Lithostratigraphy, Tephrochronology, and Rare Earth Element Geochemistry of Fossils at the Classical Pleistocene Fossil Lake Area, South Central Oregon

Home , Fort Rock, Fossil Lake (Oregon)

James E. Martin, Doreena Patrick,1 Allen J. Kihm,2 Franklin F. Foit Jr.,3 and D. E. Grandstaff4

Museum of Geology, South Dakota School of Mines and Technology, 501 East St. Joseph Street, Rapid City, South Dakota 57701, U.S.A. (e-mail: [email protected])

ABSTRACT One of the most famous fossiliferous Pleistocene sites in the Pacific Northwest is Fossil Lake, Oregon. Until recently, fossil collections from the area were not stratigraphically controlled, owing to the lack of a detailed stratigraphic and chronologic framework. Our field studies reveal at least nine exposed thin rhythmic fining-upward depositional packages, most separated by disconformities. Analysis of interbedded tephras reveals that the Rye Patch Dam (∼646 ka), Dibekulewe (∼610 ka), Tulelake T64 (∼95 ka), Marble Bluff (47 ka), and Trego Hot Springs (23.2 ka) tephra layers are present in the section, indicating deposition from more than ∼646 ka to less than 23 ka, which includes both the late Irvingtonian and Rancholabrean North American land mammal ages, a much longer time span than previously believed. Bones analyzed from eight of the defined units have distinctly different rare earth element (REE) signatures. Fossils obtain REE during early diagenesis, and signatures are probably closely related to lake water compositions. REE signatures in fossils from lower packages suggest uptake from neutral pH waters. In contrast, REE signatures become increasingly heavy REE–enriched up-section, with positive Ce anomalies in the upper units. REE signatures in fossils from the upper units are very similar to waters from modern alkaline lakes, such as Lake Abert, Oregon, suggesting diagenetic uptake in increasingly alkaline and saline waters. These REE changes suggest increasing aridity up-section, a contention reinforced by the habitat preferences of the terrestrial vertebrates preserved.

Introduction The Fossil Lake area of Lake County, Oregon, may brates. Although the region is well known, rela- represent the most important Quaternary verte- tively little dating of the deposits has been under- brate paleontological region in the Pacific North- taken. Most age speculations have been made on west. Fossil vertebrates and invertebrates have the basis of specimens collected, with the excep- been known from Fossil Lake since 1876, and since tion of a radiometric date of 29,000 yr B.P. (Allison 1989, very large collections have been made from 1966, 1979) from snail shells near the base of the these Pleistocene deposits, culminating in tens of stratigraphic section. This date and many fossils thousands of stratigraphically collected verte- suggest the Late Pleistocene (Rancholabrean North American land mammal age; NALMA), which be- Manuscript received May 4, 2004; accepted November 4, gins at 150 ka (Bell et al. 2004). However, no evi- 2004. dence of Bison, the primary index fossil for the Ran- 1 Author for correspondence: Department of Earth and En- cholabrean NALMA, had been published, and only vironmental Science, University of Pennsylvania, 254-B Hayden Hall, 240 South 33rd Street, Philadelphia, Pennsylvania 19104- two possible specimens have been identified sub- 6316, U.S.A.; e-mail: [email protected]. sequently. The lack of diagnostic Bison remains 2 Department of Geosciences, Minot State University, Minot, made the Rancholabrean designation somewhat North Dakota 58707, U.S.A.; e-mail: [email protected]. problematic. To refine the temporal relationships 3 Department of Geology, Washington State University, Pull- man, Washington 99164, U.S.A.; e-mail: [email protected]. and environmental changes during the Late Pleis- 4 Department of Geology, Temple University, Philadelphia, tocene, our investigations have concentrated on Pennsylvania 19122, U.S.A.; e-mail: [email protected]. careful stratigraphic collection of paleontological

139 140 J. E. MARTIN ET AL.

Table 1. Glass Chemistry of Composite Tephra Samples from Fossil Lake, Lake County, Oregon Rye Patch Dibekulewe Sample T64 Fossil Lake Dam S28- Fossil Lake ash bed Fossil Lake Tulelake Fossil Lake Marble Bluff Fossil Lake Trego Hot Oxide tephra A 505.8 tephra B LD-19 tephra C (17.01 m) tephra D MSH set C tephra E Springs

SiO2 71.75 (.36) 72.20 77.11 (.11) 76.7 70.19 (.26) 70.6 76.74 (.10) 76.5 75.93 (.31) 75.50 (.20)

Al2O3 14.58 (.10) 14.55 12.83 (.03) 13.2 15.09 (.06) 14.9 13.76 (.07) 14.0 13.44 (.06) 13.60 (.10)

Fe2O3 3.44 (.12) 3.40 1.32 (.03) 1.30 3.16 (.03) 3.15 .95 (.02) .99 1.64 (.10) 1.58 (.04)

TiO2 .47 (.04) .44 .07 (.01) .07 .63 (.01) .66 .10 (.01) .10 .23 (.01) .23 (.01)

Na2O 4.24 (.18) 3.70 3.58 (.02) 4.06 4.58 (.24) 4.8 4.10 (.01) 4.1 4.02 (.20) 4.50 (.10)

K2O 3.52 (.13) 3.68 4.28 (.09) 4.06 2.82 (.01) 2.92 2.49 (.02) 2.4 3.34 (.04) 3.32 (.06) MgO .42 (.05) .35 .06 (.00) .05 .81 (.01) .70 .25 (.01) .24 .24 (.01) .21 (.01) CaO 1.44 (.11) 1.46 .59 (.01) .61 2.55 (.01) 2.29 1.55 (.00) 1.50 1.05 (.03) 1.02 (.03) Cl .11 (.02) .13 .12 (.01) n.d. .14 (.01) n.d. .05 (.00) .07 .10 (.01) n.d. Totala 100 100 100 100 100 100 100 100 100 100 No. tephra samples 14324 No. shards analyzed 17 67 49 32 79 Similarity coefficient .97 .96 .96 .98 .97 Note. Also shown are compositions of reference tephras, similarity coefficients, numbers of samples and shards analyzed, and the analytical standard deviation (SD); 1 SD of the analyses given in parentheses.MSH p Mount St. Helens; n.d. p not determined. Source. Rye Patch Dam, Williams (1994); Dibekulewe ash bed, Davis (1978); Sample T64 Tulelake and Trego Hot Springs, Rieck et al. (1992); Marble Bluff MSH set C, Davis (1985). a Analyses normalized to 100% on a volatile-free basis. samples from discrete lithologies. We identified ticle presents the first major study of REE in fossils tephra layers interbedded within these lithological from a lake-dominated environment. units that aid in the temporal resolution. Herein, we describe the detailed lithostratigraphy, tephrochronology, and rare earth element (REE) signatures Analytical Methods of the fossils and their implications for the history Electron Microprobe Analysis of Tephra. Mounts and evolution of Fossil Lake. for analysis of tephra samples consisted of four 8- Since its discovery in 1876, Fossil Lake has pro- mm-diameter glass rings (cells) fixed to a standard duced thousands of invertebrate and vertebrate fos- petrographic slide using epoxy, allowing four sam- sils; many were described in the late nineteenth ples to be prepared simultaneously. A few hundred and early twentieth centuries. Fish (Cope 1883), milligrams of tephra followed by a few drops of birds (Shufeldt 1913; Howard 1946), and mammals epoxy were added to each cell, thoroughly mixed, (Elftman 1931; Martin 1996) are particularly abun- and allowed to cure. The cells were trimmed using dant. However, none of the early collections were a microdiamond saw, ground to thin section thick- stratigraphically controlled. Therefore, the area has ness, polished, and carbon coated. been recollected over the last 15 yr, resulting in The composition of the glass in the tephra was thousands of specimens tied to thin lithostrati- determined using the Cameca Camebax electron graphic units. A future goal is to analyze the new microprobe in the GeoAnalytical Laboratory lo- collections in a temporal framework; therefore, a cated in the Department of Geology at Washington detailed lithostratigraphic framework combined State University. The analytical conditions are the with tephrochronology was required. The first por- same as those described in Hallett et al. (2001), and tion of this contribution represents the results of the tephra glass compositions are presented in table these undertakings. Based on the framework de- 1. The compositions of the glasses were compared scribed herein, the second portion of the contri- to one another and to the standard glasses in the bution represents an analysis of REE signatures in GeoAnalytical Laboratory’s database of more than fossils, which was to provide an independent as- 1500 Pacific Northwest tephra samples using the sessment of paleoenvironments that could be com- similarity coefficient (SC) of Borchardt et al. (1972) pared with results derived from the fossil succes- as a discriminator. The SC is the average of eight sions. REE signatures in fossils reflect the REE oxide concentration ratios between the glasses in content of the original pore waters during the fos- the tephra sample and tephra standard. It was cal- silization process (Henderson et al. 1983; Trueman culated using a weighting of 1.0 for the Si, Al, Ca, 1999) and may be used to infer original paleoen- Fe, and K oxide concentration ratios, 0.50 for the vironmental or early diagenetic conditions (Elder- Na oxide ratio, and 0.25 for the Mg and Ti oxide field and Pagett 1986; Patrick et al. 2004). This ar- ratios. Na was a given lower weighting because of Journal of Geology FOSSIL LAKE TEPHROCHRONOLOGY 141

