Patrick K. Spencer Departmentof Geology,Whitman College, Walla Walla, 99362

A Small MammalFauna from the Touchet Beds of Walla Walla County, Washington:Support for the Multiple-FloodHypothesis

Abstract

Recent work in southeastern Washington has challenged the long-accepted hypothesis that the late Pleistocene Touchet Beds were deposited by one or a few collosal glacial outburst floods from glacial Lake Missoula in western Montana. Work conducted for this study in the northern portion of the Walla Walla Valley has revealed the presence of a diverse and exceedingly well- preserved small-mammal fauna from the middle portion of these flood-related sediments. The remains are preserved with fragile bone elements and dentition intact and virtually no evidence for post-mortem transport or reworking. Biogenic structures inter- preted as rodent burrows are prevalent throughout the section. Sedimentary structures and textures preserved at the fossil collection site suggest that the sediments were deposited under conditions of considerable turbulence, and that there were periods of time between depositional episodes characterized by erosion and colonization by small mammals. The data are overwhelm- ingly supportive of a hypothesis which invokes multiple flood episodes separated in time by decades-long periods of exposure for deposition of the Touchet Beds. lntroduction Miocene Basalt Group and cover- ing the Pliocene and earlier Pleistocene valley The Touchet Beds (late Pleistocene) are widespread in the Walla Walla Valley of south- fill deposits.The Touchet Beds and associated eastern Washington (Figure l), mantling the erosional topography of the Channeled Scablands

Figure l. Index map of the Walla Walla Valley showing approximate limits of distribution of the Late Pleistocene Touchet Beds (hatch marks), and location of study sections. Section A represents the section, section B the Burlingame Canyon section. After Biornstadt (1980).

NorthwestScience, Vol. 63, No.4, 1989 167 have long been recognized as having resulted unusually good preservation (Figure 2). This from flood waters released from the ice-dammed specimen was recovered from a clastic dike ap- glacial Lake Missoula in western Montana. Ex- proximately I m below a thin tephra layer. Bjorn- posuresof the Touchet Beds in the PascoBasin stadt (1980) analyzed,a tephra at a nearby locali- and Walla Walla Valley of southeastern Wash- ty, which occurred at the same apparent strati- ington comprise a distinctive sequenceof graded graphic level and exhibited similar structures. He rhythmites laid down as water backed up behind found this tephra to fall within the Mt. St. Helens 'S' a hydraulic dam at , the only outlet set chemical group. The Mt. St. Helens set 'S' for water entering the region (Figure l). tephra has been dated at approximately The mechanismsresponsible for deposition 13,000 ybp (Mullineaux and Crandell l98l). The dike of the Touchet Beds have been argued for some appearsto originate from below the collec- tion locality, time, and from this argument, two schools of and thus the skull most likely pre- thought have emerged. The first interprets the dates the tephra, which is here approximately 3 Touchet Beds as the result of a single massive m below the modern topographic surface. The flood or a few floods, with hydraulic surging of base of the Touchet Beds is not exposedin the the floodwaters giving rise to the characteristic Walla Walla Valley. At Burlingame Canyon (sec- graded rhythmites (Bretz 1923,1925, 1930, 1969; tion B, Figure l) the section extends for tens of 'S' Bretz et al.1956; Baker 1978; Carson et al. l97B: meters below the St. Helens set tephra. It is, Waitt, 1978; Bjornstadt 1980).The second con- thus, likely that the original depositional local- siders the Touchet Beds as having been deposited ity for the skull is above the base of the Touchet 'S' by as many as forty or more separate floods, each Beds and below the St. Helens tephra. giving rise to a graded rhythmite, separated in Of particular interest is the state of preser- time by decades-longperiods of exposure.This vation of the skull. All of the dentition is intacr view of the Touchet Beds was pioneered by Waitt (Figure 3a). The zygomatic arches, likewise, are (1980,1984, 1985a, 1985b), and supportedby At- preserved without damage (Figure 3a). The water (1984). auditory bullae, which are delicate features on recent skull This paper will examine the vertebrate fauna material, are intact (Figure 3b) and pterygoid and associated biogenic structures recovered the processes(Figure 3b) are preserved with minimal from an exposure of the Touchet Beds in the damage. Finally, the skull exhibits virtually northern Walla Walla Valley. In addition, the no evidence for abrasion as a result of post-mortem sedimentologic characteristics observed at the transport. Other cranial elements outcrop will be discussedand interpreted in light are not intact, but had disintegrated in outcrop. of the prevailing ideas regarding depositional Post-cranial remains exhibit excellent preserva- mode of the Touchet Beds. tion, with no evidence for abrasion. If the Touchet Beds were the result of a single VertebrateFauna flood, or few floods characterized by hydraulic surging, then the presence of well-preserved, Vertebratefossils belonging to the Order Ro- unabraded mammalian remains from the middle dentiahave been recoveredfrom the Touchet portion of the deposit is problematical. If the River section(section A, Figure l). Represented re- mains represent members of the local fauna in the collectionto date are membersof the fam- en- tombed by the initial flood surge, they might be ilies Sciuridae (ground sguirrel, 2 skulls), preserved in their present Cricetidae(vole, I skull),Heteromyidae (kangaroo condition, but at or near the baseof the deposit,If the remains were rat, partial skull),and post-cranialremains of at transported into the area by a later flood surge, least four other individuals from the Order they might be preserved in the middle of the Rodentia. Species-leveltaxonomy with such a deposit, but would certainly exhibit evidence paucity of material is unproductive;however, of transport, such as abraded surfaces and/or preliminary study of the fossilsshows them to broken features. The best explanation for the oc- be indistinguishablefrom specieswhich now in- currence of a well-preserved habit the region. mammalian fauna in the middle of the Touchet Beds is that it One of the sciurid skullsin particular is wor- represents the remains of an established local thy of special considerationbecause of its fauna which experienced minimal (or no)

