesicular obsidian shards), and thin beds tion. Their interpretation was based in part of dark gray airfall tuff. Interbedded ®ne- on the presence of welded tuff conglomerate grained tuffaceous claystones, siltstones, and in both Unit 2 of the Balm Creek section and very ®ne sandstones are commonly olive- in the Rose Creek Member. Here we empha- gray to pale yellow-gray and weather to size that the mean intermediate diameters and smooth slopes with occasional thin resistant clast types of the Unit 2 conglomerates of the interbeds. The debris ¯ows are lithically uni- revised Haystack Valley Member in the Balm form: welded tuff pebbles derived from the Creek sections differ from the Rose Creek Picture Gorge ignimbrite occur as well- Member conglomerates south of Kimberly as rounded matrix-supported clasts scattered shown in table 1. Unit 2A conglomerates are through ®ne-grained gray structureless tuff monomictic, welded tuff-bearing, pebble (®g. 12). These ¯ows are frequently heavily gravels, whereas Rose Creek Member con- overprinted by paleosol features: siliceous glomerates are polymictic (welded tuff and white rhizoliths, invertebrate burrows, devit- various other clasts) pebble-to-cobble gravels ri®cation, and soil brecciation. The unit is not considerably coarser than those of Unit 2. zeolitized and often terminates in sets of Furthermore, biochronologic and radiometric stacked siliceous paleosols. This lithofacies data discussed below show that Unit 2 at association is also present in Unit 5 on Balm Balm Creek and the Rose Creek Member Creek (®g. 6). along the east wall of the John Day valley An important lithologic correlation exists differ signi®cantly in age by as much as 5± between the high unit south of Kimberly on 6 million years (table 3). the east wall of the valley and Unit 5 at Balm WELDED-TUFF CONGLOMERATES OF THE 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 31 2735 to ϳ 300 ft (from Ͼ Fig. 9. Stratigraphic pro®le of the east wall of the John Day valley south of Kimberly from Spring Creek to Rose Creek. Downfaulting of 2400 ft) over a distance of 1 mi (1.6 km). The pro®le continues to the south in ®gure 10. the Rose Creekϳ Member along the Rose Creek fault and at faults in the vicinity of Spring Creek results in offset of 32 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 10. Stratigraphic pro®le of the east wall of the John Day valley south of Kimberly from Rose Creek south to McCarty Creek (Foree South). The Rose Creek Member rests with angular unconformity on older John Day rocks (Turtle Cove and Kimberly Members). The early Hemingfordian mammal fauna from the Rose Creek Member comes primarily from Picture Gorge 36 and sites at Rose Creek. 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 33

Fig. 10. Continued. The Rose Creek Member is characterized by a lithofacies association comprising polymictic welded tuff-bearing conglomerate ®ning upward to gray-black coarse detrital ¯uvial sandstone, debris ¯ows, coarse obsidian-shard tuff, and olive-gray tuffaceous siltstone and claystone, overprinted by siliceous paleosols.

ROSE CREEK MEMBER: Clasts derived from of these types can be recognized in hand the Picture Gorge ignimbrite, particularly specimen in the conglomerates and are sum- those from partially to strongly welded parts marized in table 1. Hay (1963: 206) recorded of the ¯ow, make up a signi®cant fraction of that the ``ignimbrite exhibits all gradations the pebbles and cobbles in the conglomerates between nonporous, highly welded tuff and of the Rose Creek Member. The term ignim- porous, friable lapilli tuff in which the par- brite is used here in the sense of Fisher ticles of glass are neither ¯attened by com- (1966b: 6) as a general term for all rock paction nor welded.'' types that are produced by dispersal of frag- According to Fisher (1966b), a basal air- ments from a pyroclastic ¯ow. Fisher ob- fall tuff and unwelded tuff making up the served several lithologies in the two cooling lowest two units of the ignimbrite are rhyo- units of the Picture Gorge ignimbrite, includ- litic and the higher units are dacitic in com- ing air-fall tuff, unwelded and partially weld- position. The air-fall tuff and the two over- ed lapilli tuff, and densely welded tuff. Some lying cooling units are composed mostly of 34 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 11. The Rose Creek Member at the early Hemingfordian mammal locality Picture Gorge 36. Coarse polymictic gravels, sands, and overlying tuffaceous sediments of the Rose Creek Member (rc) incise the Kimberly Member gray tuffs (k). Dashed white line marks the contact between the two units. vitric shards and pumice fragments which lapilli; (d) welded obsidian-like black glass can be readily observed in hand specimen clasts (not a true obsidian but an end-member under the binocular microscope. Lapilli in- densely welded tuff described by Ross and clude pumice and rock fragments of crystal- Smith, 1960: 24, ®g. 9); (e) gray lapilli tuff, line salic rocks and vitric welded tuff. The commonly with brown vitric welded tuff la- unwelded base of Fisher's cooling unit 1 is a pilli in a gray devitri®ed aphanitic ground- friable tuff that grades in places into welded mass. Locally, clasts of the densely welded tuff and a black obsidian-like glass. tuff display an attractive red and gray band- Many clast types in the Rose Creek Mem- ing. ber conglomerates can be attributed to the Smith (1960) observed that deformation of Picture Gorge ignimbrite. They are derived shards and pumice fragments is the only cer- from the densely welded or partially welded tain criterion for recognition of welding in portions of the ignimbrite and can be tuffs that have crystallized, which is often the grouped into discrete classes: (a) olive to case in geologically older ash ¯ows. Only the dark brown, crystal-poor, densely welded densely welded and/or crystallized units of tuff with dark pumice ®amme; (b) greenish- the ignimbrite have survived as clasts in the gray, pale yellow or red rhyolitic/dacitic conglomerates of the Rose Creek Member, welded tuff, with a few small euhedral or the unwelded tuff and air-fall tuff units being broken quartz and feldspar phenocrysts dis- too soft and friable to survive erosion and tributed in an aphanitic microcrystalline transport as cohesive clasts. Smith noted that groundmass (without dark pumice ®amme); the transition from unwelded to densely (c) partially to densely welded, gray to dark welded is accompanied by loss of pore space gray obsidian-shard tuff, often vitreous, ex- and progressive deformation of glass shards hibiting shard compaction and buff pumice and pumice. The Rose Creek conglomerate 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 35

Fig. 12. Debris ¯ows characterize Rose Creek Member outcrops along the east wall of the John Day valley south of Kimberly, at Sutton Mountain, and at Balm Creek. Rounded clasts of welded tuff derived from the Picture Gorge ignimbrite, vitric tuff from subjacent John Day units, and clasts of Deep Creek tuff, are supported in a ®ne-grained tuffaceous matrix. Masses of water-saturated volcanic ash moved downslope into stream courses where ¯uvial gravels were incorporated and deposited as local debris lenses.

TABLE 3 Biochronology and Geochronology of the Upper John Day Formation in the Haystack Valley-Kimberly Areas, John Day River Valley, Oregon 36 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Map B: Topographic map of the Kimberly area, John Day valley, Oregon, Kimberly 7.5-minute quadrangle, indicating the location of measured sections in Johnson Canyon: 19, Johnson Canyon East; 20, Johnson Canyon West. clasts derived from the ignimbrite, despite LY): Fisher's appraisal of the Kimberly and their textural and compositional variety, are Haystack Valley Members south of Kimberly nearly all entirely lacking in pore space and mentioned the stratigraphically high unit on many display deformed pumice clasts and the east wall (our Rose Creek Member) but compacted shards, indicating their derivation did not discuss or examine in detail the beds from the welded portions of the ignimbrite thought by him to be of the same age on the cooling units. It is important to consider west wall. Fisher and Rensberger (1972: ®g. whether in Miocene and older ash ¯ows such 3, measured section 9) presented a strati- as the Picture Gorge ignimbrite the compo- graphic section of the west wall at the mouth sition of the various ignimbrite zones may of Johnson Canyon; a conglomerate near the have been affected long after cooling by base in that section was questionably corre- burial and ground-water diagenesis, further lated with the basal conglomerate of the Rose obscuring the initial textural composition of Creek Member (their section 8) on the east the rocks (Smith, 1960: 153). However, wall. Haystack Valley and Kimberly Mem- transport of these clasts and subsequent dia- bers were mapped together as a single unit genesis in Rose Creek Member drainages has both on the west and east walls of the valley not prevented the identi®cation in hand spec- without differentiation (Fisher and Rensber- imen of the various ignimbrite phases. ger, 1972: Map 2). In addition to clasts derived from the ig- The best exposures of the Kimberly and nimbrite, the Rose Creek conglomerates also Haystack Valley Members (sensu Fisher and contain clasts derived from the Deep Creek Rensberger) along the west wall occur in tuff (Fisher, 1962) and from subjacent vitric Johnson Canyon 0.6 mi (1 km) southwest of tuffs of the John Day Formation incised by Kimberly and along the John Day river for the Rose Creek Member. 3 mi south of Kimberly (Maps B, C, sections JOHNSON CANYON MEMBER (WEST WALL 16±20). Here Kimberly gray tuffs are discon- OF THE JOHN DAY VALLEY SOUTH OF KIMBER- formable beneath, in fact incised by, yellow- 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 37 gray tuffaceous siltstones, sandstones, and at UCMP locality Picture Gorge 1. Named intraformational monomictic vitric tuff-peb- the ATR tuff (Fremd et al., 1994), it has been ble conglomerate (®gs. 13, 14, sections 16± dated by C.C. Swisher of Rutgers University 40 39 17, EMU1). These yellow-gray, ¯uvially at 22.6 Ϯ 0.13 Ma ( Ar/ Ar). The ATR tuff worked tuffaceous sediments and the subja- can be traced from Picture Gorge 1 north- cent gray tuffs of the Kimberly Member are ward into the principal measured section of evidently the rocks assigned to their Tjkh the Johnson Canyon beds along the canyon's map unit (Kimberly and Haystack Valley north wall (®g. 15, Picture Gorge 1 to John- Members) by Fisher and Rensberger (1972: son Canyon East). The tuff serves as a useful Map 2). It is these yellow-gray beds incised marker bed within Johnson Canyon and oc- into the Kimberly Member that we regard as curs at progressively lower elevations in out- a lithostratigraphic unit distinct from the crops from south to north, due to local block Haystack Valley Member (revised) along faulting. Balm Creek, and also lithically distinct from We designate the upper and lower units at the Rose Creek Member along the east wall. Johnson Canyon as the Johnson Canyon Outcrops at the mouth of Johnson Canyon Member. These rocks can be distinguished (Maps B, C, sections 18±20) constitute the from the Kimberly Member by the yellow- thickest measured section of these yellowish-gray color of the ®ne-grained tuffaceous gray tuffaceous rocks along the west wall of beds, the more varied lithofacies, more nu- the John Day valley south of Kimberly merous and deeply incised intraformational (ϳ355 ft, ®g. 15, from 1920 to 2278 ft). monomictic conglomerates of locally derived Here, these Johnson Canyon beds can be di- vitric tuff pebbles, the presence of the ATR vided into an upper unit conformably super- marker tuff, and superposition on the uni- posed on a lower unit (®g. 16). The lower formly gray Kimberly Member. When traced unit (®g. 15, ϳ1920±2130 ft) consists of ¯u- southward along the west wall of the John vial sequences of intraformational mono- Day valley, the Johnson Canyon Member mictic vitric tuff-pebble conglomerate, tuff- rests in erosional disconformity (EMU1)on aceous sandstone, and siltstone that ®ne up- the gray tuffs of the Kimberly Member (®g. ward, terminating in a prominent paleosol 13, Big Valley and UCMP V-6668 sections). that is continuous along the north wall of the A measured section at UCMP locality V- canyon for 0.25 mi (0.4 km). An enormous 6668 shows a basal channel of the Johnson tuff-®lled vertebrate burrow, probably exca- Canyon Member incised to a depth of at least vated by a large carnivore, can be seen in the 50 ft (®g. 13, 2310 ft) into the Kimberly north wall of Johnson Canyon penetrating Member gray tuffs. Similarly, at our Big Val- this well-developed land surface (®g. 15, sec- ley section, superposition of the Johnson tion 20, 2130 ft). Calcareous cementation Canyon Member on Kimberly beds is dem- predominates in the lower unit. Above the onstrated by a sharp erosional contact be- lower unit is an upper unit of three super- tween the underlying Kimberly gray tuff and posed massive gray airfall tuffs overprinted a channel ®ll of intraformational monomictic by siliceous paleosols (®g. 15, section 20, conglomerate at the base of the Johnson Can- 2130±2212 ft). These cliff-forming tuffs are yon Member (®g. 13, 2340 ft). succeeded by coarse obsidian-shard tuffs, The Johnson Canyon Member is best ex- each ®ning upward and ending in a devel- posed in Johnson Canyon and in isolated out- oped paleosol horizon (®g. 15, section 20, crops along the west wall of the John Day 2212±2250 ft). The upper unit concludes valley south of Kimberly above the uniform- with ϳ25±30 ft of yellow-gray tuffaceous ly gray Kimberly beds that form the base of sandy siltstones beneath the capping basalts. the cliffs. We have not observed the member The upper unit is characterized by silica ce- south of the N½,NE¼, sec. 12, T10S, R25E. mentation, especially in the paleosol hori- South of this area the unit appears to have zons rich in siliceous rhizoliths. been removed by erosion, if it was ever pre- A white vitric tuff is present in the lower sent. The member occurs north of Johnson unit of the Johnson Canyon beds 0.4 mi (0.6 Canyon in the area of the Kimberly post of- km) south of the mouth of Johnson Canyon ®ce and can be traced northwest of Kimberly 38 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 39 into the SE¼, sec. 24, T9S, R25E, and to the ®rmed by the mammal faunas from the two center, sec. 13, T9S, R25E along Bologna members (table 4) and by radiometric dating. Creek. RELATIVE AGES OF THE JOHNSON CANYON When we ®rst studied the outcrops in AND ROSE CREEK MEMBERS: High on the west Johnson Canyon and other less exposed sec- wall of the John Day valley in the NE¼, sec. tions of upper John Day rocks south of Kim- 12, T10S, R25E, opposite the mouth of berly along the west wall, we were surprised Spring Creek, is an isolated outcrop overly- to discover that there are no welded tuff- ing several hundred feet of Kimberly gray bearing conglomerates in these beds. The tuff at an elevation of 2670 feet (®gs. 13, 14, seven to eight levels of intraformational Dead Cow Gulch Section). The outcrop be- monomictic conglomerate in Johnson Can- gins with 70 feet of gray-brown claystones yon and nearby outcrops contain only locally overlain by 30±40 feet of tuffaceous sedi- derived, well-rounded, indurated vitric tuff ments that include distinctive coarse obsidi- clasts from subjacent tuffs incised by the an-shard tuff and debris ¯ow units heavily channel scours (table 1, Johnson Canyon). overprinted by paleosols. This facies associ- Also, the clast size was much smaller on av- ation is similar to Rose Creek Member lith- erage (mean I.D. Ͻ7 cm) than that for the ofacies along the east wall of the valley (®gs. basal conglomerate of the Rose Creek Mem- 9, 10) where welded tuff conglomerates and ber on the east wall (mean I.D., 12 samples, ¯uvial sandstones yielded an early Heming- 12.9 cm, table 1). This prompted us to ask fordian fauna (ϳ18.2±18.8 Ma) and were ac- whether the outcrops attributed by Fisher to companied by obsidian-shard tuff and debris the ``Haystack Valley Member'' on the west ¯ow units with strong paleosol overprints. and east walls might not represent two dif- There is an evident correlation between these ferent lithostratigraphic units that differed to topographically high obsidian-shard tuff and some degree in age. Alternatively, they might debris ¯ow units on the east and west walls be temporally equivalent sediment prisms of the valley, suggesting that the Dead Cow distinguished by lithofacies formed in differ- Gulch outcrop is an isolated outlier of the ent depositional environments. In the latter Rose Creek Member, in fact the only vestige case the polymictic welded tuff conglomer- of the member remaining along the west wall ates of the east wall and the monomictic in- of the valley south of Kimberly. traformational conglomerates of the west The Rose Creek Member outlier rests on wall might simply re¯ect stream access to a mature calcareous paleosol at the top of the geographically disjunct source areas, one de- Kimberly Member that probably represents riving welded tuff clasts from the Picture the uneroded, intact terminal surface of Kim- Gorge ignimbrite, the other from a drainage berly deposition (®g. 13, Dead Cow Gulch network of local gullies lacking access to the Section, 2670 ft). A short distance (2000 ft) ignimbrite. We looked for other evidence that to the northeast at UCMP Locality V-6668 might provide insight into this problem. It (®g. 13), a basal channel of the Johnson Can- was the unexpected discovery of a topo- yon Member that deeply incises the Kimber- graphically high remnant outcrop of the Rose ly gray tuffs at this location is 360 ft below Creek Member on the west wall of the valley the high remnant outlier of the Rose Creek that initially suggested that the Rose Creek Member. Although these outcrops are sepa- Member and the Johnson Canyon Member rated by a normal fault (®g. 13, Dead Cow might be distinct lithostratigraphic units of Gulch fault), the contrast in lithofacies and different ages. This was eventually con- lithology in such a short distance is so

