PaleoBios 26(1):21–26,26(1):21–26, MaMayy 115,5, 2006 © 2006 University of California Museum of Paleontology Magnetic of the Upper (Hemphillian) , central Oregon

DONALD R. PROTHERO1, JONATHAN M. HOFFMAN2, and SCOTT E. FOSS3 1Department of Geology, Occidental College, Los Angeles, CA 90041 2Department of Geology, University of Florida, Gainesville, FL 32611 3Bureau of Land Management, Utah State Offi ce, P.O. Box 45155, Salt Lake City, UT 84145

The Rattlesnake Formation near Picture Gorge in the John Day region of central Oregon consists of about 120 m of siltstones and conglomerates punctuated by several beds. This formation is well known for its early Hemphillian mammals, and it was originally part of the Wood Committee’s (1941) concept of the Hemphillian. Paleomagnetic samples were collected from the type section of the Rattlesnake Formation between Rattlesnake and Cottonwood Creeks 2 km south of Picture Gorge. Samples were demagnetized with both alternating fi eld and thermal demagne- tization, and yielded a stable remanence held mainly in magnetite. After cleaning, the normal and reversed directions passed a reversal test, so the remanence is interpreted to be primary. Almost the entire section is reversed in polarity except for the basal 10 m and a single site near the top of the section. Based on 40Ar/39Ar datdateses of eiteitherher 77.2.2 Ma or 7.05 ± 0.01 Ma on the Rattlesnake Ash Flow Tuff near the top of the section, we correlate the section with magnetic Chrons C3Bn to C3Br2n (6.9–7.3 Ma), or late early Hemphillian in age (Hh2 of Tedford et al., 2004).

INTRODUCTION the term “Rattlesnake Formation” to refer to the entire unit Merriam (1901) fi rst described the Rattlesnake Forma- in the original sense of Merriam (1901). tion, which is the youngest Miocene unit in the John Day The early Hemphillian age of the Rattlesnake Fauna was region. Merriam and Sinclair (1907) and Merriam et al. apparent to Wood et al. (1941), even though at that time (1925) documented its fossils and stratigraphy, and several they thought the Hemphillian was middle Pliocene. Revisions paleontologists have examined and expanded the collections of the Miocene-Pliocene boundary caused Tedford et al. since then (reviewed in Martin, 1983; see also Martin and (1987) to correlate the Hemphillian with the late Miocene. Fremd, 2001). The Wood Committee (1941) designated the Fremd et al. (1994, Fig. 12) reported 23 genera of mammals, Rattlesnake Fauna as one of their principal reference faunas along with turtle and frog remains, from the fossiliferous for the early Hemphillian. conglomerates of the Rattlesnake Formation; that faunal list Merriam (1901), Merriam and Sinclair (1907), and Mer- will be refi ned soon (Martin, pers. commun.). riam et al. (1925) used the name “Rattlesnake Formation” Numerous radiometric dates have been reported for the for the entire fossiliferous mammal-bearing unit that uncon- RAFT. Parker and Armstrong (1972) obtained K-Ar dates of formably overlies the . This terminology 6.6 ± 0.1 Ma and 6.8 ± 0.2 Ma on sanidines from the RAFT. was followed by Enlows (1973, 1976), who described the Walker (1979) reported K-Ar dates ranging from 5.95 ± 0.18 lithology of the “Rattlesnake Formation” in detail. However, Ma to 6.7 ± 0.4 Ma. Streck and Grunder (1995) reported a 40 39 most of the conglomerates and siltstones of the unit are very Ar/ Ar datedate of 7.057.05 ± 0.010.01 Ma, based on 1515 single-crystalsingle-crystal unresistant and localized, whereas the thick resistant Rattle- analyses of alkali feldspars. Swisher (in Martin and Fremd, 40 39 snake Ash Flow Tuff (RAFT) forms a prominent bench. This 2001) reported a Ar/Ar/ ArAr ddateate ooff 77.2.2 MMaa ((nono eerrorrror eestimatestimate bench occurs not only in the John Day region, but the RAFT or details given). Thus, a variety of ages for the RAFT exist was once widespread over central Oregon, covering an area in the literature, although they are beginning to converge in of at least 9000 square kilometers, and possibly as great as their age estimates and the error estimates are decreasing. 40,000 square kilometers with a tuff sheet at least 30 m thick, and up to 70 m thick in places (Streck and Grunder, 1995). METHODS Because of the prominence of this tuff unit, Walker (1979, Sampling of the lower part of the section was conducted 1990) restricted the name “Rattlesnake Formation” to just in the summer of 2001 along the south-facing exposures the RAFT. This leaves the fossiliferous sediments above and (Fig. 1) beneath the RAFT in the type section (NW SE below the RAFT without a name, and ignores the histori- NE SW section 19 T11S R26E, Picture Gorge 7.5-minute cal precedent of Merriam, Enlows and others. J.E. Martin quadrangle, Grant County, Oregon; GPS coordinates UTM (pers. commun.) is currently revising the lithostratigraphy NAD27 Zone 11 E0289782m N4931979m). Retallack et and biostratigraphy of this unit and tentatively recommends al. (2002) described this section in detail. Nine sites (three that the RAFT be designated a formation-rank unit within samples per site) were collected from the 77 m interval below the Rattlesnake Group (Martin and Fremd, 2001). Until the RAFT. The uppermost tuffaceous silty beds above the this change is formally published, we will continue to use RAFT (fi ve sites from a 30 m interval) were sampled to the 22 PALEOBIOS, VOL. 26, NUMBER 1, MAY 2006

