A two decade study of the archeology, paleoclimatology and paleogeography of the Dempsey Divide, the upland ridge separating the valleys of the Washita River and the North Fork of the Red River in southern Roger Mills County, Oklahoma.

J. Peter Thurmond¹ and Don G. Wyckoff²

  ¹Thurmond Ranch, Inc., Rt. 1, Box 62-B, Cheyenne, OK, 73628-9729; DempseyDiv@AOL.com 

²Oklahoma Museum of Natural History, 2401 Chautauqua Avenue, Norman, OK, 73072-7209; XTRambler@OU.edu

(To look at tables, charts or pictures described in article, click on bold figure numbers within the text.  For a complete list of the figures, click here)

Introduction

    Since the early 1980s, the upland area between the Washita River and the North Fork of the Red River in northern Beckham and southern Roger Mills counties, western Oklahoma, a landform we have termed the Dempsey Divide, has been the subject of a quiet study that is one of the most intensive investigations of a locality undertaken by Southern Plains prehistorians to date (Thurmond 1990, 1991a, 1991b, 1991c, 1997, 1999; Thurmond and Wyckoff 1998, 1999; Thurmond et al. 1998, 2000). Over 400 archaeological sites have been recorded, and some 150 radiocarbon dates have been obtained from these sites and from buried soils across the study area. Our overall objective has been to determine when people were here in prehistoric times, what parts of the landscape they inhabited and exploited for food and raw materials, and how their presence or absence relates to past changes in climate.  

The Dempsey Divide Study Area

  The Dempsey Divide is in the erosional zone at the eastern edge of the Tertiary period Ogallala Formation, a thick accumulation of sediment that formed as an outwash (piedmont) plain when the front range of the Rockies was uplifted between 17 and 3 million years ago. The uneroded surface of this relict plain in the Texas and Oklahoma panhandles is called the High Plains or Llano Estacado, and the eastern edge of this upper surface is termed the Caprock Escarpment. The 3-D terrain maps shown in Figure 1 and Figure 2 graphically show the relationship between the Llano Estacado and both the mountains of New Mexico to the west and the Rolling Plains to the east.

  The upland divides between the rivers of the Rolling Plains served as prehistoric highways, leading like gangplanks up onto the High Plains. Our primary study area on the Dempsey Divide, the Thurmond Ranch archaeological survey area, straddles the outcrop edge of the base of the Ogallala Formation Figure 3 and Figure 4, which is underlain by Permian period redbeds. The Permian deposits accumulated in shallow marine and shoreline environments between 290 and 250 million years ago. Given its mostly sandy texture, the Ogallala Formation is a major aquifer. All of the significant streams along the Dempsey Divide are spring fed. Springs emerge along the base of the Ogallala because the underlying Permian units are fine textured and nonporous.

The Ogallala Formation outcrop edge also forms a major ecological boundary or ecotone, because the plant communities on either side are very different. Trees, brush and tall grasses grow in the deep soils on the Ogallala side, while short grass is the typical vegetation of the Permian side of the boundary with its thin soils. The Ogallala side offered a complex mix of edible and medicinal wild plants and small game, while the Permian side was bison country. The most efficient place for prehistoric hunter-gatherers to camp was on this boundary, and a striking concentration of campsites occurs along the Ogallala ecotone Figure 4.

  The Late Pleistocene Record

    We tend to think of the "Ice Age" as a thing of the past, but we are living in it still. Known by geologists as the Pleistocene epoch, the 'ice age' or glacial cycle began some 2.5 million years ago. For most of this span, world glaciers waxed and waned every 41,000 years or so. But over the past 700,000 years, this has happened every 100,000 years Figure 5. These cycles are controlled by the orbital mechanics of the Earth, which cause the effective solar radiation reaching the Northern Hemisphere to vary over time.

The tilt of the Earth's axis of rotation relative to its orbital plane, or orbital obliquity, varies from 22.5° and 24.5° in 41,000 year periods. Earth's orbit around the sun is slightly elliptical. Elongation of this ellipse, or orbital eccentricity, fluctuates in 100,000 year cycles. These astronomical triggering mechanisms were apparently amplified by changes in deep-ocean circulation, jet stream pathways, and greenhouse gas production/sequestering by the Earth's biota (cf. Anderson and Borns 1994).