← Map C: Topographic map of the John Day valley south of Kimberly, Oregon, Mt. Misery 7.5-minute quadrangle, from Johnson Canyon to Rose Creek, indicating measured sections along the east and west walls of the valley: 7, Spring Creek; 8, Rose Creek North; 9, Rose Creek South; 15, Dead Cow Gulch; 16, UCMP locality V-6668; 17, Big Valley; 18, Picture Gorge 1. 40 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 41 marked that lateral equivalence is unlikely. gray cliff-forming tuff containing small ob- The Johnson Canyon Member to the north of sidian shards, heavily overprinted by multi- the Rose Creek outlier is capped by basalt ple stacked siliceous paleosols (®g. 17, Unit ¯ows that abut against the fault; thus, if the D, upper sequence); this upper sequence, 25± Rose Creek Member was superposed on the 30 ft thick, is expressed topographically as a Johnson Canyon Member at an earlier time, prominent bench along the south face of Sut- it was removed by erosion before in¯ux of ton Mountain (center and NW¼, sec. 33, the basalts. T10S, R21E). At no locality south of Kimberly have we Resting on Unit D, and in places incised been able to observe the Rose Creek Member into it with low relief, are ϳ100 ft of coarse in superposition on the Johnson Canyon polymictic welded tuff conglomerates, me- Member. However, biochronologic evidence dium to coarse detrital sandstones, and discussed below reveals that the mammalian brown to gray-brown tuffaceous claystones fauna obtained from the Johnson Canyon and siltstones interbedded with debris ¯ow, Member in Johnson Canyon is older than gray lapilli tuff, and coarse obsidian-shard that from the Rose Creek Member, and at tuff units: these make up a series of ¯uvial Sutton Mountain the Rose Creek Member ®ning-upward volcaniclastic units, most be- overlies rocks that closely resemble the John- ginning with welded tuff conglomerate, grad- son Canyon Member south of Kimberly (®g. ing upward through detrital sandstones to 17). brown claystones. These lithofacies are typ- UPPER JOHN DAY UNITS AT SUTTON MOUN- ical of the Rose Creek Member south of TAIN: Hay (1963: ®g. 2, sections 10±11) mea- Kimberly, suggesting correlation of the sured particularly thick sections of the upper member over an east-west distance of ϳ32 John Day beds north of Bridge Creek along mi (51.5 km). the southern face of Sutton Mountain. We In lithology, lithofacies, color, and style of studied the same area which occurs in the sedimentation, Unit D resembles the Johnson N , sec. 33, and S , sec. 29, T10S, R21E, ½ ½ Canyon Member south of Kimberly. The Sutton Mountain 7.5-minute topographic ϳ350 ft of section making up the lower se- quadrangle (®g. 17, appendices 1-10, 1-11). quence of Unit D closely corresponds to the Upper John Day rocks ®ll a large paleovalley lower unit in Johnson Canyon (®g. 15), and incised into earlier John Day beds in the center of section 33; this paleovalley ®ll is the upper sequence of gray tuff overprinted downdropped to the north along a west to by siliceous paleosols is much like the upper east-trending normal fault mapped by Hay unit in Johnson Canyon, although not as (1963: ®g. 2, sec. 33, T10S, R21E). A thick, thick. We also have collected teeth of the lithically homogeneous series of yellow-gray small chalicothere Moropus cf. oregonensis upper John Day tuffaceous sandstones, silt- from both the lower sequence of Unit D at stones, and occasional lenses of intraforma- Sutton Mountain and from the lower unit of tional monomictic vitric tuff-pebble gravel the Johnson Canyon Member in Johnson ®lls the paleovalley north of the fault, con- Canyon. Thus, we consider Unit D at Sutton tinuing upward for ϳ350 ft (®g. 17, Unit D, Mountain the lithologic and temporal correl- lower sequence6). This rock unit is overlain ative of the Johnson Canyon Member. along a sharp but conformable contact by a Superposition of the Rose Creek Member on the Johnson Canyon Member could not 6 Units A±C of our measured sections north of Bridge be demonstrated in the John Day valley south Creek are not described in this report. of Kimberly. However, along the south face

← Map D: Topographic map of the John Day valley south of Kimberly, Oregon, Mt. Misery 7.5-minute quadrangle, from Bone Creek south to McCarty Creek, indicating measured sections along the east wall of the valley: 10, Picture Gorge 36; 11, Breck Draw North; 12, Branson Creek; 13, Ancient Juniper; 14, Foree South. 42 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 43 of Sutton Mountain the superposition of the Day rock units in our study. Representative Rose Creek Member on the Johnson Canyon mammal faunas from the stratigraphically Member (Unit D) supplies this critical strati- lowest and highest rock units in the Haystack graphic relationship. Thus, the Johnson Can- Valley Member (sensu lato) of Fisher and yon Member, considered older than the Rose Rensberger suggested that its age spanned Creek Member based on the mammalian fau- about ®ve million years, from ϳ23.5±23.8 na from Johnson Canyon south of Kimberly, Ma to as young as ϳ18.2±18.8 Ma (table 3), can be documented in stratigraphic relation a much greater interval than previously sus- beneath the Rose Creek Member at Sutton pected. The biochronologic data were sup- Mountain, supporting the age relationship plemented by 40Ar/39Ar age determination of suggested independently by faunal content. feldspars from the Haystack Valley Member (revised) at Balm Creek (table 2). BIOCHRONOLOGY Mammals found by UNSM from 1993 to The richly fossiliferous mid-Cenozoic 2001, or otherwise identi®ed in paleontolog- rocks of the John Day region of central ical collections conserved at the University Oregon have attracted attention since the of California-Berkeley, Yale University, the 1860s when Thomas Condon ®rst collected American Museum of Natural History, and fossil mammals in the John Day basin. Alert- Princeton University, were assigned to the ed by Condon's discoveries, O.C. Marsh of four members identi®ed in our ®eld study Yale University visited the region in 1871 (table 4). Only specimens that could be at- and employed Leander Davis and William tributed without question to a member are Day to collect John Day mammals during ex- included in table 4. Exact stratigraphic place- tensive ®eld explorations from 1871 to 1877. ment of the fossils is documented in the ap- In 1878±1879, Jacob Wortman and Charles pendices. Sternberg made important contributions to FAUNA OF THE HAYSTACK VALLEY MEMBER E.D. Cope's John Day collection, now con- (REVISED): Fossil mammals occur throughout served in the American Museum of Natural Units 1 and 2A in the Balm Creek drainage History, and in 1889 a party led by W.B. but were not found in Unit 2B. The fauna Scott acquired a signi®cant collection for from Units 1 and 2A includes rodents, sev- Princeton University. However, more disci- eral oreodonts, two species of peccaries, a plined investigations of the geology of the horse, a dicerathere rhinoceros, temnocyo- John Day sequence began with the Univer- nine amphicyonids (beardogs), and rare spec- sity of California expeditions under John C. imens of small canids, camel, tapir, and hy- Merriam in 1899±1900. Merriam and his stu- pertragulid (table 4; appendices 1-1 to 1-7). dents established the ®rst lithostratigraphic The stage of evolution of the fauna suggests and petrographic characterization of the John an early late Arikareean age based on cor- Day beds, and at the same time added sig- relation with the Great Plains sequence. ni®cantly to knowledge of the fauna, partic- Key mammalian taxa indicative of this age ularly in terms of the biostratigraphic suc- assignment are Miohippus, Hypertragulidae, cession. temnocyonine beardogs, and advanced spe- Fossil mammals proved particularly im- cies of the rodents Entoptychus (E. indivi- portant in age determination of upper John dens) and Allomys (A. tessellatus), none of

← Fig. 13. Stratigraphic pro®le of the west wall of the John Day valley south of Kimberly. North of the Dead Cow Gulch fault the Johnson Canyon Member is incised into the Kimberly Member at UCMP V-6668 and Big Valley. The Johnson Canyon Member is downfaulted at least 400 ft between Big Valley, where the base of the member is exposed at 2340 ft, and Johnson Canyon, where the base occurs in the subsurface at unknown depth. South of the Dead Cow Gulch fault a terminal paleosol developed on the Kimberly Member is overlain at high elevation by an isolated outlier of the Rose Creek Member. Conglomerates of the Johnson Canyon Member are intraformational vitric tuff-pebble gravels; welded tuff clasts derived from the Picture Gorge ignimbrite are absent. 44 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 14. West wall of the John Day valley south of Kimberly at Dead Cow Gulch. The Rose Creek Member (a) caps the Dead Cow Gulch measured section (®g. 13) south of the Dead Cow Gulch fault (c); the V-6668 section (b) occurs north of the fault. The fault follows a west-to-east trend along the gulch; ¯ows of the Picture Gorge Basalt (d) abut the fault plane. which has been found stratigraphically high- not ®nd entoptychines above Unit 1 in the er than the rede®ned member in the study Balm Creek sections. area. Also useful in this regard are primitive The fauna together with the radiometric forms of the camel Paratylopus, the oreodont date from Unit 2A (table 2) indicate that the Paroreodon, the tapir Nexuotapirus, and a revised member is much older than previ- rodent possibly representing the earliest ously believed, certainly older than the Rose known species of Schizodontomys (or, alter- Creek Member and its fauna south of Kim- natively, a transitional form intermediate [?] berly, which previously had been correlated between Tenudomys and Schizodontomys). by Fisher and Rensberger (1972) with rocks The rodent Entoptychus individens, the of the revised member in the Balm Creek most advanced entoptychine reported by area. Rensberger (1971), occurs in the lower part FAUNA OF THE BALM CREEK MEMBER:A of Unit 1 (®g. 4, Asher Section, 0 to 48 ft) partial skull of a small peccary (cf. Hesper- ϳ250 to 280 ft below the dated tuffaceous hys; appendix 1-6) was found in the lowest siltstone (®g. 5, 25±45-ft interval, ϳ23.5± few feet of Unit 3 (®g. 5, 175±180 ft, Bear- 23.8 Ma; table 2) in Unit 2A of the Beardog dog Section: Lower); no additional fossil Section. This rodent is common in the low- mammals are known from Unit 3, or from ermost ϳ48 feet of Unit 1 as exposed along Units 4 and 5. The peccary indicates a late Balm Creek; however, because the lower Arikareean age for the base of Unit 3, an age contact of Unit 1 with subjacent rocks was which probably pertains to the lithically uni- not identi®ed anywhere in the Balm Creek form yellow-gray tuffaceous siltstones, clay- drainage, we are not able to determine the stones, and very ®ne sandstones comprising proximity of this rodent zone to the base of the lower half of the unit (®g. 5, 175±260 ft, the revised Haystack Valley Member. We did Beardog Section: Lower and Middle) that 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 45 produced this fossil. Because the lacustrine tained these collections as two separate as- beds in the upper half of Unit 3 intertongue semblages since future sampling of these with the lower yellow-gray tuffaceous sedi- rocks may indicate a difference. The assem- ments, it seems likely that the entire unit is blage below the tuff is designated here the of late Arikareean age. The age of Unit 4 Picture Gorge 1 local fauna; the assemblage remains uncertain. above the tuff is named the North Wall local FAUNA OF THE JOHNSON CANYON MEMBER: fauna. The North Wall local fauna includes Only the lower unit of the Johnson Canyon not only the fossils from the lower unit of Member contains fossil mammals; none have the member along the north wall of the can- been found in the near vertical exposures of yon, but also the mammals from UCMP Lo- the cliff-forming upper unit. Fossil mammals cality V-6431 (Little Dike Locality) on the from the lower unit are frequently found in south side of the canyon where fossils were the intraformational conglomerates as also found above the ATR tuff. stream-abraded specimens, but scattered and Almost all fossil mammals from the John- partially articulated individuals also occur in son Canyon Member come from exposures the ®ne-grained tuffaceous siltstones and at UCMP Picture Gorge 1 (V-6666) and Lit- sandstones. We examined specimens from tle Dike (V-6431) localities, and from the this unit at the University of California- north wall outcrops (UCMP locality V-6432) Berkeley and at the Los Angeles County Mu- at the mouth of Johnson Canyon (®g. 15; ap- seum of Natural History, combining these pendices 1-8, 1-9). The North Wall and Pic- identi®cations with those from our own ®eld- ture Gorge 1 local faunas compare with work. Comparison of the fauna of the lower mammals collected by University of Nebras- unit with the Great Plains sequence suggests ka and American Museum ®eld parties from a late to latest Arikareean age. the Harrison Formation and Upper Harrison Age-indicative taxa of the Johnson Can- beds in Sioux County, Nebraska, and in ad- yon Member include an advanced species of jacent Goshen and Platte Counties, Wyoming the oreodont Paroreodon; the camel Para- (Hunt, 1985, 1990). The Upper Harrison tylopus cameloides; the peccary Hesperhys; beds include the Eagle Crag Ash, a volcanic an unidenti®ed dromomerycid and the mos- tuff dated by the ®ssion track method (zir- chid deer Parablastomeryx cf. advena; the con) at 19.2 Ϯ 0.5 Ma (Hunt et al., 1983). equids Archaeohippus, cf. Kalobattipus, and The Agate Ash within the Harrison Forma- an anchithere; the small chalicothere Moro- tion has recently been dated by the 40Ar/39Ar pus oregonensis; the rhinoceroses Dicerath- method to ϳ23 Ma (sanidine, 22.9 Ϯ 0.08; erium and a small species of Menoceras; the Izett and Obradovich, 2001). In our view this tapir Miotapirus harrisonensis; the rodents provides an age range for the lower unit of Schizodontomys and Mylagaulodon; the the Johnson Canyon Member, and according- small canids Cynarctoides cf. luskensis and ly we have indicated its age in table 3 as Desmocyon thomsoni; and temnocyonine within the ϳ19.2±22.6 Ma interval. amphicyonids (table 4). We have not found The fauna of the lower unit of the Johnson Miohippus, parahippine horses, hypertragu- Canyon Member contains a number of taxa lids, or entoptychine and allomyine rodents that indicate a younger age relative to that of in the Johnson Canyon Member. Miohippus, the fauna from the Haystack Valley Member Entoptychus, and Allomys are presumably (revised). Johnson Canyon fossils of the or- extinct by the time of deposition of the mem- eodont Paroreodon, the camel Paratylopus, ber; parahippine horses ®rst appear in the de- and the rodent Schizodontomys represent spe- posits of the Rose Creek Member. cies more evolved than those of the same Fossils collected from the lower unit at the genera in the revised Haystack Valley Mem- mouth of Johnson Canyon were plotted rel- ber along Balm Creek. The absence of Mio- ative to the ATR tuff (ϳ22.6 Ma), which hippus in the Johnson Canyon Member and serves as a useful stratigraphic datum. The the ®rst appearance of dromomerycid and mammalian fauna from below the ATR tuff moschid cervoids also support this inference. does not appear to differ from the fauna col- Temnocyonine amphicyonids occur in both lected above the tuff. However, we main- the revised Haystack Valley Member and the 46 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 47