A. John Day Fossil Beds National Monument John Day

Hwy. 19 N B. J P

o S h

n B

D o k a u e y n re R d C i a

v r

k y c e o r R

“Lower Rattlesnake Section”

k ree e C nak tles Rat ek re H C w d y o . o 2 “Upper Rattlesnake nw 6 tto Section” Co 1 mile Fig. 1. A. Index map showing location of the Rattlesnake For- mation in central Oregon. B. Detail of the Picture Gorge 7.5- minute quadrangle, showing the location of the two stratigraphic Fig. 2. Orthogonal demagnetization (“Zijderveld”) plots of rep- sections in the Picture Gorge-Cottonwood Creek area. resentative samples. Solid squares indicate declination (horizontal component); open squares indicate inclination (vertical compo- nent). First step is NRM, followed by AF steps of 25, 50, and south of the type section (GPS coordinates UTM NAD27 100 Gauss, then thermal steps from 300 to 600°C. Each division Zone 11 E0290540m N4292591m). The RAFT itself was not equals 10-5 emu. sampled because its paleomagnetism has already been studied; it is reversed in polarity (Stimac and Weldon, 1996). Each oriented sample was taken with simple hand tools by (“Zijderveld”) plots, and average directions of each sample scraping a horizontal surface at the top of the block of rock. were determined by the least-squares method of Kirschvink Samples that were too crumbly in the fi eld were hardened (1980). Mean directions for each sample were then analyzed with sodium silicate. In the laboratory, the block samples using Fisher (1953) statistics, and classifi ed according to the were cored with a drill press; those too small or crumbly to scheme of Opdyke et al. (1977). withstand the drilling were fi rst molded into disks of Zircar About 0.1 g of powdered rock from several samples was aluminum ceramic . The core samples were then measured subjected to increasing isothermal remanent magnetization on a 2G cryogenic magnetometer with an automatic sample (IRM) to determine their IRM acquisition behavior, and thus changer at Caltech. After measurement of NRM (natural re- the relative abundance of magnetite or hematite. They were also manent magnetization), they were demagnetized in alternating AF demagnetized twice, once after having acquired an IRM fi elds (AF) of 25, 50, and 100 Gauss to prevent the remanence produced in a 100 mT (millitesla) peak fi eld and once after of multi-domain grains from being baked in and to examine having acquired an anhysteretic remanent magnetization (ARM) the coercivity behavior of each specimen. AF demagnetization in a 100 mT oscillating fi eld. Such data are useful in conduct- was followed by thermal demagnetization of every sample at ing a modifi ed Lowrie-Fuller test (Pluhar et al., 1991), which steps from 300 to 600°C to remove high-coercivity chemi- indicates whether single or multi-domain grains are present. cal overprints due to iron hydroxides such as goethite, and to determine how much remanence was left after the Curie RESULTS temperature of magnetite (580°C) was exceeded. Representative orthogonal demagnetization plots are Results were plotted on orthogonal demagnetization shown in Fig. 2. In Fig. 2A, the sample exhibits normal PROTHERO ET AL.—MAGNETIC STRATIGRAPHY OF THE RATTLESNAKE FORMATION 23