Each of the 100,000 year glacial cycles of the past 700,000 years has been dominated by a roughly 85,000 year (glacial) period, when average global temperatures were much lower than today, and climate was highly unstable. Each cycle has terminated in a warmer and much more climatically stable interglacial of 10,000-15,000 years. We are living in the latter part of an interglacial now, termed the Holocene by geologists, which began some 10,000 years ago.

The alluvial terraces of the Canadian River in western Oklahoma record the Late Pleistocene 100,000 year climate cycle Figure 6 and Figure 7. Each terrace is an ancient floodplain representing the relatively stable climate of an interglacial. The Canadian River downcut during the climatically unstable glacial ('ice age') maxima, and stabilized during each interglacial. A fission track date on volcanic ash near Putnam from the Lava Creek B cataclysmic eruption at Yellowstone (yes, no kidding, Yellowstone National Park) neatly caps this terrace sequence at 610,000 years ago (Figure 8; Ward 1991a, 1991b).

On the Dempsey Divide, the Pleistocene record extends back some 28,000 years. The last glacial peaked in intensity from 21,000 to 17,000 years before present (BP). This period is known by paleoclimatologists as the Last Glacial Maximum (LGM). The westerly storm track was diverted down across the American Southwest and Southern Plains during this time by the Laurentide ice sheet of Canada and the northern U.S. (Oviatt et al. 1999). Enormous freshwater lakes formed in the Great Basin region of the western United States (Trimble 1989).

Average rainfall and lake levels were higher across the region during the LGM, but fluctuated wildly. Dunes atop the divide stabilized under vegetation, then accreted new sand in a 400 year cycle of violent swings from parkland savanna to desert, in synchrony with highstands at glacial lakes Estancia and King to the west in New Mexico and trans-Pecos Texas (Figure 9; Table 1; Allen and Anderson 1993; Phillips et al. 1992; Thurmond and Wyckoff 1998; Wilkins and Currey 1997).

During the LGM, the upper valleys of Brokenleg and Sergeant Major creeks on the Thurmond Ranch were filled with sediment some 21 meters (70 feet) higher than the modern channels (Figure 10). The broad creek bottoms were networks of perennial marshes and ponds, and the whitish-gray (gleyed) color of these sediments, in stark contrast to the underlying Permian redbeds, reflects this former anoxic environment.

At Brokenleg Bend Exposure #1, a pond deposit replete with aquatic snails accumulated 28,000-21,000 BP (Figure 11). A major erosional cut separates this pond deposit from overlying marsh sediments that date 12,000-3,000 BP. At Sergeant Major Creek Exposure #1, a laminated stream deposit rapidly accumulated during the apparent gap in the Brokenleg Bend #1 sequence, 17,000-15,000 BP (Figure 12).  The sediments at Sergeant Major #1 are derived almost entirely from the Ogallala Formation, and contain little Permian redbed material.

These remnants of Late Pleistocene valley fill suggest that the basal edge of the Ogallala Formation eroded back to the south about one mile, and this entire stretch of the landscape was lowered some 21 meters (70 feet) during this 17,000-15,000 BP interval, a time when global climate was flickering between interglacial and full glacial conditions (Figure 13). The local climatic gyrations must have been staggering.

Holocene Climate, Geomorphology and Archeology

  A remarkable climate record is available for the last 10,000 years  on the Southern Plains from Hall's Cave in central Texas (Toomey et al. 1993). A detailed sequence of change in relative frequency of two moisture-sensitive species of shrew (Notiosorex crawfordi and Cryptotis parva) documents the trend in effective precipitation over time in detail (Figure 14).

The first evidence of human occupation on the Dempsey Divide occurs during one of the two wettest periods of the Holocene indicated for the Southern Plains by the Hall's Cave record, from 10,000 to 8500 BP. Small campsites with spear points in many of the Late Paleoindian styles shown in Figure 15 have been found in protected settings along Brokenleg, Currant and Sergeant Major creeks (Thurmond 1990, 1991c). Bison-hunting Late Paleoindian bands were likely coming down off the High Plains along the crest of the divide, and camping in the tributary stream valleys during the coldest winter months. These valleys would have offered running water, firewood, and shelter from the wind.