Johnson Canyon Member but are absent and gocene to mid-Miocene) of the central Great presumed extinct in the Rose Creek Member. Plains has resulted in a detailed biochronol- The presence of a small species of Menocer- ogy calibrated by radiometric and paleomag- as, temnocyonines, the canid Cynarctoides netic data (MacFadden and Hunt, 1998; cf. luskensis, and tapir Miotapirus harriso- Woodburne and Swisher, 1995; Tedford et nensis, and the absence of advanced parahip- al., 1987, 1996; Hunt, 1985, 1990). During pine horses in the lower unit at Johnson Can- our exploration of the Rose Creek Member yon indicate a fauna older than that of the on the east wall of the John Day valley south Rose Creek Member on the opposite wall of of Kimberly, we discovered fossil mammals the John Day valley. useful in age determination at two localities. The heteromyid rodent Schizodontomys is We were also able to examine specimens represented in the lower unit at Johnson Can- from this unit at the University of California- yon not only by the genoholotypic species (S. Berkeley collected in the 1960s by Lester greeni, UCMP 39435, Rensberger, 1973: 62), Kent and John Rensberger. Comparison of which was almost certainly collected from these mammals with the Great Plains se- below the ATR tuff at Picture Gorge 1, but quence suggests a precise correlation and also by several individuals of Schizodonto- supplements the earlier work of Rensberger mys found by us above the tuff at UCMP using fossil rodents. locality V-6432. The UNSM specimens of Age-indicative taxa discovered in the Rose the genus are dentally advanced and compare Creek Member include the large oreodont with a species of Schizodontomys (S. cf. Merycochoerus magnus; the dromomerycid harkseni, UW 1298, Munthe, 1981) belong- Barbouromeryx; the small moschid deer Par- ing to a fauna from the uppermost part of the ablastomeryx; the camels Paratylopus and a Arikaree Group (Upper Harrison beds) at larger species (cf. Protolabis); the equids Uva, Platte County, Wyoming. We suggest Parahippus pawniensis, Archaeohippus, and that the Uva fauna is one of the youngest an anchithere; the chalicothere Moropus or- Arikareean assemblages yet discovered in the egonensis; a large beaver (cf. Hystricops); type area of the Arikareean NALMA and, the rodents Sewellelodon and Mylagaulodon; based on the dated Eagle Crag Ash, probably and two new species of wolf-sized amphi- falls near ϳ19 Ma. Thus, because of the near cyonid carnivorans (table 4; appendices 1-12 identity of Schizodontomys from Johnson to 1-15). We have not found Miohippus, Pa- Canyon and from Uva, we consider that the roreodon, temnocyonines, entoptychine and 22.6 Ma date for the ATR tuff may be a max- allomyine rodents, the heteromyid Schizo- imum age, and that the fauna from the lower dontomys, or tapirs in the Rose Creek Mem- unit in Johnson Canyon more likely falls ber. somewhere within the younger part of the These mammals are designated the Picture ϳ19.2±22.6 Ma interval. Gorge 36 local fauna. Almost all specimens FAUNA OF THE ROSE CREEK MEMBER: Study have come from Rose Creek outcrops at of mid-Cenozoic mammal faunas from the UCMP locality Picture Gorge 36 (appendix Arikaree and Hemingford Groups (late Oli- 1-14) at the head of Bone Creek and UNSM

← Fig. 15. Stratigraphic pro®le of the Johnson Canyon Member at the mouth of Johnson Canyon, west wall of the John Day valley south of Kimberly. The Johnson Canyon East and West sections are the thickest and most representative of the member in the vicinity of and south of Kimberly. The Picture Gorge 1 section is correlated with the Johnson Canyon East and West sections at the mouth of Johnson Canyon by the ATR tuff. The lower unit of the member is fossiliferous and characterized by ¯uvial sequences of monomictic vitric tuff-pebble conglomerate ®ning upward into tuffaceous yellow-gray ®ne sandstones and siltstones overprinted by calcareous paleosols. A marked change in the paleosols occurs at the contact with the upper unit: the airfall ®ne-grained gray tuffs and coarse obsidian-shard tuffs of the upper unit are overprinted by stacked siliceous paleosols. Superposition of the upper on the lower unit is also seen at Sutton Mountain (®g. 17). 48 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 16. The Johnson Canyon Member exposed along the north wall of Johnson Canyon southwest of Kimberly. The contact (c) between the lower and upper units of the member separates tuffaceous sediments with calcareous paleosols in the lower unit from tuffs overprinted by siliceous paleosols in the upper unit. Monomictic vitric tuff-pebble conglomerates (m) are restricted to the lower unit. PG, Picture Gorge Basalt Subgroup. localities at the head of Rose Creek (appen- dont Merycochoerus magnus, which is found dices 1-12, 1-13). This local fauna compares within a sharply restricted interval in the with a fauna collected by University of Ne- Great Plains and Rocky Mountains, the ear- braska and American Museum ®eld parties liest Hemingfordian (®g. 18). In the Rose in Sioux County, Nebraska (Northeast of Ag- Creek Member this index oreodont is asso- ate local fauna, MacFadden and Hunt, 1998: ciated with the dromomerycid Barbourome- 163). The Northeast of Agate local fauna oc- ryx; the small moschid Parablastomeryx curs within sediments of the lower part of the schultzi; a large beaver near Hystricops; the Runningwater Formation paleomagnetically equid Parahippus pawniensis; and a large, calibrated at ϳ18.2±18.8 Ma (MacFadden long-limbed species of daphoenine amphi- and Hunt, 1998: 162±163) and is stratigraph- cyonid, all closely related to similar species ically above the Eagle Crag tuff in the Upper in the early Hemingfordian faunas of western Harrison beds dated by ®ssion-track (zircon) Nebraska. at 19.2 Ϯ 0.5 Ma (Hunt et al., 1983). Con- The Picture Gorge 36 local fauna is the sequently, we believe that the Picture Gorge youngest fauna known from the John Day 36 local fauna indicates an age of ϳ18.2± Formation in the type area south of Kimber- 18.8 Ma for the Rose Creek Member south ly. It dates the Rose Creek Member and pro- of Kimberly. vides an upper limit on John Day deposition The association of several Rose Creek at ϳ18 Ma for the eastern facies of the for- Member taxa support an early Hemingfor- mation. The geographic extent of the Rose dian age, equivalent to that of the lower part Creek Member also suggests, based on its of the Runningwater Formation in Nebraska. presence (without fauna) in Haystack Valley Most important is the presence of the oreo- (Balm Creek, Unit 5) and at Sutton Moun- 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 49

TABLE 4 TABLE 4 Fossil Mammals of the Upper John Day Formation (Continued) (exclusive of the Kimberly Member), Haystack Valley and Kimberly Areas, Oregon

tain, that ¯uvial incision and deposition were geographically widespread at the beginning of the Hemingfordian in the Kimberly, Mt. Misery, and Sutton Mountain topographic quadrangles.

40Ar/39Ar DATING The early late Arikareean age determined on faunal evidence for the revised Haystack Valley Member along Balm Creek is supported by age determinations on feldspars from a tan tuffaceous unit sampled within the 25±45-ft interval of the Lower Beardog Sec- tion (®g. 5, UNSM sample JD-BC-3). Three to ®ve crystals were used for each of four age determinations by Wm. McIntosh and L. Peters, New Mexico Geochronology Re- search Laboratory, New Mexico Bureau of Geology, Socorro (table 2). The higher K/Ca values for age determinations 7251-01 (23.8 Ϯ 0.06 Ma) and 7251-03 (23.6 Ϯ 0.07 Ma) suggest that all or most of the feldspar crystals in these samples were sanidines; these were considered the more reliable age data for the tuff. It is unlikely that detrital grains (xeno- 50 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 17. Uppermost rock units of the John Day Formation at Sutton Mountain, Wheeler Co., Oregon. These exposures along the southern face of Sutton Mountain demonstrate the superposition of the Rose Creek Member on the Johnson Canyon Member (``Unit D'') in the N½, sec. 33, T10S, R21E, Sutton Mountain 7.5 min. quadrangle, in an area where Hay (1963: ®gs. 2, 4, section 11) also measured one of his thickest sections of upper John Day rocks. 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 51

Fig. 18. The Great Plains oreodont Merycochoerus from the Childs Frick collection of the American Museum of Natural History provides biochronologic control for the occurrence of this indicator taxon in the upper John Day Formation, Oregon. Because many upper John Day taxa are represented by few individuals, the larger Frick samples of Arikareean/Hemingfordian mammals are critical to accurate age assessment. The graph measures the progressive fusion and posterior extension of the premaxillae in time, most likely related to ongoing development of a tapirlike proboscis in this large oreodont. The triangle labelled ``Rose Creek'' refers to UCMP 76848 from Picture Gorge 36. crysts) were included: the tuff is massive, (1989) and references therein. Upper John ®ne-grained, and appears to be made up Day rocks studied by us belong to the eastern largely if not entirely of air-fall debris. facies of the formation and occur in Wheeler The latest Arikareean age (ϳ19 Ma) sug- and Grant Counties, Oregon, south of the gested by some species in the fauna from the Blue Mountains uplift and north of the Och- lower unit of the Johnson Canyon Member oco and Aldrich Mountains (®g. 1). The appears somewhat discordant with a single- structural pattern in this area was decribed crystal laser-fusion 40Ar/39Ar date by C.C. by Fisher (1967). John Day rocks in the Hay- Swisher (Rutgers University) of 22.6 Ϯ 0.13 stack Valley-Kimberly area were deposited Ma from the Across-the-River Tuff (ATR in a basin bounded on the north by the Blue Tuff, Fremd et al., 1994), but ongoing sam- Mountains and on the south by the Middle pling and analysis may prove this to be a Mountain-Richmond anticlinal trend (Fisher, credible age for the lower unit. 1967: 119±120, ®g. 2). The tectonic history of the John Day For- REGIONAL AND LOCAL STRUCTURE mation in the Haystack Valley-Kimberly area The tectonic setting of the John Day For- during the early Miocene involved both com- mation has been discussed by Fisher (1967), pressional and extensional episodes. North- Fisher and Rensberger (1972), Walker west-southeast to north-south compression, (1977), Robinson, Brem and McKee (1984), evidenced by folding and reverse faulting in and Walker and Robinson (1990b); a region- Haystack Valley and by moderate warping al overview following eruption of the ¯ood- south of Kimberly, deformed the older John basalts is provided by Hooper and Conrey Day units and involved rocks as young as the 52 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

revised Haystack Valley Member and Balm Rose Creek Member. EMU3 occurs at the Creek Member in Haystack Valley, and the base of the Picture Gorge Basalt Subgroup Kimberly Member in its type area. This ap- (PGBS), a subdivision of the Columbia River pears to have been a progressive, syntectonic Basalt Group (CRBG). compressional deformation that initiated the The demonstration of an early Miocene deposition of coarse upper John Day clastics angular unconformity (EMU2, ®gs. 5, 6, 8± along structural lows formed by synclinal 10, 13, 17) beneath the Rose Creek Member, folding. The master synclinal structures ex- separating it at several localities from the re- tended for tens of miles (Fisher, 1967). mainder of the John Day Formation, carries An interval of stasis followed, punctuated far-reaching implications for the stratigraphy by several eruptions of pyroclastic material of the formation and regional tectonics. In from the Cascade Range (upper unit of the the Haystack Valley-Balm Creek area, upper Johnson Canyon Member). We have not es- John Day rocks of the revised Haystack Val- tablished whether a ®nal compressional epi- ley and Balm Creek Members were de- sode, or instead, an early phase of extension- formed and faulted before deposition of the al deformation, triggered the deposition of Rose Creek Member. South of Kimberly, the Rose Creek Member's coarse ¯uvial and northward tilting, deformation, and subse- debris ¯ow facies, which clearly records a quent erosion of the Turtle Cove and Kim- region-wide synchronous event. During the berly Members of the John Day Formation ®nal phase of Rose Creek deposition, how- preceded deposition of the Rose Creek Mem- ever, the ®ne-grained tuffs and stacked sili- ber. Angular unconformity of the Rose Creek ceous paleosols demonstrate that the familiar Member on subjacent John Day rocks is best pattern of sporadic pyroclastic input, soil de- demonstrated south of Kimberly along the velopment, and erosion was again reestab- east wall of the John Day valley where the lished. unit lies on Turtle Cove Member green clay- Nonetheless the evidence for post-Rose stones to the south and on Kimberly Member Creek extension is incontrovertible. Exten- gray tuffs to the north (®g. 10). Later, the sional tectonism down-dropped John Day John Day Formation, including the Rose sediments along normal faults in the Hay- Creek Member, underwent extensive faulting stack Valley and Kimberly areas, creating a and erosion during the late early Miocene series of half-graben or grabens that seem to (ϳ18.0±?17 Ma) prior to the eruption of the coincide with the earlier synclinal axes pro- Columbia ¯ood-basalts. The oldest local duced during compression (Mitchell and ¯ows are those of the Twickenham Basalt Monument synclines of Fisher). Basalt ¯ows (Picture Gorge Basalt Subgroup) that ®lled abutting nearly vertical fault scarps south of valleys and other topographically low areas, Kimberly, at Rose Creek and Dead Cow eventually completely burying the varied to- Gulch (®g. 14), suggest that the ®nal phase pography developed on the John Day For- of extensional faulting was soon followed by mation (Bailey, 1989; Hooper and Swanson, the initial basalt ¯ooding in the region 1990). The unconformity between the Co- (Twickenham Basalt). lumbia basalts and the John Day Formation

Our identi®cation of regional unconfor- in the study area is designated EMU3 (®gs. mities extending over much of the study area 5, 6±10, 13, 15). contributed to the recognition of the major Synclinal folding and subsequent down- lithogenetic depositional units, designated faulting geographically isolated remnants of here as members, within Fisher and Rens- uppermost John Day rocks so that they es- berger's (1972) Haystack Valley Member caped post-John Day erosion. Although sensu lato. These members are of early Mio- much of the upper John Day in Haystack cene age, and hence the bounding unconfor- Valley has been removed by recent erosion, mities are designated in our measured sec- in the Balm Creek area the uppermost beds tions as EMU1 (early Miocene unconformi- are preserved on a down-faulted block that ty), EMU2, and EMU3. EMU1 marks the un- incorporates the Balm Creek syncline and an conformity at the base of the Johnson adjacent anticlinal upwarp to the south (Fish-

Canyon Member. EMU2 is at the base of the er, 1967: ®g. 5). Similarly, downdropped 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 53