GRAY TUFFACEOUS SILTSTONE suggesting that the remanence is held largely in magnetite, 100 although the continuing slight increase in IRM suggests some hematite was present as well. In most samples, the ARM 90 was more resistant to AF demagnetization than the IRM, 80 suggesting that the remanence is held in single-domain or 70 pseudo-single-domain grains. M

R Mean directions for all sites are given in Table 1. All but I 60 f

o three of the 14 sites were reversed in polarity, and all but two

t 50 n of the sites were statistically signifi cant, i.e., separated from e c r 40 a random distribution at the 95% confi dence level (Class I e

P 30 sites of Opdyke et al., 1977). One site was missing a sample that crumbled, so it was considered a Class II site (Opdyke 20 et al., 1977). Another site yielded two stable directions, but 10 the third was divergent; it is a Class III site (Opdyke et al., 0 1 3 5 10 30 50 100 300 1000 1977). Magnetic Field (mT) The mean direction for all normal samples was D=0.0, I=48.3, k =4.4, α95 = 27.8 (n=9). The mean direction for all

REDDISH reversed samples was D=195.2, I = -55.1, k = 9.3, α95 = 8.8 100 (n=32). As can be seen from a stereonet (Fig. 4), the mean

90 directions are antipodal within error estimates, so the direc- tions pass a reversal test. This indicates that the overprints 80 have been removed and the resulting vectors represent the 70 primary or characteristic remanence in the samples. M R

I 60 The magnetic stratigraphy of the composite Rattlesnake f

o section is shown in Fig. 5. Almost the entire section is of t 50 n

e reversed polarity; exceptions are the two basal sites (1 and c

r 40

e 2) covering the lowest 10 m of the type section, and one P 30 site (site 12) at 100 m in the composite section (about 20 20 m above the RAFT).

10

0 1 3 5 10 30 50 100 300 1000 Table 1. Paleomagnetic data from the Rattlesnake Formation. Magnetic Field (mT) Fig. 3. IRM acquisition (ascending curve on right) and Lowrie-

Fuller test (two descending curves on left) of a representative SITE DEC INC K α95 powdered sample from the Rattlesnake Formation. Open circles = IRM; solid circles = ARM. 1 18.0 16.6 8.9 44.0 2 301.9 51.5 2.0 133.2 polarity at NRM (the direction is north and down), and the 3 221.1 -63.8 69.2 14.9 rapid decline in intensity through AF demagnetization shows that the remanence is held in a low-coercivity mineral, such as 4 171.9 -62.9 31.9 22.2 magnetite (consistent with the fact that the remanence nearly 5 189.2 -34.4 39.9 19.8 vanished by 600°C). In Fig. 2B, a typical reversed sample shows a single component of remanence, which is reversed 6 181.9 -19.8 7.4 48.9 (south and up) at NRM and decays steadily to the origin, 7 159.2 -52.5 5.0 180.0 losing all remanence as it exceeds the Curie point of magnetite (580°C). This sample apparently has a slight high-coerciv- 8 203.3 -63.3 31.0 22.5 ity overprint of goethite, judging from the gradual decline 9 184.9 -45.2 6.0 55.5 in intensity in the fi rst three AF demagnetization steps. In Fig. 2C, the sample is also reversed at NRM, but it decayed 10 184.9 -70.7 130.8 10.8 steadily to the origin and lost all remanence by 600°C; these 11 239.6 -29.7 36.2 20.8 data and a signifi cant low-coercivity component suggest that the remanence was carried by magnetite. 12 8.9 57.8 280.2 7.4 Representative IRM acquisition analyses are shown in 13 201.5 -69.3 13.9 34.3 Fig. 3. Most samples showed IRM saturation at 300 mT, 24 PALEOBIOS, VOL. 26, NUMBER 1, MAY 2006

N DISCUSSION Correlation of the Rattlesnake Formation paleomagnetic results is shown in Fig. 6. The long reversed magnetozone that covers nearly the entire section is constrained by the age of the RAFT. If either of the 40Ar/Ar/39ArAr ddateate ooff 77.05.05 ± 00.01.01 MMaa (Streck and Grunder, 1995) or the 7.2 Ma date by Swisher (in Martin and Fremd, 2001) is used, this long reversed interval must be Chron C3Br (7.1–7.3 Ma). The simplest in- terpretation is that the short normal interval above the RAFT (magnetic site 12) probably represents Chron C3Br1n, and the basal 10 m of normal polarity represents Chron C3Br2n (Fig. 6). If this age determination is correct, then the entire Rattlesnake Formation spans the interval from 6.9 to 7.3 Ma. This would indicate a late early Hemphillian age, or the Hh2 stage as suggested by Tedford et al. (1987, 2004).