well-recognized impediments to accurate measure- Ultrex-grade HNO3 and diluted to 100 mL. When ment and susceptibility to post-depositional envi- only smaller masses of bone were available, the ronmental modification. Mg and Ti were given initial solution was diluted to smaller volumes to lower weighting because of their low concentra- preserve the mass/volume ratio. Samples were ditions and, consequently, their high relative error of luted to appropriate levels with 2% ultra-pure 1 measurement. Similarity coefficients 0.94 (Rieck HNO3. Initial samples were analyzed using a Fin- et al. 1992) are generally considered permissive of nigan MAT Element/1 high-resolution magnetic a correlation, provided they are consistent with sector inductively coupled plasma–mass spectrom- stratigraphic constraints. eter (ICP-MS) at Rutgers University. The Hawaiian REE Analysis. In this study, REE concentrations Basalt (BHVO) was used as a standard; precision and relative to certified 4%ע were measured in vertebrate fossil materials from accuracy were better than each major stratigraphic unit at Fossil Lake to de- values. Analytical procedures followed those by tect differences in REE signatures between the lith- Field and Sherrell (1998). Subsequent analyses were ologic units, most of which were !0.5 m thick. conducted using a VG Plasma-Quad 3 ICP-MS in These REE signatures in fossils from the various the Geology Department at Temple University. Re- units were compared with REE signatures in nat- sults have been corrected for isobaric interferences ural waters to infer possible paleoenvironmental or where necessary. The coefficient of variation for -of the analyzed value. Ana 5%ע! early diagenetic environments. The interpretations most REEs is from this study are compared with previous studies lytical results have been normalized relative to the of paleoenvironmental conditions at Fossil Lake North American Shale Composite (NASC; Gromet and neighboring pluvial Lake Chewaucan (Allison et al. 1984). Cerium anomalies (Ce∗ ) have been cal- 1966, 1979, 1982; Cohen et al. 2000; Negrini et al. culated from 2000). Osteological elements from various species and 2Ce ∗ p N Ϫ eggshells were sampled from all of the described Ce ϩ 1, ()LaNNPr lithologic units (table 2). This allowed investigation of changes in REE signature and paleoenvironment where the subscript N indicates NASC-normalized with time, as well as possible influence of osteo- values. logical material and species. Fossils were obtained from two localities in the upper siltstone of the fourth depositional package and from five sites in Lithostratigraphy of Fossil Lake the lower coarse unit of the third depositional pack- Fossil Lake (fig. 1) is a small playa (∼16 km2) in the age (table 2) to investigate possible lateral varia- Fort Rock Basin of south central Oregon, east of tions in signatures over the area of the Fossil Lake Crater Lake (Mt. Mazama) and southeast of the Basin. The latter five sites are separated by up to Newberry volcanic complex. These and other vol- 16 km. canoes have provided tephra and volcaniclastic de- Sample preparation techniques follow those by bris to the region (Kuehn and Foit 2001). Fossil Lake Staron et al. (2001). Large intact bone specimens lies within a large fault-bounded basin in the north- were sampled by drilling using a Dremel variable- ern portion of the Basin and Range Province. This speed electric drill. Between samples, the drill was large basin trends nearly 65 km east-west and may cleaned with diluted (2%) trace-metal-grade nitric be considered to be composed of three smaller ba- acid or trace-metal-grade acetone and rinsed in dis- sins, which are connected at higher elevations: tilled water. Cortical bone samples were immersed westerly Fort Rock Basin, easterly Christmas in water and then in a diluted acetic acid solution Valley Basin, and the Silver Lake Basin to the south- in an ultrasonic bath to remove matrix and sec- west. Fossil Lake lies within the Christmas Valley ondary carbonate. Other matrix and secondary Basin. minerals were removed by handpicking. Samples During the Pleistocene, the entire basin was in- were then rinsed with distilled water and dried. undated and formed pluvial Fort Rock Lake (Alli- Selected samples were analyzed using x-ray dif- son 1940, 1979). At its maximum, the lake included fraction, revealing essentially pure apatite with the Silver Lake Basin to the south and is estimated only minor amounts of other crystalline phases. to have been as much as 60 m (200 ft) deep, ex- Cleaned bone fragments were mechanically tending over 1200 km2 (500 square miles; Snyder crushed in an acid-washed mortar and pestle. Ap- et al. 1964; Allison 1979). The lake appears to have proximately 0.2 g of powder was weighed for each waxed and waned a number of times, probably re- sample. Each sample was dissolved in about 2 mL lated to the waxing and waning of Pleistocene gla- Table 2. Results of Rare Earth Element (REE) Analyses of Fossils from Fossil Lake, Oregon (ppm) SDSMT Specimen Package No. locality La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Fish 1 10FL V893 1939 4248 522.2 2368 526.6 140.4 572.1 89.31 628.1 137.1 412.4 63.10 416.9 61.01 Enamel/horse 1 12FL V893 158.1 395.7 33.96 155.7 34.90 10.86 50.08 7.82 59.07 14.69 49.66 8.35 61.93 10.20 Cortical/mammal 1 13FL V893 529.1 1124 134.4 660.1 142.4 39.26 172.1 24.25 177.9 43.25 140.86 22.44 155.5 24.62 Fish 1 16FL V893 313.9 685.3 74.71 342.0 75.40 21.05 101.2 15.11 110.0 25.88 85.31 13.39 94.73 15.58 Fish 2 18FL V895 1573 3144 417.1 2099 564.0 174.8 803.7 135.97 1100 287.3 980.5 155.5 1090 176.8 Cortical/mammal 2 19FL V895 91.68 73.08 6.43 28.94 6.44 2.06 9.23 1.34 9.98 2.46 8.62 1.45 12.15 2.10 Fish 3 23FL V8914 484.6 999.0 113.77 520.4 112.9 31.73 146.6 22.17 157.3 36.91 122.09 19.79 137.48 23.24 Fish 3 24FL V8934 108.5 305.1 14.58 71.08 16.36 6.44 31.13 3.05 21.24 4.67 16.20 2.85 23.29 3.93 Fish 3 25FL V891 229.4 497.4 53.37 261.2 61.21 18.70 91.59 13.55 102.0 25.07 86.35 13.78 100.63 17.23 Cortical/bird 3 26FL V891 225.6 521.2 47.01 225.3 52.68 16.20 68.62 10.11 74.00 18.14 59.61 10.02 65.12 12.14 Cortical/mammal 3 28FL V8914 89.95 173.3 18.99 89.54 20.41 6.25 29.78 4.41 32.87 8.17 28.95 4.84 35.75 6.32 Cortical/bird 3 30FL V8914 102.2 183.0 19.88 93.34 20.39 6.26 31.17 4.58 34.07 8.72 29.99 5.04 38.51 7.11 Fish 3 41FL V892 119.5 226.2 24.92 114.5 27.38 8.07 33.70 4.84 35.47 8.53 28.19 4.37 30.90 5.05 Trabecular/mammal 4 43FL … 57.59 113.1 13.65 65.68 15.27 4.41 18.04 2.84 21.22 5.47 19.55 3.44 27.52 5.22 Fish 5 44FL V891 77.81 195.7 22.27 99.70 22.65 6.79 26.87 4.57 34.40 8.52 28.69 4.80 35.80 6.24 Cortical/mammal 5 45FL V891 41.07 100.7 12.15 52.32 11.88 3.44 12.46 2.14 15.90 3.95 13.73 2.39 19.44 3.35 Fish 6 32FL V8928 85.02 325.7 17.71 87.79 19.96 6.17 31.87 5.02 39.77 10.93 40.20 7.00 57.04 11.52 Cortical/mammal 6 34FL V8928 47.29 675.7 15.42 64.94 14.24 3.97 14.64 2.66 19.44 4.82 17.40 3.26 29.00 5.06 Cortical/bird 6 49FL V8928 2092 4608 605.1 2761 623.1 180.0 820.3 129.5 868.9 192.0 574.5 82.95 532.0 82.77 Cortical/mammal 7 36FL V8913 94.50 382.3 25.58 113.9 25.31 8.08 33.78 6.35 50.50 13.54 49.21 8.48 67.25 12.28 Fish 7 50FL V8913 462.7 812.1 118.0 571.9 137.9 40.49 190.7 29.76 231.6 61.14 209.7 33.65 236.5 38.45 Cortical/mammal 7 47FL V8928 15.27 115.7 4.10 17.20 3.79 1.14 4.46 .84 6.13 1.56 5.66 1.05 9.47 2.06 Trabecular/mammal 7 48FL V8928 15.51 38.8 4.54 18.08 3.90 1.07 3.59 .63 3.86 .79 2.29 .35 2.83 .45 Fish 8 38FL V8913 94.23 1236 31.85 148.2 39.89 13.52 63.75 12.99 112.09 32.90 128.02 23.17 189.2 35.64 Fish 8 39FL V8913 40.92 212.3 13.81 64.48 17.51 6.25 30.04 6.44 57.44 16.83 63.22 10.56 75.78 12.73 Rabbit 8 40FL V8913 30.13 1030 12.96 57.69 14.97 4.72 19.53 3.45 27.33 7.23 27.65 5.26 46.41 8.30 Note. Package numbers as in figure 2; REE concentrations in parts per million.No. p REE specimen number; SDSMT p South Dakota School of Mines and Technology. Journal of Geology FOSSIL LAKE TEPHROCHRONOLOGY 143