Fig. 19. Structural trends south of the Blue Mountains in the eastern subregion along the course of the John Day River (modi®ed from Fisher, 1967). Upper John Day rocks east of Spray in Haystack Valley (Haystack Valley Member [revised] and Balm Creek Member) west of the Richmond fault differ from upper John Day rocks east of the fault in the vicinity of Kimberly (Johnson Canyon Member superposed on Kimberly Member). In both areas and at Sutton Mountain, however, the Rose Creek Member unconformably overlies and incises all of these upper John Day units and is the terminal member of the John Day Formation in the region. fault blocks in the vicinity of Kimberly con- 1967: ®g. 2) that parallel the trend of the serve the principal outcrops of the Johnson Blue Mountain axis. Lesser anticlinal and Canyon Member in its type area, this west- synclinal folds of similar trend occur imme- east trend apparently coinciding with Fisher's diately north of the Richmond fault and (1967) Monument syncline. Down-faulting south of the Kahler syncline in the Haystack of the Mitchell syncline along the trend of Valley area (Fisher, 1967: ®g. 5). Although the Richmond fault also appears to be re- he did not name it, the Balm Creek syncline sponsible for the preservation of uppermost belongs to this group of lesser folds and was John Day beds at Sutton Mountain (Gordon, mapped by Fisher (1967: ®g. 5; Fisher and 1988, and personal obs.). Rensberger, 1972: Map 3, secs. 26±28, T8S, HAYSTACK VALLEYÐBALM CREEK: Fisher R25E). (1967) mapped the structural framework of The Balm Creek and Rose Creek Mem- the Haystack Valley area north of the Rich- bers, the uppermost units of the John Day mond fault. Haystack Valley lies on the south Formation identi®ed in the Haystack Valley margin of a structural province situated be- area, are preserved along the trend of the tween the Blue Mountains on the north and Balm Creek syncline. Southeast of the syn- the nearly parallel trend of the Richmond clinal axis these members have been largely fault on the south (®g. 19). In this area John removed by erosion (®g. 20). However, in Day sediments were warped into broad this same area the revised Haystack Valley southwest-northeast trending synclines Member is well exposed and can be traced (Mitchell and Kahler synclines of Fisher, from the syncline to the south and east where 54 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 20. Schematic cross-section from west to east across the valleys of Haystack Creek and Balm Creek, illustrating the interpreted structural relationships of fault blocks to the local stratigraphic sequence. Abbreviations: Tj, John Day Formation; Tjt, Turtle Cove Member; Tjk, Kimberly Member; Tjhv, Haystack Valley Member (revised); Tjbc, Balm Creek Member; PGBS, Picture Gorge Basalt Subgroup. it is truncated by the projected trace of the Section, ®g. 8) in proximity to the fault Richmond fault. These exposures extending shows that the upper part of the Balm Creek from the syncline to the fault form a rugged Member has been upwarped and removed by terrain of ridges and valleys that produced erosion, and the unconformity is overlain by the greater part of the mammalian fauna from welded-tuff conglomerate and tuffaceous the member. They are broken by a prominent units of the Rose Creek Member. Similar southwest-northeast trending reverse fault Rose Creek incision into the Balm Creek (not indicated on Fisher's maps) with vertical Member also occurred along the axis of the separation of ϳ20±50 ft that parallels the syncline but there the upper part of the Balm trend of the syncline. It extends from the Creek Member was preserved (Beardog Sec- NE¼, sec. 33 through the N½,NW¼, sec. 34, tion: Middle-Upper, ®gs. 5, 6). to the S½,SW¼,SE¼, sec. 27, T8S, R25E Northwest of the Balm Creek syncline where it disappears beneath the Columbia along the ¯oor of Haystack Valley itself, up- basalts, which are not faulted. Regional permost John Day units have been removed northwest-southeast compression produced by erosion and only the lower part of the the synclinal folding and faulted the south formation (®g. 20, Tjt, Turtle Cove Member) limb of the syncline prior to eruption of the is present (Fisher and Rensberger, 1972: Map basalts, warping and offsetting rocks of the 3). These outcrops are fossiliferous and pro- revised Haystack Valley Member and Balm duce a fauna that includes the large oreodont Creek Member immediately northwest of the Promerycochoerus, Miohippus, hypertragu- Richmond fault. lids, less advanced entoptychines, and the The Rose Creek Member apparently lies temnocyonine Mammacyon, indicating an with angular unconformity on the Balm age more typical of the middle of the for- Creek Member along both the synclinal axis mation above the Picture Gorge ignimbrite. and to the south of the axis adjacent to the On the west wall of Haystack Valley, cliff- Richmond fault (in ®g. 20 the Rose Creek forming exposures of upper John Day units Member does not occur along the cross-sec- are down-faulted against these older John tional transect). A measured section (HS-1 Day rocks. The west wall preserves a section 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 55 that includes the revised Haystack Valley into the Rudio Creek drainage where John Member and in places the lower part of the Day exposures along the creek in secs. 25, Balm Creek Member (®g. 20). These out- 26, and 36, T9S, R26E are also down-faulted crops of uppermost John Day rocks can be to the north along similar trends, preserving traced westward to the village of Spray a thick accumulation of Kimberly Member where prominent exposures occur north of sediments near the mouth of Rudio Creek. the town. Measured sections on the west wall of the JOHN DAY VALLEY SOUTH OF KIMBERLY: John Day valley south of Kimberly (®gs. 13, Southeast of the Richmond fault the regional 15) indicate signi®cant down-faulting of the structural pattern (®g. 19) is characterized by Johnson Canyon Member and the subjacent a more westerly to easterly grain of anticlines Kimberly Member, beginning at Dead Cow and fault trends (Fisher, 1967: ®g. 2Ðthe Gulch in the N½,NE¼, sec. 12, T10S, R25E, John Day, Middle Mountain, and Hamilton and continuing northward to Johnson Can- faults; Richmond and Black Butte anti- yon (W½,NW¼, sec. 31, T9S, R26E). The clines). A structural pro®le based on Fisher's Kimberly Member appears to be downfaulted (1967: ®g. 5) transect from Middle Mountain on the north at least 200 ft (60 m) at Dead on the south to Kahler Basin on the north Cow Gulch (®g. 13), the southernmost fault passes through the John Day valley south of we have identi®ed along the west wall. Im- Kimberly and includes the uppermost John mediately north of the fault the contact of the Day rocks of this report (®g. 21). Fisher's Johnson Canyon Member on Kimberly gray pro®le shows John Day sediments ®lling a tuffs is present in Dead Cow Gulch at UCMP shallow basin ϳ25 mi (ϳ40 km) in north- locality V-6668, and can be traced 0.8 mile south extent. The John Day strata south of (1.25 km) to the north at approximately the Kimberly are gently downwarped according same elevation into the Big Valley section. If to Fisher's interpretation; faulting is limited the Johnson Canyon Member is then fol- to the trace of the Richmond fault and is not lowed farther north along the west wall, the identi®ed south of Kimberly. Somewhat member is apparently down-faulted Ͼ400 ft more pronounced folding is shown northwest (Ͼ120 m) between the Big Valley section of the Richmond fault where, at the scale of and Johnson Canyon (®g. 13). This suggests the pro®le, John Day strata have been to us that the considerable thickness of the warped into southwest-northeast trending Johnson Canyon Member in the vicinity of synclines and anticlines. Johnson Canyon and Kimberly post of®ce is Fisher's (1967) structural interpretation due to preservation of the member in a half- can be augmented by our ®eld observations graben or graben setting. that identi®ed extensive faulting south of Measured sections along the east wall of Kimberly along both east and west walls of the John Day valley indicate a similar struc- the John Day valley. The location of the tural setting. Although there is some varia- faults is important in interpreting the litho- tion in the elevation of the lower contact of stratigraphy of the formation south of Kim- the Rose Creek Member from its southern- berly. From Johnson Canyon on the north to most outcrops at McCarty Creek (Foree) to Rose Creek on the south, a distance of ϳ3 Rose Creek on the north, the unit lies at a mi (ϳ4.8 km), faulting is evidenced by ver- consistently high topographic elevation along tical offset of marker units in measured sec- the east wall (®g. 10). At Rose Creek, how- tions, by the presence of springs in the valley ever, the section is abruptly down-faulted walls, and by basalt-®lled valleys approxi- with nearly 200 ft (60 m) of vertical offset mately coincident with suspected fault (Rose Creek fault, ®g. 9). One mile (1.6 km) trends. Faults are normal, down to the north, farther north at Spring Creek (®g. 9, section with vertical offsets from ϳ40 to Ͼ400 ft 7), the Rose Creek Member has been down- (ϳ12 to Ͼ120 m). Fault trends are dif®cult dropped an additional 160 ft (48 m). Beyond to determine in several locations because of Spring Creek to the north, basalt ¯ows de- the covering basalts, but can be limited to scend to low elevations along the east wall northwest-southeast to west-east alignments. and the Rose Creek Member is absent; if out- These faults apparently continue to the east crops of the Rose Creek Member were once 56 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 Picture Gorge Tji, John Day Formation; Tj, 3 mi south of the Blue Mountains uplift (based on Fisher, ϳ 40 km). Abbreviations: ϳ 25 mi ( Picture Gorge Basalt Subgroup. ϳ PGBS, pre-Tertiary rocks; pT, Clarno Formation; Tc, Fig. 21. Schematic cross-section of the John Day basin from Middle Mountain to 1967: ®g. 5). NoteCanyon, the and thick Rose upperBalm Creek John Creek Day Members), and strata and Rose (i.e., Creek to above Members). Tji) the Length near north of the of cross-section: center the of the Richmond transect fault in in the vicinity the of Balm Kimberly Creek (Kimberly, Johnson syncline (Haystack Valley Member [revised], ignimbrite; 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 57 present, they were removed by pre-basalt Baksi (1974) averaged the ages after com- erosion. bining these samples and identi®ed an age The down-to-the-north offset of John Day range of 14.7 Ϯ 0.2 to 15.9 Ϯ 0.3 Ma. Baksi lithostratigraphic units on both walls of the (1989) later adjusted these dates, arriving at John Day valley south of Kimberly suggests an age range from 15.1 to 16.3 Ma, in an that uppermost John Day rocks were pre- attempt to account for loss of radiogenic 40Ar served from erosion on down-faulted blocks due to inclusion of ®ne-grained basaltic ma- within a graben or half-graben system (®g. terial in the samples. It is conceivable that 22) that extends roughly parallel to Fisher's the earliest PGBS ¯ows, which do not occur (1967: ®g. 2) Monument syncline. The gra- at Picture Gorge, may be much older. ben includes Kimberly Member and Johnson The oldest ¯ows of the PGBS are assigned Canyon Member outcrops (1) along the John to the Twickenham Basalt made up of three Day river northwest of Kimberly as far as members, from oldest to youngest: the Don- Bologna Creek, with exposures possibly ex- nely Basin, Bologna Creek, and Muleshoe tending northwest to the Richmond fault; (2) Creek Members (Bailey, 1989). The thick- north of Kimberly to the basalt rim of the ness of the oldest ¯ows in the study area John Day valley; and (3) east and northeast varies considerably in conformity with their an unknown distance, probably as far as Por- eruption on a John Day surface of high relief. tuguese Canyon (Fisher and Rensberger, Bailey's (1989) study suggests that the 1972: Map 3, NE corner, T9S, R26E), and Twickenham Basalt ¯ows emerging from the including the down-faulted exposures at the Monument dike ®ssure system migrated mouth of Rudio Creek. Pronounced erosional along the synclinal/graben topographic lows, relief on the John Day Formation occurs west to Donnely Basin and Alder Mountain, along this graben system associated with the eastward to Monument Mountain and Zion Monument syncline, and Fisher (1967: 121) Scope, and as far south of Kimberly as realized that this relationship suggested that Holmes Creek. This ¯ow pattern closely cor- local drainages ®rst developed along these responds to the trend of the Mitchell-Monu- topographic lows in post-John Day, pre-Co- ment synclines of Fisher (1967: ®gs. 2, 4) lumbia basalt time. and his 2000-ft contour drawn at the base of ERUPTION OF COLUMBIA FLOOD BASALTS: the PGBS. Shortly after the deposition and extensional Evidently, the earliest PGBS ¯ows of the faulting of the early Hemingfordian Rose Twickenham Basalt were following the struc- Creek Member, the ®rst ¯ows of the Colum- turally controlled, topographically low syn- bia ¯ood basalts in the region erupted from clinal and/or graben structures that today cor- the Monument dike ®ssures between Kim- respond closely to the course of the John Day berly, Monument, and Courtrock, and from River from the vicinity of Monument west- ®ssures in the vicinity of Picture Gorge. ward to Kimberly, Haystack Valley, and These ¯ows belong to the Picture Gorge Ba- southwest to Donnely Basin, Twickenham, salt Subgroup (PGBS) of the Columbia River and Sutton Mountain. Of particular interest Basalt recently discussed by Bailey (1989). in support of this view is that south of Kim- The ages of the oldest PGBS ¯ows are un- berly the ®rst ¯ows of the Twickenham Ba- known since the available radiometric dates salt extended to the south only as far as the are from ¯ows at the type section in Picture vicinity of Holmes Creek (®g. 2). These early Gorge where the lower third of the PGBS is ¯ows do not extend southward beyond the absent (Bailey, 1989). Initial potassium-ar- Rose Creek fault system because they evi- gon ages on ¯ows at the type section ranged dently abut against the fault scarps that ex- from 14.4 to 16.1 Ma (Baksi, 1974: 432, isted at the conclusion of John Day deposi- based on basalt crushed to coarse to very tion south of Kimberly. For example, in sec. coarse sand size for argon extraction); more 12, T10S, R25E, PGBS ¯ows abut the Dead ®nely crushed samples yielded ages from Cow Gulch fault, demonstrating the exis- 14.4 to 15.9 Ma but unfortunately these ages tence of the fault scarp at the time of eruption obtained from 14 ¯ows did not correspond of these early ¯ows (®gs. 13, 14). Bailey's to the stratigraphic succession. Watkins and (1989) data can be interpreted to suggest that 58 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 22. Schematic diagram showing the hypothetical development (A±F) of an early Miocene half- graben or graben system in the vicinity of and south of Kimberly post of®ce, John Day valley, Oregon. Abbreviations: k, Kimberly Member; jc, Johnson Canyon Member; rc, Rose Creek Member; twb, Twickenham Basalt; DCGF, Dead Cow Gulch fault; RCF, Rose Creek fault. 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 59 in time younger ¯ows of the Monument structural and topographic high, and only lat- Mountain Basalt (above the Twickenham Ba- er did much younger ¯ows of the PGBS salt) overtopped the faults and extended eventually onlap this trend (Bailey, 1989: ®g. south to Middle Mountain. Over time the ir- 7, Branson Creek section). South of the Rich- regular John Day topography was ®lled by mond anticline a separate ®ssure system later ¯ood-basalt to create a nearly level ter- erupted PGBS ¯ows in the Picture Gorge rain: ¯ows of the Monument Mountain Ba- area. The Richmond anticline-Middle Moun- salt are of considerable areal extent and by tain high appears to have separated the ¯ows this time indicate low topographic relief of the Monument dike system from those of throughout the basin (Bailey, 1989: 79). the Picture Gorge system to the south, ``sub- Bailey (1989) remarked on the unusual stantiated by the many ¯ows at Picture Gorge thickness of ¯ows of the Twickenham Basalt that cannot be correlated with the thicker se- near their type locality along the John Day quences in the central portion of the basin'' River, and his sections show that the earliest (Bailey, 1989: 82). ¯ows coincide with the present course of the river along the North Fork from Lower Cam- DISCUSSION AND CONCLUSIONS as Creek to Monument and Kimberly, thence westward to Bologna Creek, Haystack Val- Lithostratigraphic mapping of the upper ley, Donnely Basin, and Twickenham. It be- John Day beds by Fisher (1967) and Fisher comes evident that the present-day North and Rensberger (1972) provided a useful Fork of the John Day River and its contin- framework for additional studies of these uation west to Twickenham follow this an- rocks. Our examination of upper John Day cient structural trend occupied by the earliest rocks above the Kimberly Member in Hay- PGBS ¯ows. This could only occur if at least stack Valley and south of Kimberly suggests some degree of moderate warping and/or that the Haystack Valley Member can be di- faulting of the PGBS has continued along vided into four lithostratigraphic units that this structural trend. convey the history of sedimentation, tectonic Further evidence in support of this line of activity, and mammalian faunal change from reasoning is found in the structural setting of ϳ24 to ϳ18 Ma. We propose that these units the PGBS from Kimberly south along the be informally designated as ``members'' John Day River to Picture Gorge. Fisher pending acceptance by other investigators fa- (1967: ®g. 4) indicated that the modern miliar with the geology of the John Day re- course of the river followed a north-south gion. linear topographic low ®lled by PGBS ¯ows We believe that revision of the Haystack from Kimberly to Middle Mountain, and sev- Valley Member5 contributes to the identi®- eral tectonic maps of this area indicate the cation of genetic, unconformity-bounded presence at this location of a north-south lithostratigraphic units within upper John trending syncline (Walker, 1977; Brown and Day rocks in the Haystack Valley and Balm Thayer, 1966). This synclinal axis does not Creek drainages (North American Strati- extend south beyond Middle Mountain, cut graphic Code, 1983: article 23d, e). Our in- off by the east-west trend of the Richmond tent is to restrict the member in its type area anticline (Fisher, 1967). That the Richmond to the lowermost of three superposed upper anticline was elevated and eroded toward the John Day rock units that are best exposed in end of John Day time is indicated by the an- the southern limb of the Balm Creek syncline gular unconformity of Rose Creek Member (®g. 23; ®gs. 4±5, 8). Upper John Day beds gravels on lower John Day rocks and the of the syncline are preserved in a southwest- close stratigraphic approach of the Rose northeast trending ridge held up by Columbia Creek Member to the Picture Gorge marker basalt that separates Balm Creek from Hay- ignimbrite north of Middle Mountain (®g. stack Creek. Prominent exposures of the re- 10, measured section 14). Thus, at the time vised member occur along the southern limb that the earliest Twickenham ¯ows erupted of the syncline and extend to the southeast from the Monument dike ®ssures, the Middle where they are truncated by the Richmond Mountain-Richmond anticline area was a fault of Fisher (1967). These outcrops occu- 60 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 23. Stratigraphic nomenclature applied to the upper John Day Formation in the Balm Creek syncline east of Haystack Valley: a, Fisher and Rensberger's section 10 (equates to the Asher Section of this report); b, upper John Day strata of the Balm Creek syncline as interpreted in this study. 1 Later assigned by Rensberger (1983) to the Haystack Valley Member. py the NE¼, sec. 33; NW¼, sec. 34; S½, sec. tinue to include local, monomictic welded 27; and SE¼, sec. 28, T8S, R25E, Kimberly tuff-bearing conglomerates in the lower part 7.5-minute topographic quadrangle. Addi- of the Balm Creek section in the syncline and tional exposures form cliffs along the west along the west and east walls of Haystack and east walls of Haystack Valley where they Valley in agreement with the initial de®nition are in fault contact with lower John Day of the member by Fisher and Rensberger rocks of the Turtle Cove Member that form (1972: 17). But in our revision the base of the ¯oor of Haystack Valley (®g. 20). The the member in the Haystack Valley area is strata here allocated to the revised member not identi®ed by a particular welded tuff con- include Units 1 and 2 of our Balm Creek glomerate because these conglomerate lenses measured sections. The upper boundary of occur more or less at random within the the member is placed at a disconformity ``ribbed tuff'' lithology which characterizes above the prominent paleosol developed on the revised rock unit exposed along Balm the gray massive airfall tuff (GMAT) of Unit Creek. 2 which is a mappable marker bed in the The lowest ϳ160 ft of our revised Hay- Haystack and Balm Creek valleys. The Hay- stack Valley Member in the Balm Creek stack Valley Member as revised would con- drainage comprise greenish-gray to olive- 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 61 gray zeolitized tuffaceous rocks, chie¯y clay- The stratigraphically highest unit of the stones and siltstones with some scattered ¯u- John Day Formation is best exposed along vial sandstone lenses; the Entoptychus indi- the east wall of the John Day valley south of videns zone of Rensberger (1971) corre- Kimberly. The unit is named the Rose Creek sponds to the lowest ϳ40±50 ft of the unit. Member for the central location of outcrops He considered this large, robust entoptychine at the head of Rose Creek relative to other rodent the most advanced species of the ge- outcrops to the north and south. The member nus, arguably more evolved than most spe- lies with angular unconformity on subjacent cies of Entoptychus found in the Kimberly rocks of the John Day Formation along the Member. This rodent and other mammals as- east wall where the lower contact of the sociated with it in the lowest ϳ160 ft of our member is evident at most outcrops (®gs. 9, revised member represent a younger fauna 10, 24). To the south at McCarty Creek (Fo- than that found in similar greenish zeolitized ree) the Rose Creek Member lies on the rocks of the Turtle Cove Member (Fisher and eroded surface of green zeolitized claystones Rensberger, 1972: Map 3, map symbol Tjt in of the Turtle Cove Member (®g. 24c). Farther secs. 21, 22, 28, T8S, R25E) forming the north the member rests on gray tuffs of the ¯oor of Haystack Valley to the west (®g. 20, Kimberly Member (®g. 24d). The basal Rose Tjt, sec. 29, T8S, R25E). Although the lower Creek strata comprise stream channel depos- contact of the revised Haystack Valley Mem- its incised into underlying John Day rocks ber with subjacent John Day rocks has not and typically include coarse pebble-to-cob- been observed in the Balm Creek area, the ble, polymictic welded tuff conglomerates, presence of a mammal fauna (table 4) from cross-strati®ed ¯uvial sandstones, and debris the E. individens biozone along Balm Creek, ¯ows. Ashfall events are indicated at various temporally younger than the fauna from levels in the member by coarse obsidian- greenish zeolitized rocks (Tjt, Turtle Cove shard tuffs. The upper part of the member is Member) forming the ¯oor of Haystack Val- commonly made up of ®ne-grained air-fall ley, con®rms that the revised Haystack Val- and ¯uvially reworked olive-gray to yellow- ley Member is younger. The two units are gray tuff overprinted by silica-cemented pa- mostly in fault contact along the east and leosols. Calcareous cementation is rare to west walls of Haystack Valley (®g. 20), how- nonexistent in the paleosols and throughout ever we believe that a superpositional rela- the unit. tionship is preserved in sec. 21, T8S, R25E A lithofacies association typical of most (see addendum). outcrops includes polymictic welded-tuff Rocks in the Balm Creek syncline that oc- conglomerate, debris ¯ows, coarse obsidian- cur above the revised Haystack Valley Mem- shard tuffs, and stacked siliceous paleosols ber and below the Rose Creek Member are developed on olive-gray ®ne-grained tuffa- herein named the Balm Creek Member. The ceous claystone. Debris ¯ows are invariably unit (®g. 23; ®gs. 4±6, 8) is composed of composed of ®ne-grained gray tuff with ma- ¯uvially reworked yellowish-gray tuffaceous trix-supported, well-rounded pebbles and siltstones, sandstones, and claystones in its cobbles, many of them welded tuff. These lower part that are interbedded with overly- ¯ows are often overprinted by paleosols. In ing thinly bedded lacustrine tuff. These sed- addition to particularly common densely iments are overlain by a series of ®ning-up- welded tuff clasts with dark pumice ®amme, ward sequences of tuffaceous dark gray ¯u- the conglomerates also include pebbles and vial sandstones, interbedded with rather thick cobbles of other partially to densely welded gray massive airfall tuffs. The age of the unit rhyolitic/dacitic volcanics derived from the is uncertain due to the scarcity of fossil Picture Gorge ignimbrite, including gray la- mammals, but the rocks show no compelling pilli tuff and coarse obsidian shard tuff with lithologic resemblance to the Johnson Can- buff pumice lapilli (table 1). Also present are yon Member, particularly the lacustrine beds clasts of the Deep Creek tuff. The conglom- which are unique to the Balm Creek area. A erates from the member along the east wall fossil peccary from the base of the unit sug- of the John Day valley are among the coars- gests a late Arikareean age. est known from the eastern facies of the John 62 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Fig. 24. Stratigraphic nomenclature applied to the John Day Formation south of Kimberly, Oregon: a, Fisher and Rensberger (1972); b±d, this report. Abbreviations: tc, Turtle Cove Member; k, Kimberly Member; jc, Johnson Canyon Member; rc, Rose Creek Member.