168 Spencer Figure 2. Sciurid skull recovered from a clastic dike in the Touchet River section of the Touchet Beds. A: dorsal viewl B: ven- tral view; C: left lateral oblique view.

A Small Mammal Fauna from the Touchet Beds r69 '.;:';*.*l&''

Figure 3. A: Upper lefi dentition and zygomaticarch of sciurid (ground squirrel)skull. B: Auditory butlae(l) and prerygoid process(2) of sciurid skull. Specimenfrom Touchet River section.Scale bar equals0.5 cm.

170 Spencer transport during a distinct flood episode.This shows relatively tight grouping of all data points, is consistent with other physical and biogenic and poor separationoflower, middle, and upper evidencepresented in the following section and rhythmites. The graphic data supports the field is supportive of the hypothesis which invokes observation that rhythmites are poorly developed multiple flood episodes with decades-longex- in the Touchet River section as compared to the posure between floods. Burlingame Canyon section. Sedimentary Structures: Primary sedimentary Sedimentology structuresobservable in outcrops can be usedto Two well-exposedsections of Touchet Beds were interpret the conditions of deposition in the two sampled for this study (seelocation map, Figure study sections.The Burlingame Canyon section l). Stratigraphic control within the sections is shows,from base to top within each rhythmite, provided by the occurrenceof Mt. St. Helensset a transition from planar bedding (lower-upper 'S' tephra at the top of each sampled interval. flow regime transition) to ripples and climbing The Burlingame Canyon section(section B, Fig- ripples, and finally to massivebedding reflecting ure l) representswhat is consideredto be typical deposition from suspensionand/or eolian deposi- ofthe Touchet Beds in this region. The Touchet tion. These structures suggest an initial flood River section (section A, Figure l), from which surge and deposition of a traction load followed the fossilswere recovered,represents somewhat by rapidly decreasing energ'y.In addition, paleo- different sedimentologic conditions as inter- current directions are consistentlyup-valley for preted from differences in grain size parameters, the lower to middle portions of rhythmites. The physicalsedimentary structures, and geographic Touchet River section, while displaying mod- location. erately well-developed ripples and cross-stratifi- cation, shows no clearly definable transition in The Burlingame Canyon section, where there flood energy within rhythmites. Planar beds were are as many as 40 rhythmites exposed, displays not noted, and massivebedding at the tops of well-developed graded bedding within each rhythmites was irregularly preserved.In addition, rhythmite, with the basal sand giving way upward ripples did not show a consistentorientation, sug- to silt representingstanding water, and in some gesting that paleocurrentdirections were highly cases,a thin veneer of massivesilt capping the variable. rhythmite. Mean grain size for samplesrepresent- Biogenic sedimentary structures, primarily ing the base, middle, and top of rhythmites re- small-diameter cylindrical features attributed by flects this fining-upward trend (Figure 4, lower Waitt (1980)to burrowing activity of rodents, are right). Values for standard deviation (a measure prevalent throughout both sample sections. of sorting) show correspondingvariation within Alcorn (1940) has indicated that modern rodents each rhythmite (Figure 4, upper right). Samples burrow to a maximum depth of 2 m below ground from the Touchet River section do not show the surface. The presence of burrows throughout same trends in grain size distribution. Mean thick sectionsof the Touchet Beds suggeststhat grain sizes generally show low or erratic variabil- they are a result of burrowing which was contem- ity within rhythmites as well as consistently poraneouswith deposition,rather than a result smaller mean grain size (Figure 4, lower left). of the burrowing activity of modern rodents. In Similarly, sorting values are considerablymore addition, Waitt (1980) pointed out that the consistent,again with little apparent trend within burrow-filling sediment is compacted to a degree rhythmites (Figure 4, upper left). In addition, equivalent to that of the surrounding sediment, rhythmites from the Touchet River section do not suggestingcontemporaneous compaction of the have clear upper and lower boundaries and were Touchet Beds and the burrow-filling sediment. difficult to distinguish visually while in the field. 'S' The St. Helens set tephra is extensivelyper- Plots of standard deviation (sorting) versus forated by small diameter (less than I cm) mean grain size are shown in Figure 5. Samples features,which Spencer(1987) suggested might from the Burlingame Canyon section (Figure 5, be insect burrows and which may indicate ex- right) allow reasonablegrouping of data points posure or near exposure subsequentto deposi- into lower, middle, and upper rhythmites. The tion of the tephra. An alternative explanation for Touchet River section (Figure 5, left), by contrast, these featureswas presentedby Bjornstadt (1980):