           The Hall's Cave record suggests that average annual rainfall gradually decreased on the Southern Plains after 8500 BP (Figure 14). There is only sparse evidence of a human presence on the Dempsey Divide from this time until about 2000 BP. Regional climate deteriorated into a drought, known as the Altithermal, of a severity that defies belief (cf. Antevs 1955; Holliday 1989; Reeves 1973). Regional rainfall appears to have fallen to desert levels around 7000 BP, and stayed there for most of the next four millennia. At Mustang Springs, northeast of Midland, Texas, hunter-gatherers dug a series of progressively deeper wells chasing a retreating water table at about 6800 BP (Meltzer and Collins 1980).

The desertification and floral denudation of the Dempsey Divide study area would have left the soft, unconsolidated Permian, Tertiary and Quaternary sediments extremely vulnerable to erosion when it did rain. Most of the rainfall during the Altithermal probably derived from brief, intense convective thunderstorms during the early warm season, producing high runoff. Deep canyons were eroded into the tributary stream valleys along the Dempsey Divide, much like those we see today. The Late Pleistocene valley fill deposits and Late Paleoindian campsites we have studied survived this fluvial incision event only as small erosional remnants along the canyon rims.

            Again, the Dempsey Divide and most of the surrounding region was apparently abandoned by people during most of the Altithermal. We have identified two brief incursions during wetter interludes indicated by the Hall's Cave record. Bands of the Calf Creek horizon, crafting dart points like those shown in Figure 16, entered the area sporadically during the first of these mesic periods around 5000 BP from central and eastern Oklahoma (Thurmond and Wyckoff 1999). At the very end of the Altithermal, people of the McKean complex ventured down into western Oklahoma from southeastern Colorado, probably during the pronounced wet episode indicated at Hall's Cave around 3000 BP.

At about the time of the McKean incursion, the canyons along the Dempsey Divide began to refill with sediment. The Hall's Cave record indicates that regional climate was flickering between near modern rainfall and full-Altithermal aridity at this time, on a centennial scale. It appears that the (on average) increased rainfall resulted in mass wasting of the upland slopes, washing enormous quantities of fine sediment and rock fragments down into the canyons. At archaeological site 34RM507 on the east side of the Thurmond Ranch, radiocarbon dating demonstrates that the canyon filled nearly to its modern rim between about BC 1300 and AD 100 (Figure 17 and Figure 18). The brief return to full Altithermal conditions around BC 1100 indicated at Hall's Cave presumably chased the McKean folks back to Colorado, again depopulating the area.

           People did not return to the Dempsey Divide until after BC 50, near the end of the rapid canyon filling sequence. These people of the Late Archaic Twilla phase made corner and side notched dart points, and  distinctive polished stone artifacts known as a lunate stones (examples are illustrated in (Figure 19 and Figure 20). The regional distribution of these artifact styles suggests that, as climate moderated after the end of the Altithermal, Twilla phase people entered the Southern Plains from the northeast, possibly from south-central or southeastern Kansas, and rapidly spread south along the Caprock Escarpment (Figure 21).

           For the next millennium, the Twilla phase people appear to have inhabited the study area with only very minor changes in material culture, primarily in projectile point styles. On the basis of these changing styles, we identify an Early Woodland Lake Creek phase, with ¹⁴ C dates mostly between AD 400 and 600, during the transition from the atlatl and dart to the bow and arrow. By AD 900-1000 use of the dart had ceased and cord marked pottery of styles common to the east and northeast was being crafted, leading us to define a Late Woodland Beaver Dam phase. A typical Late Woodland arrow point assemblage is illustrated in (Figure 22).

The uplands of the Dempsey Divide were virtually abandoned after AD 1000, at the time that occupation of the Hay-Cyclone-Quartermaster Creek basin began 30 kilometers (20 miles) to the northeast (Figure 23 and Figure 24; Baugh et al. 1984; Moore 1988; Thurmond 1991b). Only a handful of small, ephemeral archeological sites with artifacts diagnostic of the Late Prehistoric and Protohistoric periods (AD 1000 - 1700) have been recorded on the Dempsey Divide (Thurmond op. cit.). The artifacts of the latest Dempsey Divide sites and the earliest Quartermaster Creek sites are much the same, and these are likely the same people. It appears that the local people abandoned their former broad spectrum hunter-gatherer subsistence pattern for a specialized Plains Villager lifestyle of horticulture and bison hunting at about AD 1000. While the complex resources of the Ogallala ecotone were attractive to hunter-gatherers, the deep, fertile soils of Hay, Cyclone and Quartermaster creeks are more suited to farming (Figure 25). The interfluvial uplands of the Quartermaster basin would have provided ideal grazing for bison.