Day Formation, having mean intermediate Only a single exposure of the Rose Creek clast diameters consistently Ͼ7 cm, with 12.9 Member has survived along the west wall of cm as the average mean intermediate diam- the John Day valley south of Kimberly (®g. eter from 12 localities (table 1). Most Rose 24b). At Dead Cow Gulch a terminal paleo- Creek Member conglomerate clasts are de- sol developed at high elevation on the Kim- rived from the Picture Gorge ignimbrite, berly Member is overlain by olive-gray tuff- which was exposed to erosion by the defor- aceous claystones, coarse obsidian-shard mation of John Day rocks prior to deposition tuff, debris ¯ows, and siliceous paleosols of of the member, or are indurated vitric tuff the Rose Creek Member (®g. 13, section 15). clasts derived from the subjacent Kimberly Welded tuff conglomerate is absent from this and Turtle Cove Members. section and there is no evidence of stream 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 63 incision by a Rose Creek channel. Here the Member is evident where the southernmost basal unit of the Rose Creek Member is ol- exposures of the unit occur along the west ive-gray tuff deposited directly on the ter- wall in the SE¼,NE¼, sec. 1 and in the minal Kimberly paleosol. John Day rocks NE¼,NE¼, sec. 12, T10S, R25E, Mt. Mis- have been eroded to the north and south of ery 7.5-minute quadrangle. Here the member this outcrop and the resulting topographic is incised into the gray Kimberly Member lows ®lled by ¯ows of the Columbia River tuffs with a relief of at least 50 ft (15.2 m). Basalt Group, suggesting that here the Rose Fine-grained yellow-gray tuffaceous silt- Creek Member was preserved on an isolated stones and sandstones reworked by ¯uvial hill or ridge. The absence of conglomerate processes ®ll these narrow valleys. Only in- and ¯uvial sandstones at this outcrop sug- traformational monomictic conglomerates gests that the west wall exposures at Dead made up of locally derived, rounded gray vi- Cow Gulch represent an interchannel loca- tric tuff clasts occur in the channel in®lls; tion relative to the east wall outcrops of the welded tuff pebbles or other exotic clasts are Rose Creek Member that sample an alluvial absent. Average clast size is much smaller facies. than in the Rose Creek conglomerates (table The association of polymictic welded tuff 1). Clasts from the coarsest conglomerate conglomerate, debris ¯ows, coarse obsidian- lens identi®ed in the Johnson Canyon Mem- shard tuffs, and stacked siliceous paleosols, ber attain only 6.2 cm average intermediate not only along the east wall of the John Day diameter, and no clasts in any Johnson Can- valley south of Kimberly but also at the top yon conglomerate lens exceed 9±10 cm. of the John Day Formation at Balm Creek The Johnson Canyon Member is down- and at Sutton Mountain, demonstrates the faulted to the north along the west wall of widespread distribution of this lithostrati- the valley (®gs. 22, 24b) so that the lower graphic unit. It records the ®nal phase of contact of the member is not visible in John- John Day deposition in the eastern facies of son Canyon. These faults apparently trend the formation before the eruption of the Co- approximately west-east and are situated lumbia ¯ood-basalts. The early Hemingfor- along the southern boundary of the SW¼, dian age of the Rose Creek Member is es- sec. 31, T9S, R26E, and in the N½,NE¼, tablished by the mammal fauna found in out- sec. 12, T10S, R25E. Offset of strata to the crops along the east wall of the John Day north is documented by the white marker tuff valley south of Kimberly (table 4). Other (ATR tuff, ϳ22.6 Ma) at Johnson Canyon: outcrops of the unit we have examined at the tuff occurs high in the exposures at Balm Creek and Sutton Mountain have not UCMP locality Picture Gorge 1 and is pro- produced fossil mammals. gressively downfaulted so that 0.5 mi (0.8 The beds overlying the Kimberly Member km) to the north, it occurs 60 ft (18 m) lower along the west wall of the John Day valley at the base of the exposures at the mouth of for 2.5 mi (4 km) south of Kimberly are des- the canyon. ignated the Johnson Canyon Member. This The cliff-forming outcrops on the north unit (®g. 24b; ®gs. 13, 15) is based on the wall of Johnson Canyon at its mouth are the outstanding exposures at the mouth of John- thickest exposures of the member that we son Canyon and on correlated outcrops to the have measured (®g. 16). In Johnson Canyon south along the west wall of the valley. Ad- the member can be divided into a lower and ditional exposures of the Johnson Canyon upper part (®g. 15). The lower unit includes Member are also found 1 mi (1.6 km) north- predominantly yellow-gray ¯uvially re- west of the Kimberly post of®ce on the Bro- worked tuffaceous siltstones and sandstones ken Hammer ranch, and 2.5 mi (4 km) north- with interbedded lenses of intraformational west along the John Day River at Bologna monomictic vitric tuff conglomerate. Calcar- Creek. Basalt-®lled valleys now separate eous cementation is present throughout the these once laterally contiguous outcrops lower unit and is particularly characteristic of where late John Day erosion created a deeply the paleosols. The upper unit is separated dissected topography. from the lower unit by a disconformity with- The lower contact of the Johnson Canyon out evidence of erosion; this surface is 64 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 marked by a highly developed siliceous pa- stead, they closely match the obsidian-shard lesol horizon suggesting a prolonged tem- air-fall tuff in the upper part of the Johnson poral hiatus. The upper unit above the dis- Canyon Member. Rose Creek streams must conformity is composed of thick gray airfall have incised both the ignimbrite and air-fall tuff units overlain by a series of obsidian- obsidian-shard tuff units of the Johnson Can- shard tuffs, each terminating in a paleosol yon Member, and locally transported these horizon. All paleosols of the upper unit are clasts in Rose Creek drainages south of Kim- siliceous. The upper unit concludes with yel- berly (table 1, both types occur together in low-gray tuffaceous sediments overlain by the basal conglomerate of the member at Columbia basalt ¯ows. Rose Creek North-2560 ft). Along the walls of the John Day valley Because no fossil mammals are known south of Kimberly we have not observed the from the Johnson Canyon Member's upper Rose Creek Member in superposition on the unit, its exact age remains in doubt. The Johnson Canyon Member. The closest geo- abundant fauna from the lower unit (table 4) graphic approach of the two units is in Dead and the dated white marker tuff allow the age Cow Gulch on the west wall where a John- of the lower unit to be more reliably deter- son Canyon Member paleovalley is cut into mined as late to latest Arikareean (table 2), Kimberly tuff on the north side of the gulch, hence older than the early Hemingfordian and ϳ820 ft (250 m) to the south the topo- fauna of the Rose Creek Member. graphically high, solitary Rose Creek out- The ages of the four members recognized crop rests on a terminal Kimberly Member in this report that occur above the Kimberly paleosol without intervening Johnson Can- Member are of particular interest in estab- yon beds (®gs. 13, 14, 24b). The two out- lishing the history of tectonic activity, sedi- crops are situated on opposite sides of the mentation, and erosion immediately prior to Dead Cow Gulch fault. The lithologies and the eruption of the Columbia plateau basalts lithofacies of the two members on the north in the region. We note the following relevant and south sides of the fault are entirely dif- observations on the age of these units and on ferent, and thus the Rose Creek and Johnson their correlation with the established NAL- Canyon Members at this location are unlike- MA faunal sequence in the North American ly to be coeval facies. midcontinent: However, there is a similarity between the (1) The revised Haystack Valley Member Rose Creek Member outcrop at Dead Cow in Haystack Valley is dated by a reasonably Gulch and the upper part of the Johnson Can- abundant early late Arikareean mammal fau- yon Member in the canyon. Coarse air-fall na, and by a dated tuff (40Ar/39Ar) closely obsidian-shard tuffs are present in both sec- associated with the monomictic welded tuff- tions, and we noted earlier that such coarse bearing conglomerates mentioned by Fisher tuffs are common in the Rose Creek beds. and Rensberger in their original description Yet we doubt that these tuffs support a tem- of the unit. The two most reliable ages on poral correlation between the Rose Creek feldspar crystals (23.6 Ϯ 0.07, 23.8 Ϯ 0.06 Member and the upper part of the Johnson Ma, table 2) from the tuff indicate the mem- Canyon beds, based on the following obser- ber is of earliest Miocene age, in proximity vations. In the basal polymictic conglomer- to the Paleogene-Neogene boundary (ϳ23.8 ates of the Rose Creek Member on the east Ma, Berggren et al., 1995), and the fossil wall of the John Day valley, occasional mammals are in agreement with this. Some- rounded clasts of air-fall obsidian-shard tuff time after this, the member was zeolitized, are found intermixed with dark gray, welded then later folded and faulted during a time of obsidian-shard tuff and other pebbles derived regional northwest-southeast compression. from the Picture Gorge ignimbrite. These This compression produced the northeast- pale yellowish-gray, obsidian-shard air-fall southwest trending anticlinal and synclinal tuff clasts are porous and uncompacted, folds typical of the region southeast of the hence readily distinguished from the partially Blue Mountains such as the Mitchell and to densely welded, dark gray obsidian-shard Kahler synclines (Fisher, 1967: ®g. 2). Fold- tuff clasts derived from the ignimbrite. In- ing of the Balm Creek syncline and devel- 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 65 opment of the low-angle reverse faulting that the member then is thought to fall some- parallels the syncline probably occurred at where between these two dates. Hence the this time. The creation of the graben (or half- faulting that cuts these rocks occurred after graben) that Haystack Creek and Balm Creek 22.6 Ma, and more probably after 19.2 Ma. apparently occupy today must have occurred (4) We have not found Rose Creek Mem- during a later interval of extensional faulting, ber sediments south of Kimberly deposited resulting in preservation of the uppermost in topographic lows developed on the John- John Day units. We conclude then that the son Canyon beds as might be expected if regional compression that folded the rede- Rose Creek dissection occurred after the con- ®ned Haystack Valley Member took place af- clusion of Johnson Canyon Member deposi- ter ϳ23±24 Ma, but prior to the deposition tion. In fact, Columbia basalts ®ll an eroded of the Rose Creek Member which lies with topography cut in the Johnson Canyon Mem- apparent angular unconformity across the ber (Fisher and Rensberger, 1972: pl. 6). A older John Day rocks. Since the Rose Creek possible explanation for this is that the Rose Member can be reliably dated by its fauna Creek Member may represent a lateral facies from outcrops south of Kimberly at ϳ18.2± of the Johnson Canyon Member and hence 18.8 Ma, the northwest-southeast compres- is age-equivalent to all or part of it. We doubt sion and consequent folding must have taken this because the lower part of the Johnson place prior to that time in the early Miocene. Canyon Member is lithically distinct from (2) The Turtle Cove and Kimberly Mem- the Rose Creek beds, not only in the different bers along the east and west walls of the John clast content of their respective conglomer- Day valley south of Kimberly were tilted and ates, but because of the absence of coarse warped at some time prior to the deposition obsidian-shard tuffs and debris ¯ows in the of the Rose Creek Member which overlies Johnson Canyon lower unit. Also, fossil them with angular unconformity. The Rose mammals from the lower unit appear to be- Creek Member was not involved in the fold- long to an older assemblage than the fauna ing of the older John Day rocks. The member from the Rose Creek Member; the dated is dated by its early Hemingfordian mammal ATR tuff (ϳ22.6 Ma) in Johnson Canyon fauna at ϳ18.2±18.8 Ma. Thus if the defor- supports this. Unfortunately, the upper unit mation south of Kimberly and the folding in in Johnson Canyon has not yielded a mam- Haystack Valley are part of the same episode mal fauna and has not been radiometrically of regional compression, this took place be- dated, and thus its age is uncertain. However, tween ϳ24 and ϳ18.2 Ma in the early Mio- because obsidian-shard air-fall tuffs in the cene. upper part of the Johnson Canyon Member (3) The Johnson Canyon Member rests are the likely source for similar obsidian- with erosional disconformity on the Kimber- shard tuff clasts that occur in the basal con- ly Member south of Kimberly post of®ce glomerates of the Rose Creek Member, it along the west wall of the John Day valley. seems that Rose Creek streams incised and Bedding within the member is essentially hence postdate the upper part of the Johnson horizontal. There is no evidence that the Canyon Member. We suspect that the pro- member was tilted or faulted during the de- found interval of erosion that preceded de- formational event that warped the Kimberly position of the Rose Creek Member removed and Turtle Cove Members. Except for normal a signi®cant part of the upper John Day For- faulting that offsets the member south of mation from Middle Mountain north to Kimberly, it appears to be largely undis- Spring Creek and Dead Cow Gulch, and this turbed throughout its outcrop area. The lower may have included units of the Johnson Can- part of the Johnson Canyon Member contains yon beds. Rose Creek sediments were then the ATR tuff of Fremd et al. (1994) dated by deposited across the truncated erosion sur- the 40Ar/39Ar method at 22.6 Ϯ 0.13 Ma, and face on the older John Day units. the fauna from this part of the unit is of late We also have observed that at Sutton or latest Arikareean age, at least as old as Mountain (sec. 33, T10S, R21E, Sutton ϳ19.2 Ma (based on biocorrelation with Mountain 7.5-minute quadrangle) Rose Great Plains Arikareean faunas). The age of Creek beds are superposed on a thick unit of 66 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 yellow-gray tuffaceous siltstones, sand- viously reported (19±22 Ma, Robinson et al., stones, and monomictic intraformational con- 1984) were from the western facies of the glomerates that resembles the Johnson Can- formation in the Warm Springs area (Wood- yon Member south of Kimberly. We suggest burne and Robinson, 1977). A precise deter- that the Sutton Mountain section documents mination of biochronologic age for the youn- the superposition of the two members that is gest stratigraphic units within the classic absent south of Kimberly. eastern facies of the John Day Formation has (5) After deposition of the Rose Creek been lacking since the earliest investigations Member, pedogenesis preceded normal fault- of these rocks. This situation resulted from ing of the member south of Kimberly from inadvertently combining the mammal faunas Rose Creek northward to Spring Creek; this from the Rose Creek, Johnson Canyon, and faulting produced escarpments that are abut- revised Haystack Valley Members into a ted and preserved by the earliest PGBS ¯ows composite fauna that spanned more than ®ve (Twickenham Basalt). Younger basalt ¯ows million years within the early Miocene then completely buried the existing John Day (ϳ18.2±23.8 Ma). topography. Fossil mammals discovered during our The oldest ¯ows of the PGBS at the type ®eld studies from 1992 to 2001, supple- section at Picture Gorge are possibly as old mented by specimens in museum collections, as ϳ16 Ma (Baksi, 1989; Bailey, 1989); thus, establish a biochronologic age of ϳ18.2±18.8 there is reason to suggest that the stratigraph- Ma for the youngest fauna from the eastern ically lowest ¯ows of the PGBS (Twicken- facies of the John Day Formation. This fau- ham Basalt), which are not present in the Pic- na, herein designated the Picture Gorge 36 ture Gorge type section, are older than ϳ16 local fauna, comes from a laterally continu- Ma (e.g., Tolan et al., 1989: 17), perhaps as ous, mappable stratigraphic unit, the Rose old as ϳ17 Ma. If Rose Creek deposition Creek Member, located at high elevations ended by ϳ18 Ma, followed by erosion and immediately below the Columbia basalts on weathering of the resulting land surface, then the eastern margin of the John Day valley the terminal Rose Creek paleosols marking south of Kimberly. Outcrops of the Rose the end of John Day deposition in the region Creek Member that have produced the de®n- developed during this ϳ1 million-year inter- ing mammal assemblage are from two local- val, prior to eruption of PGBS ¯ows from ities in the west half of sec. 17, T10S, R26E. the Monument feeder dikes. The ®rst is along Rose Creek in the NW¼, (6) The Balm Creek Member includes the NW¼,NW¼, sec. 17, and the second on ¯uvial and lacustrine volcaniclastics and air- Bone Creek in the NE¼,NW¼,SE¼,SW¼, fall tuffs that occur stratigraphically above sec. 17, continuing into the SW¼,SE¼, the revised Haystack Valley Member and be- NE¼,SW¼, sec. 17, T10S, R26E, Grant low the Rose Creek Member in the Haystack County, Oregon. Valley-Balm Creek area. If the Haystack Val- (8) Is there a signi®cant interval of time ley Member is dated at ϳ23±24 Ma, and the between the conclusion of Rose Creek Mem- Rose Creek Member is as old as ϳ18.2 Ma, ber deposition and the initiation of Columbia then the Balm Creek Member falls within ¯ood basalt ¯ows in the region? The oldest this interval in the early Miocene. Although lava ¯ows of the Columbia River Basalt it is possible that the Balm Creek Member, Group were believed to be the Imnaha ba- all or in part, is the same age as the Johnson salts of eastern Oregon and western Idaho Canyon Member, we lack de®nitive evidence dated at ϳ17.0±17.5 Ma (Reidel and Tolan, of this at present. 1992; Baksi, 1989; Hooper, 1982; McKee et (7) Mammalian faunas from the youngest al., 1981). More recent 40Ar/39Ar dating of stratigraphic units of the John Day Formation tholeiitic basalts in southeastern Oregon has have been involved in estimates of the identi®ed ¯ows of the Steens Basalt and the amount of time that elapsed between the ces- basalt of Malheur Gorge as the earliest erup- sation of John Day deposition in the region tions of the Columbia River ¯ood-basalt and the initiation of the Columbia lava ¯ows. province with ages from 16.6 Ϯ 0.02 Ma to The youngest John Day mammal faunas pre- 16.9 Ϯ 0.8 Ma (Swisher et al., 1990; Hooper 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 67 et al., 2002). Newer dates of ϳ15.4±15.5 Ma at ϳ23.5±23.8 Ma. Thus, zeolitization oc- have been obtained on Imnaha basalts, and curred after ϳ23.5 Ma and ended before re- the earlier dates at ϳ17.0±17.5 Ma are con- gional compression deformed the member, sidered less reliable (Hooper et al., 2002). which took place prior to deposition of the Flows of the Picture Gorge Basalt Sub- overlying Rose Creek beds at ϳ18.2±18.8 group were erupted from the Monument dike Ma. Zeolitization of John Day sediments in swarm (Bailey, 1989) situated in the area the region may have concluded by ϳ23 Ma south of Kimberly and as far east as the town if the ATR marker tuff in the unzeolitized of Monument (Fisher, 1967; Wilcox and Johnson Canyon Member is reliably dated at Fisher, 1966). Bailey (1989) proposed a sep- 22.6 Ma. arate eruptive center near Picture Gorge (10) The mammal fauna of the revised south of the Monument dike complex for the Haystack Valley Member is clearly more ar- ¯ows in Picture Gorge. The K/Ar-dated chaic in its taxonomic composition than the ¯ows of the PGBS that occur in Picture faunas from the Johnson Canyon and Rose Gorge are reported to be as old as ϳ16 Ma Creek Members. Miohippine horses, entop- (Bailey and Conrey, 1992; Baksi, 1989; Bai- tychine and allomyine rodents, and hypertra- ley, 1989). However, the oldest PGBS ¯ows gulids are conspicuous holdovers from older (Twickenham Basalt) probably predate the John Day faunas, whereas advanced taxa ¯ows at Picture Gorge and it is these ¯ows such as parahippine horses, Archaeohippus, that occur in proximity to Rose Creek Mem- dromomerycid and moschid cervoids, the ber sediments along the east wall of the John chalicothere Moropus, large beavers, the ro- Day valley south of Kimberly. Dating these dents Mylagaulodon, Sewelleladon, and ad- ¯ows would be of considerable interest. If vanced species of Schizodontomys, and the the Twickenham basalts are as old as 17 Ma, large index oreodont Merycochoerus are ab- and the Rose Creek sediments as young as sent. ϳ18 Ma, the interval in question is brief. (11) The upper John Day faunas of Incision of the Rose Creek Member into Oregon indicate more mesic and better veg- older John Day strata may record regional etated environments in the Paci®c Northwest elevation of the Richmond Anticline (Fisher, relative to the seasonally arid grassland 1967: ®g. 2) in early Hemingfordian time, plains of the North American midcontinent perhaps somehow related to magmatic in¯a- in the early Miocene. Almost all channel tion prior to eruption of the PGBS, resulting conglomerates in the revised Haystack Val- in northward tilting and local folding of the ley, Johnson Canyon, and Rose Creek Mem- John Day Formation. Northward tilting of bers contain fossil wood together with fossil John Day strata and PGBS ¯ows from Mid- mammal remains, whereas fossil wood of dle Mountain to Kimberly may have contin- any kind in Arikaree Group rocks of the ued during post-PGBS tectonism in the mid- Great Plains is extremely scarce. It is not un- to late Miocene. til the early Hemingfordian deposits of the (9) Recognition of the different ages of the Runningwater Formation that stream depos- revised Haystack Valley, Johnson Canyon, its carry more abundant fossil wood in the and Rose Creek Members clari®es the timing midcontinent. of zeolitization in the Haystack Valley Mem- (12) The Hemingfordian Land Mammal ber of Fisher and Rensberger (1972: 18). `àge'' is used here in a different sense from Zeolitization in the Rose Creek and Johnson that originally proposed by the Wood Com- Canyon Members is, if present at all, ex- mittee of the Paleontological Society (Wood tremely rare. It appears to be restricted to the et al., 1941) and followed by Fisher and revised Haystack Valley Member in the Hay- Rensberger (1972: ®g. 8). We place the Ari- stack Valley-Balm Creek area where the kareean-Hemingfordian boundary in the ear- member was zeolitized before the beds were ly Miocene at ϳ18.8 Ma (see MacFadden deformed and faulted. Beds of the revised and Hunt, 1998, for discussion), based upon Haystack Valley Member are the youngest recent reevaluation of the succession of John Day rocks to have undergone extensive mammalian faunas from Arikaree and Hem- zeolitization in the local area, and are dated ingford rocks of western Nebraska in the 68 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 type area of these ages. The following ob- the Upper Harrison beds was regarded by servations are relevant: some investigators as the basal or earliest The de®ning fauna of the Arikareean Hemingfordian fauna, notwithstanding its NALMA is the Agate local fauna found in clearly Arikareean aspect. This usage of the the base of the Upper Harrison beds in west- term Hemingfordian was consequently ap- ern Nebraska (Wood et al., 1941: 11±12; plied by Fisher and Rensberger (1972: ®g. 8) Hunt, 1990). Many of its component taxa to the identi®cation of John Day faunas; this ®rst occur in the subjacent Harrison Forma- was a reasonable approach, for this was action and continue into the Upper Harrison cepted practice at the time. beds, a formation-rank term now replaced by More recently it has been recognized that Anderson Ranch Formation (Hunt, 2002). the mammalian faunas of the Runningwater Thus, the mammals of the Harrison and Up- Formation of western Nebraska, which over- per Harrison beds constitute an integral fauna lies the Upper Harrison beds, represent a that is central to the de®nition of the Arika- readily identi®able biochronologic datum reean Land Mammal `àge''. Note that this within the early Miocene of the Great Plains fauna is not necessarily con®ned to the geo- (Tedford et al., 1987; MacFadden and Hunt, chron of the Harrison and Upper Harrison 1998; Woodburne and Swisher, 1995: 345). rock units, and is recognized as a biochron- Numerous genera of mammals appear for the ologic entity entirely on the basis of its com- ®rst time in these Runningwater faunas and ponent taxa. serve as biochronologic markers for the be- Despite the lithologic similarity of the Up- ginning of the Hemingfordian NALMA. per Harrison to the Harrison Formation and Also, a number of Arikareean genera contin- lower stratigraphic units of the Arikaree ue into the Runningwater Formation where Group (Hunt, 1990), the Upper Harrison distinctive early Hemingfordian congeneric beds were included as the basal unit of the species can be identi®ed by their more ad- overlying Hemingford Group under the term vanced stage of evolution. This faunal com- ``Marsland Formation'' (Lugn, 1939, and ref- plex continues in time and can be recognized erences therein)7. The Hemingfordian age in the mammals of the late Hemingfordian was implicitly considered as the geochron of Box Butte and Sheep Creek Formations of the Marsland and Sheep Creek Formations of western Nebraska. Thus, the sequential fau- the Hemingford Group by the Wood Com- nas from the Runningwater, Box Butte, and mittee (1941), and as a result the fauna of Sheep Creek Formations are currently regarded as representative of the Hemingfor- 7 The Marsland Formation, as used by Lugn (1939) dian age (Tedford et al., 1987). and Schultz (1938), combined the Upper Harrison beds Key indicator taxa marking the beginning (Peterson, 1909), an upper Arikaree rock unit primarily of the Hemingfordian in the Great Plains in- composed of volcaniclastic eolian ®ne-grained sand- clude the oreodonts Merycochoerus magnus stones, with an overlying rock unit comprising ¯uvial sands and granitic gravels marking the initiation of and an advanced form of Merychyus; the Hemingford Group deposition in western Nebraska. equid Parahippus pawniensis and a second They referred to the Upper Harrison beds as the ``lower larger parahippine, P. tyleri (the ®rst hypso- Marsland'' and the ¯uvial sands and gravels as the `ùp- dont species of Parahippus in the Great per Marsland'' beds. Cook (1965) recognized that the Upper Harrison beds were lithically more similar to Plains); the small moschid deer Machaerom- rocks of the Arikaree Group, and separated them from eryx; the ®rst merycodont antilocaprids; a the `ùpper Marsland'' sands and gravels, which he then large long-limbed daphoenine amphicyonid; named the Runningwater Formation. He regarded the an ancestral mylagaulid Mesogaulus panien- Runningwater Formation as evidence of a major tectonic sis and species of the large beavers Anchithe- episode to the west, resulting in stream transport of granitic sands and gravels from the Rocky Mountains onto riomys and Hystricops; a species of the rhi- the Great Plains. Because of the confusion surrounding noceros Menoceras somewhat evolved be- the term ``Marsland'' and the lithologic heterogeneity of yond M. arikarense from the Agate Quarries; the lower and upper rock units included in the ``forma- and various species of camelids (Michenia, tion'', its use is no longer recommended, and a new formation-rank term (Anderson Ranch Formation) has Protolabis, Oxydactylus longirostris) and been proposed to replace `Ùpper Harrison'' (Hunt, protoceratids (Syndyoceras). The absence of 2002). certain taxa is also signi®cant: these include 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 69 the oreodont Promerycochoerus; merychip- Smith, and Regan Dunn (John Day Fossil pine horses with cement-covered cheek teeth, Beds National Monument); J.A. Gauthier, and terminal Arikareean parahippines (P. ne- Lyndon Murray, and M.A. Turner (Yale Uni- brascensis, P. wyomingensis); felid and nim- versity); and David Taylor (Portland State ravid cats; the large mustelid Megalictis; and University). We also thank the National Park temnocyonine beardogs. Several small arc- Service staff at John Day Fossil Beds Na- toid carnivores also ®rst appear in North tional Monument who aided with logistics America in the ¯uvial channels of the Run- and made our work an enjoyable experience. ningwater Formation. The earliest Heming- We gratefully acknowledge the assistance of fordian fauna in the type area in western Ne- Bureau of Land Management archeologist braska is the Northeast of Agate local fauna. John Zancanella who provided us ®eld ac- Paleomagnetic calibration of rocks that yield- cess to BLM lands. ed this fauna suggests an age of ϳ18.2±18.8 The ®eld program could not have been ini- Ma (MacFadden and Hunt, 1998). It is with tiated, or completed, but for the interest and species in this local fauna that the mammals remarkable generosity of the region's ranch- of the Rose Creek Member of the John Day ers and landowners. Our appreciation goes Formation show the closest correspondence. out to the Leonard Breck family, and to Rob and Becky Williams of Kimberly, Oregon, ACKNOWLEDGMENTS for permission to study the geology of the Longview Ranch, John Day valley, Oregon, Our interest in the upper John Day For- and to the Asher, Foster, Hopper, Kintz, and mation was the result of an invitation from Withrow families for access to their lands in National Park Service paleontologist Ted the Haystack Valley-Balm Creek area. Mike Fremd to examine fossils and localities of Fingerhut shared his ®rsthand knowledge of amphicyonid carnivores in the Haystack Val- the John Day beds in Haystack Valley and ley-Kimberly-Monument area. A brief recon- Balm Creek, allowed us to study fossils col- naissance with Fremd in 1992 led to the be- lected by him in the area, and aided us in ginning of our ®eld studies on upper John numerous ways throughout our work. We Day units. Without his enthusiasm and logis- also appreciate the assistance and continued tical assistance, this study could not have interest of National Park Service paleontol- been completed. We also acknowledge the ogists Scott Foss, Ted Fremd, and preparator earlier work of J.C. Merriam, Richard Hay, Matt Smith at John Day Fossil Beds National R.V. Fisher, and John Rensberger. Their pub- Monument. lished reports formed the initial framework For production of illustrations we are in- for our observations on the upper John Day debted to Angie Fox, artist/illustrator for the beds in the Haystack Valley and Kimberly University of Nebraska State Museum. areas. We particularly bene®ted from thoughtful We are grateful to Dr. Wm. C. McIntosh reviews of the manuscript by L. Barry Al- and Lisa Peters of the New Mexico Bureau bright, Anthony Barnosky, and Richard H. of Mines and Mineral Resources, Socorro, Tedford. who generously attempted the dating of a number of tuffs and tuffaceous rock units REFERENCES from the John Day Formation. This study would not have been possible Bailey, D.G., and R.M. Conrey. 1992. Common without access to museum and university parent magma for Miocene to Holocene ma®c collections conserving John Day fossils: we volcanism in the northwestern United States. express our appreciation to W.A. Clemens, Geology 20: 1131±1134. Mark Goodwin, Anthony Barnosky, and Pat Bailey, M.M. 1989. Revisions to stratigraphic nomenclature of the Picture Gorge Basalt Sub- Holroyd (University of California-Berkeley); group, Columbia River Basalt Group. In S.P. R.H. Tedford, M.C. McKenna, and Jin Meng Reidel and P. R. Hooper (editors), Volcanism (American Museum of Natural History, New and tectonism in the Columbia River ¯ood-ba- York); William and Elizabeth Orr (University salt province. Geological Society of America of Oregon); Ted Fremd, Scott Foss, Matt Special Paper 239: 67±84. 70 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