A Small Mammal Fauna from the Touchet Beds l7l 11 10 x I x 8 x ut x o- 6 X E 5 x ttl x x 4 3 x x 2 x a '( I I 2 SD,PHI SD, PHI

x tJ'J x o- x = Lrt

3 4 MEAN,PHI MEAN,PHI TOUCHETRIVER BURTINGAMECANYON

Figure 4. Plots of StandardDeviation versus sample (upper graphs)and Mean grain sizeversus sample (lower graphs) for the Touchet River section(left) and the BurlingameCanyon section (right). Solid horizontallines representrhythmite boundaries.Mount St. Helens set 'S' tephra locatedat top of each sampledinterval. he proposed that they represent hydroplastic ness and purity of the tephra suggests a valley injection structures. which was drained or nearly drained at the time of deposition. The tephra itself may be used to suggestex- posure or near exposure at the time of its deposi- The sedimentologic data is best explained as tion. It is difficult to envision an airfall tephra a result of the size and location of tributary accumulating to a thickness of a few centimeters valleys(Touchet River section)with respectto the if the valley was occupied by deep, turbulent Walla Walla Valley (Burlingame Canyon section). waters resulting from a single flood. The thick- The center of the Walla Walla Valley (section B,

172 Spencer - I o- z

llt = a SD,PHI TOUCHETRIVER BURLINGAMECANYON

Figure 5. Plots of Standard Deviation versus Mean grain size for the Touchet River section (left) and the Burlingame Canyon section (right). Samples grouped according to position within rhythmite (+: upper; x: middle; o: lower).