phy of the area came in 1974 when one of us (J. E. Martin) studied environmental impacts at the site for the Bureau of Land Management. Lithological units were differentiated initially by weathering profile, which mirrors composition. All of the units at Fossil Lake are thin, and although Allison (1966) recognized only three tephra layers interbedded within the detrital units, we determined that five separate tephra layers exist. These tephra layers are normally associated with a rhythmic depositional pattern of fining-upward sequences. Normally, each sequence is about a meter and a half thick and composed of a basal basalt conglomerate grading upward into a tuffaceous, friable, basaltic sandstone (Bertog et al. 1997) overlain by a silty claystone, representing lacustrine deposition. We recognize at least nine fining-upward depositional packages exposed in the Fossil Lake area. Erosional processes, Figure 1. A, Outline map of Oregon. The shaded area providing the key for recognition, differentially pre- contains the Fossil Lake region in B. B, Map of the north- serve the upper claystone of each depositional ern portion of Lake County, Oregon, showing location of package. These claystones were noted by Allison modern lakes and maximum extent of pluvial Fort Rock (1966), but he miscorrelated some, did not discern Lake (dashed line; Allison 1982). The Fossil Lake area, the subtle differences between them, and thus did from which most stratigraphic sections and fossils were not recognize the repeated depositional pattern. obtained, is shown by the filled triangular area in the Once the pattern was recognized, it was tested by p Christmas Valley Basin (CV).FRB Fort Rock Basin, extensive trenching and correlation of the tephra FR p town of Fort Rock,CA p California ,ID p Idaho , layers. NV p Nevada,.WA p Washington These efforts resulted in the following composite section (fig. 2). The base of the exposed section is ciers. Following final desiccation, eolian processes represented by a yellow claystone, which is more created extensive dune fields, particularly obvious than a meter thick; its base was not located. This in the Fossil Lake area, where active dunes remain. claystone, which hosts tephra layer A (Rye Patch The fossils have been exposed due to wind and dune Dam; table 1; fig. 2), may represent the uppermost migration, augmented by erosion from sheet wash. interval of a depositional package (package 1; fig. The geology of Fossil Lake was discussed by War- 2). This claystone is overlain by basaltic gray-brown ing (1908), who agreed with Russell (1884) that the unconsolidated sandstone, which contains an in- region was flooded by large Pleistocene lakes. terbedded, discontinuous tephra layer (tephra layer Smith (1926) first used the Fossil Lake Formation B, Dibekulewe; table 1), representing the base of for 2.4–3 m of lake silts that contained numerous the second depositional package. Scattered basaltic Pleistocene fossils. Some of the first detailed geo- pebbles lie at the base of the sand, and weathering logical investigations at Fossil Lake were those by has resulted in ferruginous staining. Eolian pro- Ira Allison, who began investigations in 1939 and cesses result in depressions within this unit published a series of articles culminating in his re- throughout the Fossil Lake area, and these are char- views of the geology and paleontology of Fossil acterized by concentrations of snails. From this Lake (1966) and the regional geology (1979). When level, Allison (1966) obtained a 14C date on snails Allison noted tephra layers at Fossil Lake, he used of 29,000 yr B.P. This sand grades upward into a them for correlation but not for absolute dating. transitional brownish siltstone with some sand in- Allison did obtain a radiocarbon date of 29,000 yr terbeds grading into a white claystone, totaling 90 B.P. from fossil snails (Allison 1966, 1979) lying cm. The white claystone is composed principally below a tephra layer at a site where prolific bird of reworked tephra and represents the uppermost remains occur. We have reinvestigated the site, unit of the second depositional package. The base known as the Bird Site, and sampled the associated of the third package represents a major disconform- tephra (tephra layer B, Dibekulewe; table 1; fig. 2) ity and contains a pebble to small cobble basal ba- and snails. salt gravel, only a few centimeters thick, which is Our efforts in understanding the lithostratigra- laterally discontinuous. The conglomerate is over- 144 J. E. MARTIN ET AL.

Figure 2. Composite stratigraphic sections, analyzed tephra layers (table 1), and tephra ages at Fossil Lake, Oregon. Fossil Lake sediments have been subdivided into numbered depositional packages comprising lithostratigraphic units. Grain sizes for individual units are indicated by lengths of the bars (scale at base of section). Major unconformities occur between packages 3 and 4 and at the base of package 5. lain by a basaltic, loosely consolidated sandstone posed of poorly consolidated basaltic sandstone (approximately 80 cm), which contains a distinc- with scattered pebbles from 12 to 70 cm thick over- tive 8-cm pumice bed (tephra layer C, Tulelake lain by a brown siltstone/claystone, ranging from T64), also noted by Allison (1966). The sand is in 12 to 24 cm. Within the siltstone/claystone of the turn overlain by a brown, silty claystone, up to 60 seventh package is a bright white, powdery tephra cm thick, capping the third rhythmite. Another layer (tephra layer E, Trego Hot Springs), the young- sandstone (30 cm thick) with scattered pebbles oc- est tephra within the succession. The widespread curs at the base of the fourth package. The upper- upper package of ferruginous sand (up to 145 cm most layer of the fourth package, a distinctive, tuff- thick) grades upward into a brown siltstone (12 cm), aceous tan siltstone ranging from 24 to 46 cm thick, which is in turn overlain by a white claystone (55 overlies this sand. Another tephra layer (tephra cm) characterized by salmonid remains. In most layer D, Marble Bluff; table 1; ﬁg. 2) is associated areas, the white claystone represents the upper- with the siltstone, resulting in the distinctive color. most unit; however, one area exhibited remnant Above this package, lateral variation exists. To the pockets of sand a few centimeters thick, indicating east, at least three additional packages lie between the succession was thicker, but erosion has reduced the tan siltstone and the base of the uppermost the section essentially to the top of the white clay- package. To the west, a disconformity exists, and stone. Overall, nine ﬁning-upward depositional the base of the suprajacent package consists of dis- packages have been distinguished, which appear to tinctively cross-bedded, ferruginous sands. The ad- have been formed during repeated waxing and wan- ditional packages to the east (5–7) are each com- ing of the Pleistocene lakes. Journal of Geology FOSSIL LAKE TEPHROCHRONOLOGY 145