CONCLUSIONS The 120 m-thick type section of the Rattlesnake For- mation (sensu Merriam,Merriam, 1901)1901) includes an importantimportant earlyearly Fig. 4. Stereonet of mean of normal and reversed sites. Solid Hemphillian mammal fauna. Although a variety of dates have circle and lines indicate mean and ellipse of confi dence of normal been obtained for the RAFT, either of the recently obtained samples (lower hemisphere projection); open circle and dashed 40Ar/39Ar agesages of 7.27.2 Ma or 7.057.05 ± 0.10.1 Ma constrainconstrain its cor-cor- line indicate mean of reversed samples (upper hemisphere projec- relation. Magnetostratigraphic analysis shows that the Rattle- tion). Solid square indicates projection of reversed mean to the snake Formation is almost entirely reversed in polarity, and lower hemisphere of the stereonet.

LITHO- MAGN. DECLINATION INCLINATION LOGY SITES 90 180 270 0 90 -90 +90 Ft M 14 300 13 100 12 Gray tuff 11 10 Rattlesnake Ash-flow tuff

9

8 200 7 50

6

100 5 4

3 2 1 0 0

Fig. 5. Lithostratigraphy and magnetic stratigraphy of the type section of the Rattlesnake Formation. Stratigraphy after Retallack et al. (2002). Declination and inclination of magnetic sites are shown. Solid circles are sites that are statistically removed from a random dis- tribution at the 95% confi dence level (Class I sites of Opdyke et al., 1977). Hachured circle indicates that only two samples remained (Class II sites of Opdyke et al., 1977). Open circle indicates site in which two directions gave a clear polarity preference, but the third was divergent (Class III sites of Opdyke et al., 1977). PROTHERO ET AL.—MAGNETIC STRATIGRAPHY OF THE RATTLESNAKE FORMATION 25

Y Martin, J.E. 1983. Additions to the early Hemphillian (Miocene) T I N H A R Rattlesnake fauna from central Oregon. South Dakota Academy O C A M L R L O of Sciences Proceedings 62:23–33. H A P O

Ma E P C N Martin, J.E., and T.J. Fremd. 2001. Revision of the lithostratig- 6 raphy of the Hemphillian Rattlesnake units of central Oregon. C3An n a i

1n l

l PaleoBios 2121 (suppl. ttoo no 2):89. C3An1n i h

C3An p Merriam, J.C. 1901. A contribution to the geology of the John Day e m

2n t e

a Basin. University of California Bulletin, Department of Geology L H 2(9):269–314. e C3Ar n

e Merriam, J.C., and W.J. Sinclair. 1907. Tertiary faunas of the John c n o

i Day region. University of California Publications in Geological

C3Bn a

7 i l

M E. Hemphillian

C3Br1n l i Sciences 5:171–205.5:171–205. e h Rattlesnake Ash-flow tuff t

C3Br p a E. Hemphillian Merriam, J.C., C. Stock, and C.L. Moody. 1925. The Pliocene L m