Baksi, A.K. 1974. Isotopic fractionation of a Oregon. U.S. Geological Survey Professional loosely held atmospheric argon component in Paper 400-B: B302±B304. the Picture Gorge basalts. Earth and Planetary Fremd, T., E.A. Bestland, and G.J. Retallack. Science Letters 21: 431±438. 1994. John Day Basin Paleontology Field Trip Baksi, A.K. 1989. Reevaluation of the timing and Guide and Road Log: for 1994 Society of Ver- duration of extrusion of the Imnaha, Picture tebrate Paleontology annual meeting, Seattle, Gorge, and Grande Ronde Basalts, Columbia WA. Publication of John Day Fossil Beds Na- River Basalt Group. In S.P. Reidel and P.R. tional Monument 94-1, 62 pp. Hooper (editors), Volcanism and tectonism in Gordon, I.R. 1988. Stratigraphy and syntectonic the Columbia River ¯ood-basalt province. Geo- sedimentation in the John Day Formation, Sut- lological Society of America Special Paper ton Mountain, Wheeler County, Oregon. Hon- 239: 105±111. ors thesis, Bachelor of Science, Department of Berggren, W.A., D.V. Kent, C.C. Swisher, and Geology, Oregon State University, Corvallis, M.-P. Aubry. 1995. A revised Cenozoic geo- Oregon, 21 pp. chronology and chronostratigraphy. In W.A. Hay, R.L. 1962. Origin and diagenetic alteration Berggren, D.V. Kent, M.-P. Aubry, and J. Har- of the lower part of the John Day Formation denbol (editors), Geochronology, time scales near Mitchell, Oregon. In A.E.J. Engel, H.L. and global stratigraphic correlation. SEPM Spe- James, and B.F. Leonard (editors), Petrologic cial Publication 54: 129±212. studies: a volume in honor of A.F. Buddington: Brown, C.E., and T.P. Thayer. 1966. Geologic map 191±216. New York: Geological Society of of the Canyon City Quadrangle, northeastern America. Oregon. U.S. Geological Survey, Map I-447 Hay, R.L. 1963. Stratigraphy and zeolitic diagen- (Miscellaneous Geologic Investigations Series). esis of the John Day Formation of Oregon. Uni- Calkins, F.C. 1902. A contribution to the petrog- versity of California Publications in Geological raphy of the John Day Basin. University of Cal- Sciences 42(5): 199±262. ifornia Publications in Geological Sciences Hooper, P.R. 1982. The Columbia River basalts. 3(5): 109±172. Science 215(4539): 1463±1468. Cook, H.C. 1965. Runningwater Formation, mid- Hooper, P.R., and D.A. Swanson. 1990. The Co- dle Miocene of Nebraska. American Museum lumbia River Basalt Group and associated vol- Novitates 2227: 1±8. canic rocks of the Blue Mountains province. In Fisher, R.V. 1962. Clinoptilolite tuff from the John G.W. Walker (editor), Geology of the Blue Day Formation, eastern Oregon. Ore Bin 24: Mountains region of Oregon, Idaho, and Wash- 197±203. ington: Cenozoic geology of the Blue Moun- Fisher, R.V. 1964. Resurrected Oligocene hills, tains region. U. S. Geological Survey Profes- eastern Oregon. American Journal of Science sional Paper 1437: 63±99. 262: 713±725. Hooper, P.R., G.B. Binger, and K.R. Lees. 2002. Fisher, R.V. 1966a. Textural comparison of John Ages of the Steens and Columbia River ¯ood Day volcanic siltstone with loess and volcanic basalts and their relationship to extension-relat- ash. Journal of Sedimentary Petrology 36(3): ed calc-alkalic volcanism in eastern Oregon. 706±718. Bulletin of the Geological Society of America Fisher, R.V. 1966b. Geology of a Miocene ignim- 114: 43±50. brite layer, John Day Formation, eastern Hooper, P.R. and R.M. Conrey. 1989. A model for Oregon. University of California Publications the tectonic setting of the Columbia River ba- in Geological Sciences 67: 1±73. salt eruptions. In S.P. Reidel and P.R. Hooper Fisher, R.V. 1967. Early Tertiary deformation in (editors), Volcanism and tectonism in the Co- north-central Oregon. Bulletin of the American lumbia River ¯ood-basalt province. Geological Association of Petroleum Geologists 51(1): Society of America Special Paper 239: 293± 111±123. 306. Fisher, R.V. 1968. Pyrogenic mineral stability, Hunt, R.M., Jr. 1985. Faunal succession, lithofa- lower member of the John Day Formation, east- cies, and depositional environments in Arikaree ern Oregon. University of California Publica- rocks (lower Miocene) of the Hartville Table, tions in Geological Sciences 75: 1±39. Nebraska and Wyoming. In J.E. Martin (editor), Fisher, R.V., and J.M. Rensberger. 1972. Physical Fossiliferous Cenozoic deposits of western stratigraphy of the John Day Formation, central South Dakota and northwestern Nebraska. Dak- Oregon. University of California Publications oterra 2(2): 155±204. in Geological Sciences 101: 1±33. Hunt, R.M., Jr. 1990. Taphonomy and sedimen- Fisher, R.V., and R.E. Wilcox. 1960. The John tology of Arikaree (lower Miocene) ¯uvial, eo- Day Formation in the Monument Quadrangle, lian, and lacustrine paleoenvironments, Nebras- 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 71