Tephrochronology and Radiocarbon Dates seventh depositional package, is a very good match ka Trego Hot Springs 0.3 ע SC p 0.97 ) to the 23.2) Although fossils from Fossil Lake are useful for age tephra (Benson et al. 1997). constraints, only a few independent absolute age These dates are not consistent with the 14C date yr B.P. (Isotopes I-1641) taken 2000 ע dates have been obtained (Allison 1966, 1979). Rec- of29,000 ognition of the interbedded tephra layers within the from gastropod shells collected by Allison (1966) depositional packages indicated that a chronologic from the tephra layer near the base of the second framework could be developed to complement exposed depositional package. The ages of all but the paleontological and stratigraphic frameworks. the uppermost Trego Hot Springs tephra layer are Once the composite section (fig. 2) was proven and significantly older than this 14C age. Studies (Tan- tested through widespread outcrop and trench cor- ners 1970; Goslar and Paxlut 1985; Yates 1986) have relation, the tephras were analyzed, and geochem- suggested that 14C dates from snails may not be ical comparisons were made to known regional reliable because of recrystallization of the carbon- tephra layers. The results were previously sum- ate or uptake of isotopically old carbonate carbon. marized by Martin and Kihm (2003), and more re- In an attempt to resolve this inconsistency, new cent unpublished dates of some tephras are in- mollusk shells with original shell coloration were cluded in the following discussion. collected from Allison’s (1966) locality. Scanning The lowest tephra layer, tephra layer A (fig. 2), electron microscopic (SEM) examination did not re- lies at the top of the lowermost exposed package veal any recrystallization of the newly collected at the top of a yellow claystone. The glass in this shell material. However, an even younger 14C date -yr B.P. (Beta no. 173726) was ob 70 ע tephra is a very good match (SC p 0.97 ; table 1) to of13,420 that in the Rye Patch Dam tephra believed to be tained for one of these shells. Extensive excavation equivalent to the Desert Springs ash flow near within Allison’s sampling site revealed a probable Bend, Oregon, estimated to be 630 ka (Rieck et al. explanation for this date incongruence: every gas- 1992) and dated by laser fusion 40Ar/39Ar to tropod found in situ was found in a rodent burrow, ka (R. J. Fleck, pers. comm. 2004). How- indicating intrusion by bioturbation. Therefore, it 50 ע 686 ever, based on the age and position of the overlying appears that the buried gastropods at this level are ,ka; Lanphere et al. not the result of primary deposition. Moreover 0.2 ע Lava Creek B tephra (639 2004) and the Bruhes/Matuyama Chron boundary fish, bird, and mammal remains were seldom found in the Tulelake core (Rieck et al. 1992), Sarna- in the burrows but were ubiquitous in the sur- Wojcicki (pers. comm. 2004) estimates the Rye rounding bedded sand unit of the second deposi- Patch Dam ash bed to be ∼646 ka. The second pack- tional package. The variability in the gastropod age contains tephra layer B in the basal sand unit dates and the high degree of fossilization of the and compares well (SC p 0.96 ) with the Dibeku- mammalian bones from this low in the section lewe ash bed of the western United States, similarly make the radiocarbon dates suspect. As noted pre- ka (A. Sarna-Wojcicki, viously, although they might be environmentally 20 ע estimated to be ∼610 pers. comm. 2004). Fossil gastropods described and or geographically excluded from this area, the ab- 14C dated by Allison (1966) come from this level. sence of the Rancholabrean index fossil Bison also Tephra layer C, which lies in the lower sandstone casts some doubt on the younger ages proposed by of the third depositional package, consists of pum- Allison (1966, 1979). ice fragments, the composition of which is a very Moreover, Fossil Lake contains the same tephra good match (SC p 0.97 ) to sample T64 from a sequence, encompassing roughly the same time in- depth of 17.01 m in Tulelake, estimated to be be- terval as neighboring Summer Lake (Davis 1985; ka (A. Sarna- Negrini et al. 2000; Kuehn and Foit 2001). Although 23 ע tween 72 and 119 ka or ∼95 Wojcicki, pers. comm. 2004). A basal basalt lag sug- all of the tephra layers at Fossil Lake appear to be gests a significant basal unconformity compared slightly reworked and therefore hosted in poten- with those of most other depositional packages. tially younger sediments, the significant discon- Tephra layer D is associated with the claystone of formity observed at the base of sedimentary pack- the fourth depositional package. The composition age 3, which hosts the Tulelake T64 tephra layer of the glass in this tephra is an excellent match (tephra layer C), is stratigraphically consistent with (SC p 0.98 ) to that in the Marble Bluff bed (Mount the unconformity reported at Summer Lake (Davis St. Helens set C) of Davis (1978, 1985), which has 1985; Negrini et al. 2000). This disconformity not ka (Berger and Busacca 1995; only explains the absence at Fossil Lake of the 2 ע been dated at47 Negrini et al. 2000). The uppermost tephra (tephra many tephra layers in the age range of 130–250 ka layer E), which occurs in the siltstone unit of the reported in the Ana River section at Summer Lake 146 J. E. MARTIN ET AL.

(Davis 1985) but also suggests that the tephrochron- ficients are also large, resulting in near-quantitative ological age of the sediments above the disconform- removal of REE from solution (Kemp and Trueman ity is accurate. A deep core taken close to and ex- 2003; see however, Reynard et al. 1999). tending approximately 50 m below the Ana River REE incorporation into the fossil may be largely outcrop revealed the Dibekulewe ash bed (Sarna- accomplished within a few thousand years (True- Wojcicki et al. 2001), which is consistent with its man 1999; Patrick et al. 2001; Trueman and Tuross stratigraphic position at Fossil Lake. To date, the 2002). If fossilization is rapid relative to the sedi- Rye Patch Dam tephra has not been observed in ment deposition rate, the REE signature in the fos- this core. sil will primarily reflect the surface water chem- Thus, it appears that the Fossil Lake section is istry; however, if deposition is rapid, REE may be significantly older than suggested by Allison (1966, taken up primarily from sediment pore waters (Eld- 1979), that the lower two sedimentary packages erfield and Pagett 1986; Kemp and Trueman 2003). (fig. 2) were deposited during the Irvingtonian Incorporation of REE, fluorine, and other TEs, and NALMA, and that packages 3–8 were deposited growth of larger, more crystalline apatite decreases during the Rancholabrean NALMA as previously the solubility, specific surface area, and reactivity suggested by Martin and Kihm (2003). A major dis- of the resulting fossilized material. Most studies conformity below package 3 represents much of (see discussion in Trueman 1999; Armstrong et al. the later Irvingtonian and early Rancholabrean 2001) have suggested that once REEs are incorpo- ∼ NALMA ( 600–150 ka) based on tephrochronology. rated in the bone, they are retained and provide a Therefore, the later part of Irvingtonian II, Irving- stable signal reflecting the depositional or early dia- tonian III, and early Rancholabrean of Bell et al. genetic environment. Exceptions occur only when (2004) are degraded. the fossil material is dissolved and reprecipitated or recrystallized during high-grade metamorphism, REE Signatures in Fossils processes that have not taken place in the Fossil Lake sediments. Factors Influencing REE Signatures in Fossils. Ex- If REE signatures in fossil bones are similar to tensive discussions concern sources of REE, mech- those of the early diagenetic fluids from which the anisms controlling the REE content of fossils, and REE were obtained, as suggested by the previous the amount of fractionation of REE between fluids discussion, an understanding of REE signatures in and osteological materials (Bernat 1975; Wright et modern natural waters and how they are affected al. 1984, 1987; Elderfield and Pagett 1986; Reynard by environmental properties and geological vari- et al. 1999; Trueman 1999; Armstrong et al. 2001; ables is required to infer paleoenvironmental con- Trueman and Tuross 2002; Kemp and Trueman ditions. REE concentrations have been measured in 2003). Biogenic apatite crystals in living organisms seawater, rivers, lakes, and continental pore and are very small (Lowenstam and Weiner 1989; Wei- ground waters (Elderfield et al. 1990; Johannesson ner and Traub 1992); their structures have low de- et al. 1993, 1996, 2000; Mo¨ ller and Bau 1993; Jo- grees of crystallinity (Person et al. 1995) and con- tain relatively high concentrations of carbonate, hannesson and Zhou 1999; Dia et al. 2000; Biddau sodium, and other species. Therefore, biogenic ap- et al. 2002; and review articles by Byrne and Shol- atite crystallites are relatively soluble and reactive. kovitz 1996; Johannesson and Xiaoping 1997). Fac- During fossilization, the crystallites of biogenic ap- tors controlling REE concentrations, speciation, atite recrystallize (Brophy and Hatch 1962; Person and signatures in natural waters are still a matter et al. 1995) and grow. REE and most other trace of discussion (Byrne and Kim 1993; Johannesson elements (TEs) are adsorbed from solutions onto and Xiaoping 1997; Johannesson et al. 1995; Tang the biogenic apatite crystal surfaces (Tuross et al. and Johannesson 2003). Byrne and Kim (1993) and 1989a, 1989b) and then incorporated into the fos- Johannesson et al. (1995) suggested that many oce- silizing material during apatite crystal growth anic and continental waters are approximately sat- (Trueman 1999; Armstrong et al. 2001). Henderson urated with respect to REE phosphate coprecipi- et al. (1983) suggested that the values of distribu- tates. REE concentrations and signatures in natural tion coefficients for all of the REE, except perhaps waters appear to be largely controlled by solution Ce, are similar (see also Trueman 1999; Trueman pH and redox; concentrations of complex-forming and Tuross 2002). Therefore, signatures and ratios ligands, particularly carbonate and possibly organ- of REE in fossils should be very similar to the av- ics; sorption on hydrous ferric oxides (HFOs) and erage composition of the fluids from which they manganese oxides, as well as by the composition were obtained, particularly if the adsorption coef- of source rocks or minerals providing REE to fluids Journal of Geology FOSSIL LAKE TEPHROCHRONOLOGY 147