C3Br2n e

H Rattlesnake Formation and fauna of eastern Oregon, with notes C4n1n y l

r on the geology of the Rattlesnake and Mascall deposits. Carnegie

C4n1r a E Ar/Ar dates: Institution of Washington Publication 347:1–92.347:1–92. C4n 7.05 ± 0.01 Ma fide Streck Opdyke, N.D., E.H. Lindsay, N.M. Johnson, and T. Downs. 1977. or 7.26 Ma fide Swisher 8 The paleomagnetism and magnetic polarity stratigraphy of the Fig. 6. Correlation of the Rattlesnake Formation paleomagnetic mammal-bearing section of Anza-Borrego State Park, California. section, based on the dates and age constraints discussed in the Quaternary Research 7:316–329.7:316–329. text. Radiometric dates after Swisher (in Martin and Fremd, Parker, D., and R.C. Armstrong. 1972. K-Ar dates and Sr-isotope 2001) and Streck and Grunder (1995). Time scale after Berggren ratios for volcanic rocks of the Harney Basin, Oregon. Isochron/ et al. (1995) and Tedford et al. (2004). West 5:7–12.5:7–12. Pluhar, C., J.L. Kirschvink, and R.W. Adams. 1991. Magnetostratig- probably correlates with Chrons C3Bn to C3Br2n (6.9-7.3 raphy and clockwise rotation of the Plio-Pleistocene Mojave River Ma), or the late early Hemphillian. Formation, central Mojave Desert, California. San Bernardino County Museum Association Quarterly 38(2):31–42. ACKNOWLEDGEMENTS Retallack, G.J., S. Tanaka, and T. Tate. 2002. Late Miocene advent We thank Elizabeth Draus for help with sampling, and of tall grassland paleosols in Oregon. Palaeogeography, Palaeo- Dr. Joseph Kirschvink for access to the Caltech paleomag- climatology, Palaeoecology 183:329–354. netics lab. We thank Ted Fremd for access to National Park Stimac, J.P., and R.J. Weldon. 1996. Relation in the NW Basin and Service land. We thank Bruce MacFadden, Jim Martin, and Range in Oregon recorded in the Rattlesnake and Devine canyon an anonymous reviewer for helpful comments on this paper. ash-fl ow tuffs. EOS (1996 Fall AGU Meeting Abstracts):F159. This research was supported by grants to Prothero by the Streck, M.J., and A.L. Grunder. 1995. Crystallization and welding Donors of the Petroleum Research Fund of the American variations in a widespread sheet: the Rattlesnake Tuff, Chemical Society, and by NSF grant 00-00174. eastern Oregon, USA. Bulletin of Volcanology 57:151–169. Tedford, R.H., L.B. Albright III, A.D. Barnosky, I. Ferrusquia- LITERATURE CITED Villafranca, R.M. Hunt Jr., J.E. Storer, C.C. Swisher III, M.R. Voorhies, S.D. Webb, and D.P. Whistler. 2004. Mammalian Berggren, W.A., D.V. Kent, C.C. Swisher III, and M.-P. Aubry. biochronology of the Arikareean through Hemphillian interval 1995. A revised Cenozoic and chronostratigra- (late Oligocene through early Pliocene epochs). Pp. 169–231 in phy. SEPM Special Publication 54:129–212.54:129–212. M.O. Woodburne (ed.). Late and Cenozoic Mam- Enlows, H.E. 1973. Rattlesnake Formation. Oregon Department of mals of North America: Biostratigraphy and Geochronology. Geology and Mineral Industries Bulletin 77:24–27. Columbia University Press, New York. Enlows, H.E. 1976. Petrography of the Rattlesnake Formation in Tedford, R.H., T. Galusha, M.F. Skinner, B.E. Taylor, R.W. Fields, its type area, central Oregon. Oregon Department of Geology and J.R. Macdonald, J.M. Rensberger, S.D. Webb, and D.P.Whistler. Mineral Industries Short Paper 25:1–34.25:1–34. 1987. Faunal succes sion and biochronology of the Arikareean Fisher, R.A. 1953. Dispersion on a sphere. Proceedings of the Royal through Hemphillian interval (late Oligocene through earliest Society A217:295–305.A217:295–305. Pliocene Epochs) in North America. Pp. 153–210 in M M.O..O. Fremd, T., E.A. Bestland, and G.J. Retallack. 1994. John Day Basin Woodburne (ed.). Cenozoic Mammals of North America, paleontology fi eld trip guide and road log.Society of Geochronology and Bios tratigraphy. Columbia University Paleontology Field Trip Guide, 56 pp. Press, New York. Kirschvink, J.L. 1980. The least-squares line and plane and the Walker, G.W. 1979. Revisions to the Cenozoic stratigraphy of the analysis of paleomagnetic data: examples from Siberia and Mo- Harney Basin, southeastern Oregon. U.S. Geological Survey rocco. Geophysical Journal of the Royal Astronomical Society 62: Bulletin 1475:43–92.1475:43–92. 699–718. 26 PALEOBIOS, VOL. 26, NUMBER 1, MAY 2006

Walker, G.W. 1990. Miocene and younger rocks of the Blue Moun- Wood, H.E., III, R.W. Chaney, J. Clark, E.H. Colbert, G.L. Jepsen, tains region, exclusive of the Columbia River Group and J.B. Reeside Jr., and C. Stock. 1941. Nomenclature and cor- associated mafi c lava fl ows. U.S. Geological Survey Professional relation of the North American continental Tertiary. Geological Paper 1437:101–118.1437:101–118. Society of America Bulletin 52:1–48.