  The 400 Year Rainfall Cycle

  Intensive radiocarbon dating of buried soils that formed during the last two millennia on and around the Dempsey Divide has documented an approximate 400 year rainfall cycle during the Late Holocene. As discussed above, this same cycle appears to have been operating during the Late Pleistocene, judging by the few dates we have from LGM dune soils. Dark, organically stained topsoils formed during periods of greater rainfall termed pluvials. We have sought out localities where these soils are separated from one another by lighter-colored sediment that is not organically enriched, deposited during intervening drier periods known as interpluvials. Radiocarbon dating of the organic carbon in the top and bottom five centimeters (two inches) of each soil establishes beginning and end dates for its formation, and thus for the start and end of each pluvial (Figure 26 and Figure 27).

           Five pluvials have been defined on the basis of this soils study, with the first pluvial beginning at BC 50, as specified below. A detail of the radiocarbon dates on which these pluvial definitions are based is provided as Table 2. Note that the first well-documented pluvial, the Finch Canyon pluvial of BC 50 - AD 100, coincides with the arrival of the people of the Twilla phase. This is surely not a coincidence. People presumably returned to the area when the biota had recovered sufficiently from the long Altithermal drought to support hunter-gatherers.

However, we believe we see hints of at least three earlier Late Holocene pluvials at the end of the Altithermal. The laminated fill at 34RM507 that accumulated from BC 1300 to AD 100 (illustrated in Figure 17 and Figure 18) contains two distinct bands of greater woody charcoal density that date BC 500-300 and BC 900-700. Counter-intuitive as it may seem, local range fires may have been more likely during the first few post-Altithermal pluvials. Given the late arrival of people back on the scene, the post-Altithermal environmental recovery was apparently a long process, and you can't have a fire without fuel.

  ______________________________________________________________________________

Dempsey Divide Late Holocene Climate Sequence

A Four Century Cycle in Average Annual Effective Precipitation

  Defined Climate Intervals:

Average Pluvial Duration: 185 Years

Average Interpluvial Duration: 205 Years

Average Climate Interval Duration: 390 Years

We infer that a pluvial of ca. BC 1300-1100 was responsible for the inception of Late Holocene canyon system sedimentation on the Dempsey Divide. In fact, the 400 year rainfall cycle is likely a permanent feature of Southern Plains climate, and operated even during the Altithermal. Century-scale cycles during the Altithermal would have been, for the most part, between more and less arid conditions. The Calf Creek and McKean regional incursions apparently mark the two significant pluvials of the four altithermal millennia.

  Comparative Climate Data

  From a review of the relevant paleoclimatology literature, it is clear that the century-scale rainfall cycle we see on the Dempsey Divide is part of a global climate pattern. There are examples of  equivalent cycles in temperature and rainfall as widely separated as Antarctica; the Andes; Tibet; California; the American Northwest, Southwest and Great Plains; Mexico; the Caribbean; Greenland; Scandinavia; the Russian Arctic; Western Europe; and North Africa (Barber 1981; Benson et al. 1998; Cannariato 1999; Dugas 1998; Gordiyenko et al. 1980; Jiang et al. 1997; Lamb and van der Kaars 1995; Leventer et al. 1996; Lubinski et al. 1999; Madole 1995; Magny 1993; Mandel 1994; Mehringer and Wigand 1986; O'Hara and Metcalfe 1995; Petersen & Mehringer 1976; Peterson et al. 1991; Pfister et al. 1998; Scuderi 1993; Stokes and Swinehart 1997; Svensen and Mangerud 1997; Thompson et al. 1997; Thompson et al. 1998; Wright et al. 1985; Yu and Ito 1999).