(Johannesson and Xiaoping 1997; Hannigan and Sholkovitz 2001; Tang and Johannesson 2003). Some representative NASC-normalized REE signatures are shown in figure 3. Signatures from acidic environments tend to be middle REE (MREE) to light REE (LREE) enriched; those from neutral environments may be MREE-, LREE-, or high REE (HREE)-enriched or have flat patterns (fig. 3A) showing no specific enrichment (Johannesson and Xiaoping 1997). Seawater and alkaline lakes (fig. 3B) tend to be HREE enriched (Johannesson et al. 1993; Mo¨ ller and Bau 1993). The strength of different REE-ligand complexes varies along the REE series. For example, HREEs form stronger complexes with carbonate than do the LREEs (Wood 1990; Haas et al. 1995) and may thus be more soluble and less adsorbed than LREE in higher pH alkaline solutions (Johannesson et al. 1993; Mo¨ ller and Bau 1993; Byrne and Sholkovitz 1996). Alkaline lakes are of- ten relatively more HREE enriched than seawater (fig. 3B) as a result of greater carbonate complexing (Johannesson et al. 1993). In areas dominated by silicate rocks, major features of dissolved REE signatures are not strongly affected by rock type. How- ever, river and ground water in limestone terrains may have LREE-depleted signatures and negative Ce anomalies that are inherited from the marine limestones and seawater REE values (Johannesson and Xiaoping 1997). Dissolution of HFOs or phosphates may also provide an LREE-enriched or MREE-enriched source for some river and lake waters (Johannesson and Zhou 1999; Hannigan and Sholkovitz 2001). Variations of REE signatures of natural waters Figure 3. North American Shale Composite (NASC)– and fossils may be represented on a ternary diagram normalized rare earth element (REE) signatures for some (fig. 4) with NASC-normalized Yb, Gd, and Nd representative natural waters from various environments. A, Acidic to circumneutral pH environments. B, (YbN,GdN, and NdN) at the vertices (Patrick et al. 2002b, 2004). Yb is a representative HREE, Gd an Alkaline lakes and seawater. Reported pH values have MREE, and Nd an LREE. These elements have been been truncated to one significant figure. References: chosen to facilitate comparisons with previous re- Summer, Mono, Abert Lakes (Johannesson et al. 1993); PZ-264, PJ1-335 (Dia et al. 2002); Army, Grapevine, J-13, search on aqueous REE (see Patrick et al. 2002b, Tippipah (Johannesson et al. 2000); Quaking 4.1, Mullica 2004 for further discussion). The ternary diagram (Elderfield et al. 1990); Lake Van (Mo¨ ller and Bau 1993); allows the basic shape of the REE pattern to be Sardinia sample 5 (Biddau et al. 2002); Seawater (Pacific represented. NASC-normalized samples that plot average near-surface seawater; Piepgras and Jacobsen in the middle of the diagram (33% YbN, 33% GdN, 1992). and 33% NdN) have equal amounts of these elements and will have flat, shalelike REE patterns. Patterns plotting near the vertices will be enriched 2000; Stetzenbach et al. 2001; Biddau et al. 2002) in that and neighboring REE; for example, patterns tend to plot in the upper left part to middle of the plotting near Yb will be HREE-enriched. ternary diagram as they tend to be LREE enriched REE compositions of some representative natural and MREE enriched or have flat patterns. In con- waters are plotted in figure 4. Representative acid trast, seawater (Piepgras and Jacobson 1992) and the mine drainage (Worrall and Pearson 2001), river, alkaline lakes (Mo¨ ller and Bau 1993; Stetzenbach spring, and circumneutral pH ground waters (Krea- et al. 2001), which are HREE-enriched, plot toward mer et al. 1996; Dia et al. 2000; Johannesson et al. the lower right (Yb-enriched) portion of the dia- 148 J. E. MARTIN ET AL.

solution. However, oxidizing highly alkaline lake waters, such as those of lakes Van and Abert (fig. 3B), may also have positive Ce anomalies, which may result from the stabilization of Ce4ϩ pentacarbonate complexes in the highly alkaline, high ionic strength waters (Mo¨ ller and Bau 1993; Johan- nesson et al. 1996). The differences in REE signatures in natural waters, which are a function of environmental pH, redox, and alkalinity, may be compared with variations of REE signatures in fossils. Because the fossil REE signatures reflect the original early diagenetic water composition from which the REEs were obtained, some aspects of the ancient water chemistry and early diagenetic paleoenvironment may be inferred. REE in Vertebrate Fossils. Results of REE analyses of fossils from the various lithostratigraphic units at Fossil Lake are given in table 2. Various available Figure 4. Ternary diagram showing variations of rare osteological materials were analyzed from each de- earth elements in terrestrial waters from various relevant positional package. REE signatures in cortical bone environments. Circumneutral pH (Johannesson et al. materials from fish, mammals, birds, and avian egg- 1996, 2000; Kreamer et al. 1996; Stetzenbach et al. 2001), shell from sedimentary package 3 are shown in fig- Alkaline lakes (Johannesson et al. 1993; Mo¨ ller and Bau ure 6. Although the concentrations differ, REE sig- 1993), Pacific surface seawater (Piepgras and Jacobsen natures from bones of this unit show similar 1992; most analyses of REE in open ocean waters plot in patterns. This confirms that REE signatures in fos- this area), acid mine drainage (Worrall and Pearson 2001), sils are not affected by species or diet. Samples were shallow ground water (Dia et al. 2000), Sardinia (Biddau obtained from three different localities separated by et al. 2002). up to 16 km. This suggests that REE signatures may be somewhat laterally extensive in lacustrine sed- gram. The GdN/YbN ratio in various selected nat- iments. Sample locations are not near the probable ural waters as a function of pH is shown in figure margins of pluvial Fort Rock Lake (Allison 1966, 5. Although significant scatter occurs in the data, 1979), and signatures in fossils from marginal lake which may result from differences in concentration sediments could possibly differ. The REE signature of complex-forming ligands (particularly carbon- of a fossil bird eggshell (ϩ) is also shown in figure ate), the GdN/YbN ratio generally decreases with increasing pH. Alkalinity also increases, particularly in waters having pH values greater than about 7; thus, the decrease in GdN/YbN ratio at higher pH also reflects increased carbonate concentration and complexing of the HREE. Some terrestrial waters have pronounced positive or negative cerium anomalies (see examples in fig. 3). Negative cerium anomalies, such as that in seawater (fig. 3B), in which the Ce data point lies below the line connecting the La and Pr or Nd points, are commonly interpreted as resulting from oxidizing conditions (Wright et al. 1984, 1987; German and Elderfield 1990; Sholkovitz and Schneider 1991). Under oxidizing conditions, cerium may be oxidized from Ce3ϩ to Ce4ϩ and adsorbed or copre- cipitated with iron or manganese oxides or other phases. Positive anomalies are usually interpreted as indicating reducing conditions, under which Figure 5. Variation of GdN/YbN with pH in some rep- Ce4ϩ is reduced and released from solid phases into resentative natural waters. References as in figure 4. Journal of Geology FOSSIL LAKE TEPHROCHRONOLOGY 149

2002a) of this data shows that these REE signatures from the various units are signiﬁcantly different, indicating that these signatures can be used to determine or constrain fossil provenance.

Figure 8 shows an increase in YbN/GdN (HREE/ MREE) ratios up the lithostratigraphic section. Re- sults for different depositional packages are also

shown in a NdN-GdN-YbN ternary diagram (fig. 9). Samples from the lowest unit plot toward the cen- ter of the diagram, indicating a relatively flat signature with equal NASC-normalized concentrations of Nd, Gd, and Yb, as shown in figure 7. Samples from higher depositional packages plot to-

ward the YbN vertex, indicating progressive HREE- enrichment in the fossils. The relatively linear trend in the data suggests that the variety of signatures in the fossils results from a mixing or evo- Figure 6. Rare earth element (REE) signatures of vari- lutionary relationship between two end members, ous osteological materials from package 3 (unit 5) sam- one near package 1, having a relatively ﬂat REE pled at sites separated by up to 16 km (see table 2). The signature, and the other near package 8, having a similarity in these REE signatures suggests that they are strongly HREE-enriched signature. As shown in ﬁg- laterally similar within stratigraphic units of the Fossil ure 7, this trend is also accompanied by an increase Lake and that species and osteological material type has in the Ce anomaly. Because REEs in fossil bone little effect on REE ratios.

6. This sample has REE concentrations similar to the fossil bones (table 2), consistent with previous analyses of dinosaur eggshell and bone by Samoilov and Benjamini (1996). However, this signature dif- fers slightly from the others, having less HREE enrichment. This difference is not surprising, given the difference in mineralogy (calcite; Lowenstam and Weiner 1989), but suggests that REE signatures in eggshells may not provide results consistent with bone for taphonomic studies or paleoenvironmental studies. Therefore, results from eggshell have not been used in this study. In a few cases, some variation in LREE and MREE was also observed in REE signatures from trabecular material. This seems to be due to difficulty in cleaning matrix from the porous trabecular bone. Therefore, for this study, only cortical bone, enamel, and dentine, which produce consistent patterns, were used for stratigraphic comparisons. Although REE signatures were similar within depositional packages, they differed between units (fig. 7). The lower units have generally flat patterns, with no significant relative REE enrichment, whereas the upper units have much higher slopes to their signatures and relatively greater HREE enrichment. Lower units lack significant Ce anomalies, whereas upper units have positive Ce anom- Figure 7. Representative rare earth element (REE) sig- alies. The unique signatures in the fossil material natures in fossils from various units at Fossil Lake, act as fingerprints for each of the defined lithostrat- Oregon. A, Upper units (sedimentary packages 5–8); B, igraphic units. Discriminant analysis (Patrick et al. lower units (packages 1–4). 150 J. E. MARTIN ET AL.