ka and Wyoming: a paleobiota entombed in biotic interchange. Topics in Geobiology 4: ®ne-grained volcaniclastic rocks. Geological 219±247. Society of America Special Paper 244: 69±111. Peterson, O.A. 1909. A revision of the Entelo- Hunt, R.M., Jr. 2002. New amphicyonid carnivor- dontidae. Memoirs of the Carnegie Museum ans (Mammalia, Daphoeninae) from the early 4(3): 41±158. Miocene of southeastern Wyoming. American Pickford, M. 1986. Sedimentation and fossil pres- Museum Novitates 3385: 1±41. ervation in the Nyanza rift system, Kenya. In Hunt, R.M., Jr., X.X. Xue, and J. Kaufman. 1983. L.E. Frostick et al. (editors), Sedimentation in Miocene burrows of extinct beardogs: indica- the African rifts. Geological Society of London tion of early denning behavior of large mam- Special Publication 25: 345±362. malian carnivores. Science 221: 364±366. Reidel, S.P. and T.L. Tolan. 1992. Eruption and Izett, G.A., and J.D. Obradovich. 2001. 40Ar/39Ar emplacement of ¯ood basalt: an example from ages of Miocene tuffs in basin-®ll deposits the large-volume Tepee Butte Member, Colum- (Santa Fe Group, New Mexico, and Trouble- bia River Basalt Group. Bulletin of the Geolog- some Formation, Colorado) of the Rio Grande ical Society of America 104: 1650±1671. system. The Mountain Geologist 38(2): 77±86. Rensberger, J.M. 1971. Entoptychine pocket go- Le Conte, J. 1874. American Journal of Science phers (Mammalia, Geomyoidea) of the Early 7(ser. 3): 167. Miocene John Day Formation, Oregon. Uni- Lugn, A.L. 1939. Classi®cation of the Tertiary versity of California Publications in Geological system in Nebraska. Bulletin of the Geological Sciences 90: 1±163. Society of America 50: 1245±1276. Rensberger, J.M. 1973. Pleurolicine rodents (Geo- MacFadden, B.J., and R.M. Hunt, Jr. 1998. Mag- myoidea) of the John Day Formation, Oregon. netic polarity stratigraphy and correlation of the University of California Publications in Geo- Arikaree Group, Arikareean (late Oligocene- logical Sciences 102: 1±95. early Miocene) of northwestern Nebraska. In D. Rensberger, J.M. 1983. Successions of menisco- Terry, H. LaGarry, and R.M. Hunt, Jr. (editors), myine and allomyine rodents (Aplodontidae) in Depositional environments, lithostratigraphy, the Oligo-Miocene John Day Formation, and biostratigraphy of the White River and Ari- Oregon. University of California Publications karee Groups (Late Eocene to Early Miocene, in Geological Sciences 124: 1±157. North America). Geological Society of Ameri- Robinson, P.T., and G.F. Brem. 1981. Guide to ca Special Paper 325: 143±165. McKee, E.H., P.R. Hooper, and W.D. Kleck. 1981. geologic ®eld trip between Kimberly and Bend, Age of Imnaha BasaltÐoldest basalt ¯ows of Oregon, with emphasis on the John Day For- the Columbia River Basalt Group, northwest mation. In D. A. Johnston and J. Donnelly-No- United States. Isochron West 31: 31±33. lan (editors), Guides to some volcanic terranes Marsh, O.C. 1875. American Journal of Science in Washington, Idaho, Oregon, and Northern 9(ser. 3): 52. California. U.S. Geological Survey Circular Merriam, J.C. 1900. Classi®cation of the John 838: 29±40. Day beds. Science 11: 219±220. Robinson, P.T., G.F. Brem, and E.H. McKee. Merriam, J.C. 1901. A contribution to the geology 1984. John Day Formation of Oregon: a distal of the John Day Basin. University of California record of early Cascade volcanism. Geology Publications in Geological Sciences 2(9): 269± 12(4): 229±232. 314. Robinson, P.T., G.W. Walker, and E.H. McKee. Merriam, J.C., and W.J. Sinclair. 1907. Tertiary 1990. Eocene(?), Oligocene, and Lower Mio- faunas of the John Day region. University of cene rocks of the Blue Mountain Region. In G. California Publications in Geological Sciences W. Walker (editor), Geology of the Blue Moun- 5(11): 171±205. tains region of Oregon, Idaho, and Washington: Munthe, K. 1981. Skeletal morphology and func- Cenozoic geology of the Blue Mountains re- tion of the Miocene rodent Schizodontomys gion. U.S. Geological Survey Professional Pa- harkseni. Paleobios 35: 1±33. per 1437: 29±62. North American Stratigraphic Code. 1983. North Ross, C.S., and R.L. Smith. 1960. Ash-¯ow tuffs: American Commission on Stratigraphic No- their origin, geologic relations, and identi®ca- menclature. Bulletin of the American Associa- tion. U.S. Geological Survey Professional Pa- tion of Petroleum Geologists 67(5): 841±875. per 366: 1±81. Pascual, R., M. Vucetich, G. Scillato-Yane, and Schultz, C.B. 1938. The Miocene of western Ne- M. Bond. 1985. Main pathways of mammalian braska. American Journal of Science 35: 441± diversi®cation in South America. In F. Stehli 444. and S.D. Webb (editors), The great American Sinclair, W.J. 1901. The discovery of a new fossil 72 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282