  There are decadal, centennial and millennial scale cycles in Earth climate coincident with and presumably driven by equivalent cycles in solar output (Bard et al. 1997; Damon and Sonett 1991; Jirikowic and Damon 1994; Karlen and Kuylenstierna 1996; Lean and Rind 1996; Leventer et al. 1996; Reid and Gage 1988; Stuiver and Braziunas 1989; Suess and Linick 1990; Wigley 1988) and amplified by the ultraviolet absorption characteristics of stratospheric ozone (Haigh 1996; Shindell et al. 1999; van Geel et al. 1999). As indicated in Figure 29, two of the most pronounced cycles in solar output have (probably harmonic) periods of 200 and 400 years. Other striking cycles in the paleoclimate record have (also presumably harmonic) cycles of 800, 1600 and 2400 years (cf. Hereford et al 1998; Thompson et al. 1997; van Geel et al. 1999).

           The big picture to be gained from this 30,000 year record from the Dempsey Divide, and that of the last 10,000 years in particular, is sobering. As indicated below, hunter-gatherers found the uplands of Brokenleg and Sergeant Major creeks uninhabitable nearly half the time, and only marginally habitable another fourth of the time.  

Dempsey Divide Inferred Habitation Intensity, 10,000-1000 BP

No or low intensity habitation 72% of the time span.

_____________________________________________________________________________________

Further, recall from Figure 24 that over half of the archaeological sites on the Dempsey Divide date to the Late Archaic/Woodland period of the first millennium AD, or just 10% of the relevant timespan. If one considers the 28 radiocarbon dates assayed from the Late Archaic/Woodland campsites within the study area (shown below), all but four are coeval with pluvials, and most date to the Herring Creek pluvial of AD 400-600. Thus, a majority of the human habitation of the Dempsey Divide occurred during this single 200 year climatic event.

__________________________________________________________

Calibration Curve Intercepts of ¹⁴C Dates from

Dempsey Divide Late Archaic/Woodland Campsites

Late Woodland Beaver Dam phase dates coeval with the Higgins Creek pluvial

Early Woodland Lake Creek phase dates coeval with the Herring Creek pluvial

Early Woodland Lake Creek phase dates coeval with the Herring Creek interpluvial

Late Archaic Twilla phase dates coeval with the Finch Canyon pluvial

  _________________________________________________________________

So Now What?

  Modern residents of Western Oklahoma live in a climatically marginal region, to say the least. And we are seeing the region in the best of times, at the height of a pluvial. Even on a scale of tens of thousands of years, the paleoclimate and archaeological records indicate that it doesn't get much better than what Western Oklahoma inhabitants have experienced over the past century, and the last few decades in particular.  When a thing is as good as it gets, any potential change is worrisome.

  So where are we headed next in regional climate?  The instrumental rainfall record for western Oklahoma since 1893 suggests that a pluvial began around AD 1900, and continues to this date (Figure 28 and Table 3). There was a major climatic correction in the 1930s (no great surprise). Average annual rainfall gradually increased from around 500 mm (20") at the start of the 20^th century to over 750 mm (30") at its end, an increase of fifty percent. Based on the soils record of the last two millennia, we should be about halfway through a pluvial, and residents of Western Oklahoma should be able to look forward to roughly another century of high rainfall.

           The wrinkle in this scenario is the great climate experiment humanity is currently (albeit inadvertently) conducting in the burning of fossil fuels. Contrary to popular opinion, the verdict is in on the subject of global warming as far as most climatologists are concerned (cf. Bradley 2000). Earth's oceans and atmosphere are indeed warming (cf. Deming 1995; Pollack et al 1998). About half of the 0.55° C rise in mean global surface temperature since 1860 can be attributed to solar forcing, and the rest to greenhouse effect (Lean et al. 1995; Lean and Rind 1996). Surface temperature warming has been strongly moderated by heat absorption in Earth's oceans to a depth of 3,000 meters, else the surface temperature rise would have been greater (Levitus et al. 2000). Atmospheric concentrations of the two major greenhouse gases, carbon dioxide and methane, are 20% and 100% higher, respectively, than they have ever been in the last 400,000 years, and are still climbing (Figure 30; Raynaud et al. 2000). Atmospheric greenhouse gas concentrations are higher than during the Altithermal. We find that a bit disquieting.