waters in which units 5–8 were deposited were highly saline. Therefore, increased HREE enrichment in fossils coupled with positive Ce anomalies through the section indicate that waters of pluvial Fort Rock Lake evolved from circumneutral pH with relatively low salinity and, with minor fluctuations, became progressively more alkaline, basic, and saline over time. Fossils from the upper units have REE signatures similar to those of waters from some modern alkaline lakes, such as Lake Abert, Oregon, or Lake Van, Turkey (fig. 3B). Such changes in diagenetic water chemistry might result from increasing aridity due to progressive decrease in ef- fective precipitation in this area during the later Pleistocene. These suggestions are consistent with the faunal record, which exhibits a decrease in tax- onomic diversity in the upper units, as well as the Figure 8. Yb /Gd ratios and cerium anomalies in fos- N N presence of species, such as jackrabbits and sage sils from depositional packages at Fossil Lake, Oregon. Lines connect average values for each package or unit. grouse, adapted to more arid conditions. This interpretation may be compared with those of Allison and others for Fort Rock Lake, as well reflect the composition of lake bottom or pore wa- as those from other pluvial lakes in southern ters during fossilization, the patterns observed may Oregon. Allison (1966, 1979, 1982) observed a num- be related to changes in water composition and pa- ber of wave-cut lake beach terraces at different lev- leoenvironment during deposition of the Fossil els, which he attributed to stadial-interstadial lake Lake sediments. level fluctuations; however, he was not able to determine any relative or absolute dates for these terraces. After deposition of the uppermost salmonid- Discussion and Paleoenvironmental bearing layer, the lake became isolated and under- Interpretation went progressive desiccation (Allison and Bond

Circumneutral pH waters tend to have LREE- enriched or MREE-enriched or flat patterns, whereas basic and alkaline waters have HREE-enriched patterns (Johannesson et al. 1993; Kreamer et al. 1996; Johannesson and Xiaoping 1997). Thus, com- parison of REE signatures in fossils with those of modern waters suggests that Fort Rock Lake waters became progressively more basic and alkaline with time. REE signatures in fossils from the lower part of the section, packages 1–4, have no major Ce anomaly, whereas fossils from the upper part of the section may have large positive anomalies (fig. 8). Pos- itive Ce anomalies may arise from reducing redox conditions, due to the reduction of Ce4ϩ to Ce3ϩ and release from sediments or to stabilization of the soluble Ce-pentacarbonate complex in highly alkaline waters (Johannesson et al. 1993; Mo¨ ller and Bau 1993). There is no sedimentary evidence, such as the occurrence of organic-rich layers, py- rite, or other sulfides, that lake waters from the upper units were strongly reducing; therefore, the Figure 9. Ternary diagram showing variations of rare occurrence of positive Ce anomalies suggests that earth elements in lithologic units at Fossil Lake, Oregon. Journal of Geology FOSSIL LAKE TEPHROCHRONOLOGY 151

1983). Greenspan (1984) suggested that lake waters more highly saline during deposition of package 6 had essentially completely disappeared in the Fossil and the lower part of package 7 (unit 13), less saline Lake Basin by about 10,000 yr ago. Lake waters may during deposition of the upper part of package 7 have then reappeared, at least ephemerally, during (unit 14), and returning to highly saline conditions Neoglacial or other higher rainfall periods (Smith in package 8, before final desiccation of the lake. 1926; Allison and Bond 1983). These interpreta- Such salinity variations are broadly similar to those tions could be consistent with a long-term shift seen by Cohen et al. (2000) in Lake Chewaucan; from neutral to alkaline waters with time but pro- however, given the low resolution of REE sampling vide little additional detail. Palacios-Fest et al. from this study, it is difficult to determine precise (1993), Negrini et al. (2000), and Cohen et al. (2000) stratigraphic correlations between the REE varia- have produced detailed paleoenvironmental rec- tions at Fossil Lake and the reported salinity var- ords for Summer Lake, Oregon, the remnant (with iations at Summer Lake. High-resolution paleocli- Lake Abert) of pluvial Lake Chewaucan (Allison mate studies, similar to those of Negrini et al. 1982; Davis 1985), which covered an area just to (2000) and Cohen et al. (2000), should be under- the south of the pluvial Fort Rock Lake, of which taken at Fossil Lake to determine possible corre- the Fossil Lake Basin forms the easternmost part. lations between these two basins. Also, further lat- The exposed stratigraphic interval at Summer Lake eral studies need to be conducted to determine (Lake Chewaucan) covers the time period from whether distinguishable variations in the REE pat- about 250,000 yr ago to the present (Negrini et al. terns from west to east can be seen as a result of 2000; Kuehn and Foit 2001), generally equivalent the variations due to proximity of the fossil ma- to the middle and upper portions (above deposi- terial to the margins of the lake. tional package 3; fig. 2) of the Fossil Lake sequence. Paleosalinity estimates for Summer Lake based Conclusions on ostracode assemblages suggest moderate to high salinities from 230 to 190 ka, decreasing salinities Fossil Lake, Oregon, has been studied extensively between 190 and 165 ka, and then increasing sa- over the last century and is well known for its abun- linities between 165 and 162 ka, below the base of dance of Pleistocene fossils; however, reliable dat- a major unconformity (Cohen et al. 2000). Gener- ing and stratigraphic correlations were lacking. The ally very highly saline waters are indicated between original dates did not agree with the preservation ca. 102 and 80 ka and between 50 and 30 ka, sep- of the fossil assemblages, and the repeated packages arated by a period of low salinity (Cohen et al. 2000, of sands and gravels grading into clays had not been their figs. 8, 9, and 17). Late Pleistocene and Ho- previously recognized. This study establishes a lith- locene waters were apparently highly variable in ostratigraphic framework consisting of at least nine salinity, with Holocene waters being highly saline exposed rhythmic depositional packages with ap- (Cohen et al. 2000). At present, Summer and Abert proximate ages based on five interbedded tephras. Lakes are highly saline and have basic pH and al- The tephra layers, which include the Rye Patch kaline waters (Van Denburgh 1975; Johannesson et Dam (∼646 ka), Dibekulewe (∼610 ka), Tulelake al. 1993; Palacios-Fest et al. 1993; Cohen et al. T64 (∼95 ka), Marble Bluff (47 ka), and the Trego 2000). Johannesson et al. (1993) measured REE sig- Hot Springs (23.2 ka), indicate that the deposits at natures of waters from both of these modern lakes Fossil Lake are much older than previously pro- (fig. 3B). It is likely that, given the similarity in posed and reveal the ephemeral nature of the Late geologic features and stream and spring inputs, wa- Pleistocene lakes in Oregon during the Irvingtonian ters of pluvial Fort Rock Lake would have followed and Rancholabrean NALMAs. a similar evaporitic evolutionary trend (Hardie and The REE signatures of fossil bones are consistent Eugster 1970). Therefore, it seems likely that high- within the defined units but vary vertically be- salinity periods in Fort Rock Lake would also pro- tween different units. Therefore, REE signatures duce highly alkaline and high pH waters. may be used to determine or constrain fossil prov- Ce anomalies in fossils from the various litho- enance. Consistency of REE signatures from vari- logic units (fig. 8) may record salinity variations, ous fossil types indicates that the REE signature is such as those found at Summer Lake (Cohen et al. not dependent on diet or taxon. REE signatures in 2000). If the positive Ce anomalies result from high fossils from the lower units were relatively flat; salinities and alkalinities rather than redox varia- comparisons with REE patterns in modern waters tions, as discussed previously, lake waters may suggest near-neutral pH low alkalinity waters. REE have had low salinity during deposition of packages signatures in fossils from the upper units of Fossil 1–5. Thereafter, salinities fluctuated, becoming Lake are greatly HREE-enriched and have positive 152 J. E. MARTIN ET AL.

Ce anomalies (fig. 7), similar to signatures from ogist, Lakeview District, Bureau of Land Manage- extant, highly saline alkaline lakes, such as Lake ment, has been particularly instrumental in over- Abert or Van (fig. 3B). These signatures can be ex- seeing the long-term project to completion. R. plained by increasing aridity through the Pleisto- Hanes of the Oregon State Office, Bureau of Land cene, leading to an increase in alkalinity of the lake. Management, who is in overall charge of the pa- This shift in water chemistry in the upper units is leontology of the area, has guided our project over coincident with a general decrease in diversity of the years. Many students and colleagues have co- fossil taxa and those adapted to more xeric condi- operated in field studies during the past 15 years, tions, further corroborating the geochemical data. particularly W. and B. Harrold, who have partici- pated in nearly every field expedition. We thank P. ACKNOWLEDGMENTS Field and L. Bolge for aid with the ICP-MS analyses. This research was supported through financial as- The manuscript was improved by reviews by R. sistance from the United States Bureau of Land Feranec, A. Sarna-Wojcicki, and C. Trueman. We Management, Temple University, and the Desert particularly thank A. Sarna-Wojcicki for sharing Research Institute. W. Cannon, District Archaeol- unpublished age estimates of certain tephras.