tapir in Oregon. Journal of Geology 9(8): 702± Map I-902 (Miscellaneous Geologic Investiga- 707. tions Series). Sinclair, W.J. 1903. Mylagaulodon, a new rodent Walker, G.W., and P.T. Robinson. 1990a. Paleo- from the Upper John Day of Oregon. American cene(?), Eocene, and Oligocene(?) rocks of the Journal of Science 15: 143±144. Blue Mountains Region. In G.W. Walker (edi- Smith, R.L. 1960. Zones and zonal variations in tor), Geology of the Blue Mountains region of welded ash ¯ows. U.S. Geological Survey Pro- Oregon, Idaho, and Washington: Cenozoic ge- fessional Paper 354-F: 149±159. ology of the Blue Mountains region. U.S. Geo- Swisher, C.C., J.A. Ach, and W.K. Hart. 1990. logical Survey Professional Paper 1437: 13±27. Laser-fusion 40Ar/39Ar dating of the type Steens Walker, G.W., and P.T. Robinson. 1990b. Ceno- Mountain Basalt, south-eastern Oregon, and the zoic tectonism and volcanism of the Blue age of the Steens geomagnetic polarity transi- Mountains region. In G. W. Walker (editor), tion (abstract). Eos (Transactions of the Amer- Geology of the Blue Mountains region of ican Geophysical Union) 71: 1296. Oregon, Idaho, and Washington: Cenozoic ge- Tedford, R.H., M.F. Skinner, R.W. Fields, J.M. ology of the Blue Mountains region. U.S. Geo- Rensberger, D.P. Whistler, T. Galusha, B.E. logical Survey Professional Paper 1437: 119± Taylor, J.R. Macdonald, and S.D. Webb. 1987. 135. Faunal succession and biochronology of the Watkins, N.D., and A.K. Baksi. 1974. Magneto- Arikareean through Hemphillian interval (late stratigraphy and oroclinal folding of the Co- Oligocene through earliest Pliocene epochs) in lumbia River, Steens and Owyhee Basalts in North America. In M.O. Woodburne (editor), Oregon, Washington, and Idaho. American Cenozoic mammals of North America: 153± Journal of Science 274: 148±189. 210. Berkeley: University of California Press. Wilcox, R.E. and R.V. Fisher. 1966. Geologic map Tedford, R.H., J.B. Swinehart, C.C. Swisher, D.R. of the Monument Quadrangle, Grant County, Prothero, S.A. King, and T.E. Tierney. 1996. Oregon. U.S. Geological Survey, Map GQ-541. The Whitneyan-Arikareean Transition in the Wood, H.E., R.W. Chaney, J. Clark, E.H. Colbert, High Plains. In D.R. Prothero and R.J. Emry G.L. Jepsen, J.B. Reeside, Jr., and C. Stock. (editors), The terrestrial Eocene-Oligocene 1941. Nomenclature and correlation of the transition in North America: 312±334. New North American continental Tertiary. Bulletin York: Cambridge University Press. of the Geological Society of America 52: 1±48. Tolan, T.L., S.P. Reidel, M.H. Beeson, J.L. An- Woodburne, M.O., and P.T. Robinson. 1977. A derson, K.R. Fecht, and D.A. Swanson. 1989. new Late Hemingfordian mammal fauna from Revisions to the estimates of the areal extent the John Day Formation, Oregon, and its strati- and volume of the Columbia River Basalt graphic implications. Journal of Paleontology Group. In S.P. Reidel and P.R. Hooper (editors), 51(4): 750±757. Volcanism and tectonism in the Columbia River Woodburne, M.O., and C.C. Swisher. 1995. Land- ¯ood-basalt province. Geological Society of mammal high resolution geochronology, inter- America Special Paper 239: 1±20. continental overland dispersals, sea level, cli- Walker, G.W. 1977. Geologic map of Oregon east mate, and vicariance. SEPM Special Publica- of the 121st Meridian. U.S. Geological Survey, tion 54: 335±364.

ADDENDUM During our 1993±1996 study of the upper John represented in our study, has not been mapped Day Formation, we examined the Haystack Valley outside of the Haystack Valley-Balm Creek Valley Member in its type area along Balm Creek, where exposures. Fisher and Rensberger had placed their de®ning Our structural cross-section of the Haystack section. But, as noted, along Balm Creek the base and Balm Creek valleys (®g. 20) shows green of the member was not evident in any Balm Creek zeolitized claystones of the Turtle Cove Member outcrop (®gs. 4±5). In 2003 we attempted to re- forming the ¯oor of Haystack Valley. These beds solve the relationship of the basal contact of the are in certain fault relationship with beds of the member to older John Day units. Only in Hay- Haystack Valley Member (revised) along the west stack Valley itself, which lies immediately west wall of the valley. We also show in ®gure 20 a of the Balm Creek valley, can this stratigraphic hypothetical fault (at Route 207) along the east relationship be determined due to the presence of side of the valley at its southern end. In 2003 we older John Day beds in proximity to the revised found conclusive evidence of faulting here and Haystack Valley Member. The revised member, as also farther north along the east valley wall. The 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 73 sense of these faults could not always be deter- pying the center of Haystack Valley. Critical to mined due to cover, but reverse faulting of green that end are Fisher and Rensberger's (1972) ``Sec- Turtle Cove Member rocks (hanging wall) on tion 11'' and our outcrops C and D (Map E). younger beds (footwall) was certainly established Fisher and Rensberger (1972: ®g. 3) published at one locality (center, sec. 23, T8S, R25E). These only one measured section from Haystack Valley faults, and the presence of springs, seeps, and itself, Section 11. It lies to the east of the west zones of shattered and jointed rock along the east Haystack Valley fault, as mapped by Fisher. Sec- side of Haystack Valley (from SW¼, sec. 28 tion 11's upper part occurs just north of the center northeast to SE¼, sec. 22, thence eastward into of sec. 21, T8S, R25E; the outcrop continues sec. 23, T8S, R25E) suggests the east side of the along a ridge extending to the southern boundary valley is fault-controlled. The trend of this fault of section 21. Fisher and Rensberger's section in- zone is from southwest to northeast (Map E, east cludes 620 ft (189 m) of green Turtle Cove Mem- Haystack Valley fault zone). Reverse faulting ob- ber rocks (including the Deep Creek Tuff) above served here is coincident with reverse faulting of the Picture Gorge ignimbrite, overlain by 140 ft the revised Haystack Valley and Balm Creek (43 m) of gray nonzeolitized Kimberly Member Members in the valley of Balm Creek, where tuff. In the ®eld, we observed a yellowish gray northwest-southeast compression of the formation transition zone (ϳ60±80 ft) between the green and produced the anticlinal-synclinal folds mapped by gray rocks. Fisher (1967). We walked from Section 11 to the west, follow- However, a basal contact of the revised Hay- ing the trend of the gray ``Kimberly tuff'', and stack Valley Member with older John Day rocks discovered that the westernmost outcrop of gray is not evident along the east Haystack Valley fault tuff (outcrop D in Map E) includes gray, coarse zone. Green zeolitized beds of the Turtle Cove ¯uvial sandstone grading along strike into dark Member are in proximity to rocks of the revised green, coarse sandstone typical of Unit 2 of the Haystack Valley Member but the faults leave open Haystack Valley Member (revised). Also present the question of whether some part of the John Day were green celadonite siltstone ``benches'' similar Formation is missing along these fractures. to those in the revised Haystack Valley Member. This problem appears to be resolved by expo- At D, as at Section 11, the gray tuff unit overlies sures along the west wall of Haystack Valley in green claystones typical of the Turtle Cove Mem- secs. 20, 21, and 29, T8S, R25E, Spray and Kim- ber. But also at outcrop D, overlying the gray tuff berly 1:24,000 quadrangles that occur in proxim- unit and separated by a sharp disconformable con- ity to the west Haystack Valley fault (Map E). tact, are the typical olive to yellow-brown silt- Four principal outcrops (A±D) are observed along stones, sandstones and granule to pebble gravel of the west valley wall, separated by vegetated to- the Balm Creek Member. pographic ridges. The southernmost outcrop (A) We observed, in effect, that what had been re- is a cliff section exposing about 140 ft (43 m) of garded as massive Kimberly tuff, near the center the revised Haystack Valley Member, equivalent of sec. 21, T8S, R25E, now contained coarse to rocks of Unit 2 along Balm Creek. This section sandstones and celadonite siltstones like those of occurs west of the west Haystack Valley fault, the revised Haystack Valley Member. We sus- here obscured by vegetation. The lower 60±80 ft pected that a previously unappreciated facies re- (18±24 m) are tan to pale greenish-gray ribbed lationship was being demonstrated along the west tuffs with several levels of dark green, coarse ¯u- wall of Haystack Valley: massive-appearing gray vial sandstones like those observed along Balm tuffs that had been regarded as the Kimberly Creek in Units 1 and 2A. Member graded, to the south, into gray tuff with The revised Haystack Valley Member can be more numerous ¯uvial sandstone interbeds, even- traced to the northeast 0.4 mi (0.64 km) to a sec- tually continuing southward into a thicker, pre- ond, particularly conspicuous outcrop (B), a spec- dominantly ¯uviatile sequence made up of the tacular cliff in the SE¼,SW¼,SE¼, sec. 20, T8S, gray massive tuff (GMAT), ribbed tuffs, more R25E, Spray 1:24,000 quadrangle. Here one can abundant ¯uvial coarse green sandstones, and observe the same section that was seen at outcrop monomictic welded tuff conglomerates of the A, but now visibly faulted against green clay- Balm Creek outcrops (in the NE¼, sec. 32, T8S, stones of the Turtle Cove Member (the fault ap- R25E). The ``Kimberly'' gray tuff of Fisher and pears to be normal as illustrated in ®g. 20). The Rensberger's Section 11 apparently correlated exposures of the revised member represent Unit 2 with the upper part of the revised Haystack Valley of the Balm Creek valley. But here, once again, Member. outcrops A and B have not established the strati- This supposition received additional support graphic relationship of the revised Haystack Val- when we examined a section intermediate in dis- ley Member to the green Turtle Cove beds occu- tance between outcrops B and D. At outcrop C 74 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 75

(SE¼,SW¼,NE¼,SE¼, sec. 20, T8S, R25E) overlain by the revised Haystack Valley Member, gray massive tuff is faulted against green clay- in turn succeeded by the Balm Creek and Rose stones of the Turtle Cove Member. The gray tuff, Creek Members. The revised Haystack Valley which might be mistaken for the Kimberly Mem- Member re¯ects a signi®cant detrital input into ber, contains interbeds of coarse detrital sand- beds of this age, herein dated at ϳ23.5±23.8 Ma, stone. This in itself means little: such coarse ¯u- hence postdating Turtle Cove Member rocks south vial sandstones appear in typical Kimberly Mem- of Kimberly in Turtle Cove that are probably ber exposures. However, microscopic examination mostly older than ϳ25 Ma (see Fremd et al., of the detrital sandstone from C shows grains of 1994: ®g. 14). Thus we can establish de®nite ev- welded tuff among a variety of other clast types, idence of stream incision and ®lling in the John suggesting this unit correlates with the coarse Day Formation in the Haystack Valley region at sandstones and welded tuff gravels of Unit 2B 23.5±23.8 Ma. The intraformational gravels with- along Balm Creek (®g. 8, Section HS-1). Thus, in Kimberly Member channels south and south- several lines of evidence suggest that the revised Haystack Valley Member in south Haystack Val- east of Kimberly may be somewhat older, based ley, a correlative of Unit 2 at Balm Creek, if on Rensberger's entoptychine rodent faunas from traced to the north overlies the green claystones these rocks, when compared to rodents from the attributed to the Turtle Cove Member. lower part of the revised Haystack Valley Member Placing these observations in perspective, the (E. individensÐA. tessellatus zone). However, the John Day Formation in Haystack Valley and along presence of welded tuff conglomerate in the re- Balm Creek emerges as a distinctive sequence of vised Haystack Valley Member (both Units 2A beds best segregated into a set of member-rank and 2B) does mark the earliest record of these lithostratigraphic units unique to this geographic ignimbrite clasts in the eastern facies of the John area. Turtle Cove-like green zeolitized rocks are Day Formation.

APPENDICES 1-1 to 1-15 The biochronology of the upper John Day For- Day Fossil Beds National Monument, Kimberly, mation presented in this study is documented by Oregon; LACM-CIT, Los Angeles County Museum the stratigraphic placement of fossil mammals de- of Natural History, Los Angeles, California; NM, tailed in the appendices. Lithic descriptions of Northwest Museum, Portland State University, Port- stratigraphic units and allocation of mammalian land, Oregon; UCMP, University of California Mu- taxa to speci®c horizons are indicated for each seum of Paleontology, Berkeley, California; UNSM, measured section to facilitate collection of fossils University of Nebraska State Museum, Lincoln, Ne- at critical localities in future studies. braska; UW, University of Wyoming, Geology and Institutional Abbreviations: JODA and TF, John Geophysics, Laramie, Wyoming.

← Map E: Topographic map of the Haystack Creek and Balm Creek area, John Day valley, Oregon, Spray and Kimberly 7.5-minutes quadrangles showing Haystack Valley, bounded by the west and east Haystack Valley faults, adjacent to the Balm Creek syncline. Outcrops A±D in proximity to the west Haystack Valley fault collectively support the inference that the revised Haystack Valley Member is superposed on green Turtle Cove Member rocks in the NW¼, sec. 21, T8S, R25E. Abbreviations of rock units as in ®g. 20. 76 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 77 78 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 79 80 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 81 82 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 83 84 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 85 86 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 87 88 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282 2004 HUNT AND STEPLETON: JOHN DAY FORMATION 89 90 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 282