Figure l) would be expected to experience the which was destroyed during a distinct flood most variability in conditions during advancing episode. floods. In these areas,sedimentary structures and statistical parameters of sediments reflect rapidly Sedimentologic parameters for samples from changing energy conditions during flooding. two measuredsections of the Touchet Beds sug- Tributary valleys (Touchet River section, section gest that the conditions of deposition of these A, Figure l) would be expectedto be more con- sediments were variable dependent upon tinuously turbulent (conditions more uniform) as geographic location within the Walla Walla a result of the fact that they are both shallower Valley region. The Burlingame Canyon section, and narrower than the main valley. In addition, located in the center of the Walla Walla Valley, the higher elevation tributary valleys would be was characterized by rapidly decreasing energy the first to drain after a flood episode, and would during flood episodes, which resulted in well- experiencemore intense post-flooderosion, and developed graded rhythmites. The Touchet River thus removal of upper rhythmites, as a result of section,located in a narrower, shallower tributary their higher gradient. Observed structures and valley, was characterized by uniformly turbulent statisticalparameters for the Touchet River sec- conditions during deposition, which resulted in tion reflect these phenomena. poorly-developedgraded rhythmites. In addition, the higher elevationand steepergradient of the Conclusions tributary valley gave rise to more rapid drain- Small-mammal fossils recovered from an ex- age, and more vigorous post-flooderosion. This, posure of the late PleistoceneTouchet Beds are in turn, resulted in removal of the fine-grained exceedingly well-preserved, with fragile bone upper portion of the rhythmites in the Touchet elementsand dentition preservedunbroken. In River section. addition, the fossils show no evidence for abra- sion or reworking, suggesting little or no The presence of a well-preserved small- transport prior to burial and fossilization. mammal fauna within these sediments strongly Associated with the fossils, and present supports the hypothesis which invokes multiple throughout both study sections,are abundant ro- flood episodes,separated by decades-longperiods dent burrows. The available evidence indicates of exposure characterized by erosion and/or non- that the fossils represent the remains of members deposition, to explain the deposition of the of an establishedlocal community of mammals Touchet Beds.

A Small Mammal Fauna from the Touchet Beds t73 References Cited Carson,R. J., C. F. McKhann,and M. H.Pizey. 1978.The Touchet Beds of the Walla Walla Valley. 1n V. R. Alcorn,J. R. 1940.Life history of the Piute ground squirrel. Bakerand Dag Nummedal(eds.) The ChanneledScab- J. Mammal.2l:160-170. land.Washington, D.C., NASA. Pp. 173-177. Atwater,B. F. 1984.Periodic floods from glacial Lake Missoula Mullineaux,D. R. and D. R. Crandell.l98I. The eruptiye into the Sanpoilarm of glacialLake Columbia,north- historyof Mt. St. Helens.U.S. Geol. Suru. Prof. Paper easternWashington. Geology 12:464-467. 1250.Pp. 3-15. Baker, V. R. 1978. geology of the channeled Quaternary Spencer,P. K. 1987.A small-mammalfauna from Pleistocene scablandand adjacentareas. ln V. R. Bakerand Dag MissoulaFlood depositsnear Touchet,southeastern Nummedal (eds.) The Channeled Scabland. Washingron.Geol. Soc. Amer. Abstractswith Program, Washington,D.C., NASA. p. 17-35. vol. 19,no. 6. Bjornstadt,B. N. 1980.Sedimentology and depositionalen- Waitt, R. B., Jr. 1978.Post-Miocene stratigraphy and tec- vironment of the Touchet Beds,Walla Walla River tonism of parts of the GreatColumbia Plain and ad- basin,Washington. Richland, Washington, Rockwell- jacent Cascades,Washington. In Tectonics and Hanford Operations, Document RHO-BWI-SA-44. Seisnicity of the ColumbiaPlateau Workshop meeting 107p. volume. Richland, Washington, Rockwell-Hanford Bretz, H. 1923.The channeledscablands J on the Columbia Operations. Plateau.J. Geol.3l:617-649. 1980.About forty last-glacialLake Missoula 1925.The SpokaneFlood beyond the channeled Jokulhlaupsthrough southernWashington. J. Geol. scablands. Geol. 33:97-115, 236-259. J. 88:653-679. 1930.Valley depositsimmediately west of the 1984.Periodic Jokulhlaups from Pleisrocene channeledscabland. J. Geol. 38:385-422. glacial Lake Missoula-New evidencefrom varyed 1969.The Lake Missoulafloods in the chan- sedimentin northernIdaho and Washington. neled scabland. Geol. 77:505-543. Quater- J. nary Res.22:46-58. Bretz,J. H., H. T. U. Smith,and G. E. Neff. 1956.Channeled I985a. Casefor periodic,colossal jokulhlaups scablandof Washingron:new data and interpretations. from Pleistoceneglacial Lake Missoula.Geol. Soc. Geol. Soc. Amer. Bull., 67:957-1049. Amer. Bull. 96:127I -1286. 1985b. Reply to commenr on "Periodic jokulhlaups Receiued12 1988 from Pleistoceneglacial Lake Missoula- luly New evidencefrom vaned sedimentin northern Idaho Accepted,for publication I February 1989 and Washington.Quaternary Res. 24;357-360.

174 Spencer