REFERENCES CITED

Allison, I. S. 1940. Study of Pleistocene lakes of south- mentology of the coarse-grained deposits from the central Oregon. Carnegie Inst. Year Book 39:299–300. Quaternary Fossil Lake area in south-central Oregon. ———. 1966. Fossil Lake, Oregon: its geology and fossil Proc. S. D. Acad. Sci. 76:273–274. faunas. Oreg. State Monogr. Stud. Geol. 9:1–48. Biddau, R.; Cidu, R.; and Frau, F. 2002. Rare earth ele- ———. 1979. Pluvial Fort Rock Lake, Lake County, ments in waters from the albitite-bearing granodior- Oregon. State of Oregon, Dept. Geol. Miner. Indust. ites of central Sardinia, Italy. Chem. Geol. 182:1–14. Spec. Pap. 7:1–72. Borchardt, G. A.; Aruscavage, P. J.; and Millard, H. T., Jr. ———. 1982. Geology of pluvial Lake Chewaucan, Lake 1972. Correlation of the Bishop Ash, a Pleistocene County, Oregon. Oreg. State Monogr. Stud. Geol. 11: marker bed, using instrumental neutron activation 1–79. analysis. J. Sediment. Petrol. 42:301–306. Allison, I. S., and Bond, C. E. 1983. Identity and probable Brophy, G. P., and Hatch, T. M. 1962. Recrystallization age of salmonids from surface deposits at Fossil Lake, of fossil horse teeth. Am. Mineral. 47:1174–1180. Oregon. Copeia 1983:563–564. Byrne, R. H., and Kim, K. H. 1993. Rare earth precipi- Armstrong, H. A.; Pearson, D. G.; and Griselin, M. 2001. tation and coprecipitation behavior: the limiting role 3Ϫ Thermal effects on rare earth element and strontium ofPO4 on dissolved rare earth concentrations in sea- isotope chemistry in single conodont elements. Geo- water. Geochim. Cosmochim. Acta 57:519–526. chim. Cosmochim. Acta 54:435–441. Byrne, R. H., and Sholkovitz, E. R. 1996. Marine chem- Bell, C. J.; Lundelius, E. L.; Barnosky, A. D.; Graham, R. istry and geochemistry of the lanthanides. In Gschnei- W.; Lindsay, E. H.; Ruez, K. R., Jr.; Semken, H. A., Jr.; der, K. A., Jr., and Eyring, L., eds. Handbook on the Webb, S. D.; and Zakrzewski, R. J. 2004. The Blancan, physics and chemistry of rare earths 21:497–593. New Irvingtonian, and Rancholabrean mammal ages. In York, Elsevier. Woodburne, M. O., ed. Late Cretaceous and Cenozoic Cohen, A. S.; Palacios-Fest, M. R.; Negrini, R. M.; Wi- mammals of North America: biostratigraphy and geo- gand, P. E.; and Erbes, D. B. 2000. A paleoclimate rec- chronology. New York, Columbia University Press, ord for the past 250,000 years from Summer Lake, 391 p. Oregon, USA. II. Sedimentology, paleontology, and Benson, L. V.; Smoot, J. P.; Kashgarian, M.; Sarna-Wojci- geochemistry. J. Palaeolimnol. 24:151–182. cki, A.; and Burdet, J. W. 1997. Radiocarbon ages and Cope, E. D. 1883. On the ﬁshes of the Recent and Plio- environments of deposition of the Wono and Trego cene lakes of the western part of the Great Basin, and Hot Springs tephra layers in the Pyramid Lake sun- of the Idaho Pliocene lake. Proc. Natl. Acad. Sci. Phil- basin, Nevada. Quat. Res. 47:251–260. adelphia, p. 134–166. Berger, G. W., and Busacca, A. J. 1995. Thermolumines- Davis, J. O. 1978. Quaternary tephrochronology of the cence dating of Late Pleistocene loess and tephra from Lake Lahontan area, Nevada and California. Univer- eastern Washington and southern Oregon and impli- sity of Nevada, Reno. Archaeol. Surv. Res. Pap. 7, 137 cations for the eruptive history of Mount St. Helens. p. J. Geophys. Res. 100:22,361–22,374. ———. 1985. Correlation of late Quaternary tephra lay- Bernat, M. 1975. Les isotopes de l’uranium et du thorium ers in a long pluvial sequence near Summer Lake, et les terres rares dans l’environment marin. Cah. Oregon. Quat. Res. 23:38–53. O.R.S.T.O.M. Ser. Geol. 7:65–83. Dia, A.; Gruau, G.; Olive-Lauquet, G.; Riou, C.; Molenat, Bertog, J. L.; Martin, J. E.; and Kihm, A. J. 1997. Sedi- J.; and Curmi, P. 2002. The distribution of rare earth Journal of Geology FOSSIL LAKE TEPHROCHRONOLOGY 153

elements in groundwaters: assessing the role of elements in natural terrestrial waters and the signif- 3Ϫ 2Ϫ source-rock composition, redox changes, and colloidal icance ofPO43 and CO in limiting dissolved rare particles. Geochim. Cosmochim. Acta 64:4131–4151. earth concentrations; a review of recent information. Elderfield, H., and Pagett, R. 1986. REE in ichthyoliths: Aquat. Geochem. 1:157–173. variations with redox conditions and depositional en- Johannesson, K. H.; Lyons, W. B.; Yelken, M. A.; Gau- vironment. Sci. Total Environ. 48:175–197. dette, H. E.; and Stetzenbach, K. J. 1996. Geochemistry Elderfield, H.; Upstill-Goddard, R.; and Sholkovitz, E. R. of the rare-earth elements in hypersaline and dilute 1990. The rare earth elements in rivers, estuaries, and acidic natural terrestrial waters: complexation behav- coastal seas and their significance to the composition ior and middle rare-earth element enrichments. of ocean waters. Geochim. Cosmochim. Acta 54:971– Chem. Geol. 133:125–144. 991. Johannesson, K. H., and Xiaoping, Z. 1997. Geochemistry Elftman, H. O. 1931. Pleistocene mammals of Fossil of the rare earth elements in natural terrestrial waters: Lake, Oregon. New York, Am. Mus. Nat. Hist., Am. a review of what is currently known. Chin. J. Geo- Mus. Novitates 481:1–21. chem. 16:20–42. Field, M. P., and Sherrell, R. M. 1998. Magnetic sector Johannesson, K. H., and Zhou, X. 1999. Origin of middle ICPMS with desolvating micronebulization: interfer- rare earth element enrichments in acid waters of a ence-free subpicogram determination of rare earth el- Canadian High Arctic lake. Geochim. Cosmochim. ements in natural samples. Anal. Chem. 70:4480– Acta 63:153–165. 4486. Johannesson, K. H.; Zhou, X.; Guo, C.; Stetzenbach, K. German, C. R., and Elderfield, H. 1990. Application of J.; and Hodge, V. F. 2000. Origin of rare earth element the Ce-anomaly as a paleoredox indicator: the ground signatures in groundwaters of circumneutral pH from rules. Paleoceanography 5:823–833. southern Nevada and eastern California, USA. Chem. Goslar, T., and Paxlut, M. F. 1985. Contamination studies Geol. 164:239–257. on mollusk shell samples. Radiocarbon 27:33–42. Kemp, R. A., and Trueman, C. N. 2003. Rare earth ele- Greenspan, R. L. 1984. Holocene lake level fluctuations ments in Solenhofen biogenic apatite: geochemical inferred from fish remains in the Fort Rock Basin, clues to the paleoenvironment. Sediment. Geol. 155: Oregon. Tebiwa 21:50–55. 109–127. Gromet, L. P.; Dymek, R. F.; Haskin, L. A.; and Korotev, Kreamer, D. K.; Hodge, V. F.; Rabinowitz, I.; Johannesson, R. L. 1984. The North American Shale Composite: its K. H.; and Stetzenbach, K. J. 1996. Trace element geo- composition, major, and trace element characteristics. chemistry in water from selected springs in Death Geochim. Cosmochim. Acta 48:2469–2482. Valley National Park, California. Ground Water 34: Haas, J. R.; Shock, E. L.; and Sassani, D. C. 1995. Rare 95–103. earth elements in hydrothermal systems: estimates of Kuehn, S. C., and Foit, F. F., Jr. 2001. Updated tephra standard partial molal thermodynamic properties of stratigraphy at Summer Lake, Oregon, a sub-basin of aqueous complexes of the rare earth elements at high pluvial Lake Chewaucan. In Guidebook for the pressures and temperatures. Geochim. Cosmochim. Friends of the Pleistocene Ninth Annual Pacific Acta 59:4329–4350. Northwest Cell Field Trip. SK1–9. http://geology Hallett, D. J.; Mathewes, R. W.; and Foit, F. F., Jr. 2001. .csustan.edu/kuehn/newberry/fop/fop2001.pdf Mid-Holocene Glacier Peak and Mount St Helens We Lanphere, M. A.; Champion, D. E.; Clynne, M. A.; Low- tephra layers detected in lake sediments from south- enstern, J. B.; Sarna-Wojcicki, A. M.; and Wooden, J. ern British Columbia using high resolution tech- L. 2004. Age of the Rockland tephra, western USA. niques. Quat. Res. 55:284–292. Quat. Res. 62:94–104. Hannigan, R. E., and Sholkovitz, E. R. 2001. The devel- Lowenstam, H. A., and Weiner, S. 1989. On biominer- opment of middle rare earth element enrichments in alization. Oxford, Oxford University Press, 324 p. freshwaters: weathering of phosphate minerals. Martin, J. E. 1996. First occurrence of Cynomys from Chem. Geol. 175:495–508. west of the Rocky Mountains. J. Soc. Vertebr. Paleo. Hardie, L. A., and Eugster, H. P. 1970. The evolution of 16:51A. closed-basin brines. Mineral. Soc. Am. Spec. Pap. 3: Martin, J. E., and Kihm, A. J. 2003. Geochronology of the 273–290. Pleistocene tephra sequence at Fossil Lake, Oregon. Henderson, P.; Marlow, C. A.; Molleson, T. I.; and Wil- Cordilleran Sec., Geol. Soc. Am. Program Abstr. 35: liams, C. T. 1983. Patterns of chemical change during 21. bone fossilization. Nature 306:358–360. Mo¨ ller, P., and Bau, M. 1993. Rare-earth patterns with Howard, H. 1946. A review of the Pleistocene birds of positive cerium anomaly in alkaline waters from Lake Fossil Lake, Oregon. Carnegie Inst. Publ. 551:143–195. Van, Turkey. Earth Planet. Sci. Lett. 117:671–676. Johannesson, K. H.; Lyons, W. B.; and Bird, D. A. 1993. Negrini, R. M.; Erbes, D. B.; Faber, K.; Herrera, A. M.; Rare earth element concentrations and speciation in Roberts, A. P.; Cohen, A. S.; Wigand, P. E.; and Foit, alkaline lakes from the Western U.S.A. Geophys. Res. F. F., Jr. 2000. A paleoclimate record for the past Lett. 21:773–776. 250,000 years from Summer Lake, Oregon, U.S.A. I. Johannesson, K. H.; Lyons, W. B.; Stetzenbach, K. J.; and Chronology and magnetic proxies for lake level. J. Pa- Byrne, R. H. 1995. The solubility control of rare earth laeolimnol. 23:125–149. 154 J. E. MARTIN ET AL.

Palacios-Fest, M. R.; Cohen, A. L.; Ruiz, J.; and Blank, B. tional material collected there. Bull. Am. Mus. Nat. 1993. Comparative paleoclimatic interpretations from Hist. 32:123–178. nonmarine ostracodes using faunal assemblages, trace Smith, W. D. 1926. Physical and economic geology of element shell chemistry and stable isotope data. In Oregon: the southeastern lake province. Oregon Uni- Swart, P. K.; Lohman, K. C.; McKenzie, J.; and Savin, versity, Commonw. Rev. 8:199–253. S., eds. Climate change in continental isotopic rec- Snyder, C. T.; Hardman, G.; and Zdenek, F. F. 1964. Pleis- ords. Am. Geophys. Union Geophys. Monogr. 78:179– tocene lakes in the Great Basin. U.S. Geol. Surv. Misc. 190. Invest. Map I-416, scale 1 : 1,000,000. Patrick, D.; Martin, J. E.; and Grandstaff, D. E. 2002a. Staron, R. M.; Grandstaff, B. S.; Gallagher, W. B.; and Application of rare earth element (REE) signature anal- Grandstaff, D. E. 2001. REE signatures in vertebrate ysis at Pleistocene Localities in Fossil Lake, Oregon. fossils from Sewell, NJ: implications for location of Geol. Soc. Am., Abstr. Program 34:A-57. the K-T boundary. PALAIOS 16:255–265. Patrick, D.; Martin, J. E.; Parris, D. C.; and Grandstaff, Stetzenbach, K. J.; Hodge, V. F.; Guo, C.; Farnham, I. M.; D. E. 2002b. Rare earth element signatures of fossil and Johannesson, K. H. 2001. Geochemical and sta- vertebrates compared with lithostratigraphic subdi- tistical evidence of deep carbonate groundwater visions of the upper Cretaceous Pierre Shale, central within overlying volcanic rock aquifers/aquitards of South Dakota. Proc. S. D. Acad. Sci. 81:161–179. southern Nevada. J. Hydrol. 243:254–271. ———. 2004. Paleoenvironmental interpretations of rare Tang, J., and Johannesson, K. H. 2003. Speciation of rare earth element signatures in mosasaurs (Reptilia) from earth elements in natural terrestrial waters: assessing the upper Cretaceous Pierre Shale, central South Da- the role of dissolved organic matter from the modeling kota, USA. Palaeogeogr. Palaeoclimatol. Palaeoecol. approach. Geochim. Cosmochim. Acta 67:2321–2339. 212:277–294. Tanners, M. A. 1970. The validity of radiocarbon dates Patrick, D.; Terry, D. O., Jr.; and Grandstaff, D. E. 2001. on terrestrial snail shells. Am. Antiq. 35:94–100. The influence of osteological material and time on Trueman, C. N. 1999. Rare earth element geochemistry rare earth element (REE) concentrations in vertebrate and taphonomy of terrestrial vertebrate assemblages. fossils. J. Vertebr. Paleontol. 21:88A. PALAIOS 14:555–568. Person, A.; Bocherens, H.; Salie`ge, J.-F.; Paris, F.; Zeltoun, Trueman, C. N., and Tuross, N. 2002. Trace elements in V.; and Ge´rard, M. 1995. Early diagenetic evolution of Recent and fossil bone apatite, in phosphates: geo- bone phosphate: an x-ray diffractometry analysis. J. chemical, geobiological, and materials importance. Archaeol. Sci. 22:211–221. Am. Mineral. Soc. Rev. Mineral. Geochem. 48:1–742. Piepgras, D. J., and Jacobson, S. 1992. The behavior of Tuross, N.; Behrensmeyer, A. K.; and Eanes, E. D. 1989a. rare earth elements in seawater; precise determination Strontium increases and crystallinity changes in taph- of variations in the North Pacific water column. Geo- onomic and archaeological bone. J. Archaeol. Sci. 16: chim. Cosmochim. Acta 56:1851–1862. 661–672. Reynard, B.; Lećuyer, C.; and Grandjean, P. 1999. Crystal- Tuross, N.; Behrensmeyer, A. K.; Eanes, E. D.; Fisher, L. chemical controls on rare-earth element concentra- W.; and Hare, P. E. 1989b. Molecular preservation and tions in fossil biogenic apatites and implications for crystallographic alterations in a weathering sequence paleoenvironmental reconstructions. Chem. Geol. of wildebeest bones. Appl. Geochem. 3:261–270. 155:233–241. Van Denburgh, A. S. 1975. Solute balance at Abert and Rieck, H. J.; Sarna-Wojcicki, A. M.; Meyer, C. E.; and Summer Lakes, south-central Oregon. U.S. Geol. Surv. Adam, D. P. 1992. Magnetostratigraphy and tephro- Prof. Pap. 502-C:1–29. chronology of an upper Pliocene to Holocene record Waring, G. A. 1908. Geology and water resources of a in lake sediments at Tulelake, northern California. portion of south-central Oregon. U.S. Geol. Surv. Wa- Geol. Soc. Am. Bull. 40:409–428. ter Supply Pap. 220:1–86. Russell, I. C. 1884. A geological reconnaissance in south- Weiner, S., and Traub, W. 1992. Bone structure: from ern Oregon. U.S. Geol. Surv. Annu. Rep. 4:431–464. a˚ngstroms to microns. FASEB J. 6:879–885. Samoilov, V. S., and Benjamini, C. 1996. Geochemical Williams, S. K. 1994. Late Cenozoic tephrastratigraphy features of dinosaur remains from the Gobi Desert, of deep sediment cores from the Bonneville Basin, South Mongolia. PALAIOS 11:519–531. northwest Utah. Geol. Soc. Am. Bull. 106:1517–1530. Sarna-Wojcicki, A. M.; Negrini, R. M.; Adam, D. P.; Wal- Wood, S. A. 1990. The aqueous geochemistry of the rare- ker, J. P.; and Wan, E. 2001. The Anna River core: earth elements and yttrium. I. Review of available extending the record of the Ana River section back in low-temperature data for inorganic complexes and the time. An addendum to the Guidebook for the Friends inorganic REE speciation of natural waters. Chem. of the Pleistocene Ninth Annual Pacific Northwest Geol. 82:159–186. Cell Field Trip. SK1–9. Worrall, F., and Pearson, D. G. 2001. The development Sholkovitz, E. R., and Schneider, D. L. 1991. Cerium re- of acidic groundwaters in coal-bearing strata. I. Rare dox cycles and rare earth elements in the Sargasso Sea. earth element fingerprinting. Appl. Geochem. 16: Geochim. Cosmochim. Acta 55:2737–2743. 1465–1480. Shufeldt, R. W. 1913. Review of the fossil fauna of the Wright, J.; Schrader, H.; and Holser, W. T. 1987. Paleo- desert region of Oregon, with a description of addi- redox variations in ancient oceans recorded by rare Journal of Geology FOSSIL LAKE TEPHROCHRONOLOGY 155

earth elements in fossil apatite. Geochim. Cosmo- D. L., ed. Conodont biofacies and provincialism. Geol. chim. Acta 51:631–644. Soc. Am. Spec. Pap. 196:325–340. Wright, J.; Seymour, R. S.; and Shaw, H. F. 1984. REE and Yates, T. J. S. 1986. Studies of non-marine mollusks for Nd isotopes in conodont apatite: variations with geo- the selection of shell samples for radiocarbon dating. logical age and depositional environment. In Clark, Radiocarbon 28:457–463.