Holocene paleoenvironments in southeastern Minnesota ...people.ku.edu/~lgonzlez/NewFiles/Publications/Bakeretal02.pdf · southeastern Minnesota, USA. Early Holocene forests were replaced

Holocene paleoenvironments in southeastern Minnesota ^chasing the prairie-forest ecotone

R.G. Baker *, E.A. Bettis III, R.F. Denniston 1, L.A. Gonzalez,L.E. Strickland 2, J.R. Krieg

Department of Geoscience, University of Iowa, Iowa City, IA 52242-1379, USA

Accepted 16 August 2001

Abstract

This paper uses a multi-proxy approach involving pollen, plant macrofossils, speleothem isotopes, and alluvialhistory of streams to reconstruct the history of prairie expansion and contraction along the prairie-forest border ofsoutheastern Minnesota, USA. Early Holocene forests were replaced by prairie along this border, but eastwardexpansion of prairie stalled for 2000 yr when the prairie-forest ecotone stabilized. Prairie invaded the area from the west8000^9000 yr B.P., but mesic forest remained less than 100 km to the east until about 6000 yr B.P. Changes in N13Cvalues in speleothem calcite, that reflect the rise of C4 grasses, correlate well with the presence of C4 grass speciesidentified in the plant macrofossil record. Periods of large floods correlate with speleothem evidence of dry summers,increased cool-season precipitation (both resulting in less plant cover to absorb moisture), and change to prairievegetation. ß 2002 Elsevier Science B.V. All rights reserved.

Keywords: grasslands; paleoenvironment; pollen; plant macrofossils ; speleothems; alluvial history

1. Introduction

Many researchers working on the paleoenviron-mental history of grasslands have lamented thatwe could only identify main grass groupings (C3

vs C4, or various tribes), because no fossil evi-dence (including pollen, phytoliths, and carbonisotopes) permitted the identi¢cation of grassand forb species. If this level of identi¢cation

were possible, we could determine exactly whatspecies assemblages were present in past commun-ities. This paper presents such evidence as part ofan integrated study of Holocene environmentalchange involving the prominent prairie-forest eco-tone in southeastern Minnesota and northeasternIowa, USA. The paper uses pollen and especiallyplant macrofossils from small-stream deposits toreconstruct the chronology of vegetational changealong this ecotone, and these changes are used toinfer climatic conditions. The history of £uvialactivity of the stream is also presented, and theimportance of both climate and vegetation arediscussed as potential forcing functions of changesin the pattern of erosion, stability, and depositionalong the stream. These environmental changes

0031-0182 / 02 / $ ^ see front matter ß 2002 Elsevier Science B.V. All rights reserved.PII: S 0 0 3 1 - 0 1 8 2 ( 0 1 ) 0 0 3 5 4 - 6

1 Present address: Geology Department, Cornell College,Mt. Vernon, IA 52314, USA.

2 Present address: Earth Surfaces Processes, U.S. Geologi-cal Survey, Box 25046 MS 980, Denver Federal Center,Lakewood, CO 80225, USA.

* Corresponding author.

PALAEO 2715 22-1-02 Cyaan Magenta Geel Zwart

Palaeogeography, Palaeoclimatology, Palaeoecology 177 (2002) 103^122

www.elsevier.com/locate/palaeo

are then correlated with high-resolution carbon-and oxygen-isotopic records from calcite in near-by cave speleothems.

The history of Holocene prairie expansion andcontraction in the Midwest has been discussed forover 75 yr beginning with Gleason (1922). For thelast 35 yr paleoecologists have accumulated abun-dant and reliable pollen data (Cushing, 1965;Wright 1968; Grimm, 1984). Yet there are stillmajor questions. It is well established frommany sites that drier conditions began in the earlyHolocene from the western Great Plains eastwardto northwestern and central Minnesota (Webb etal., 1983; Barnosky et al., 1987). In northwesternMinnesota, this history has been worked out inexquisite detail (McAndrews, 1966; Bradbury andDean, 1993), indicating especially arid conditionsfrom about 8000 to 4000 yr B.P. (uncalibrated 14Cyr before present). The scarcity of lakes and wet-lands south of the Late Wisconsin glacial borderhas led us to investigate stream sediments andcave speleothems as sources of paleoenvironmen-tal reconstruction (Baker et al., 1996, 1998; Den-niston et al., 1999a,b).

Recent work has indicated that maximum arid-ity in southern Wisconsin and eastern Iowa waslater, from about 5500 to ca. 3500 yr B.P., andclimate remained moist until mid-Holocene time(Baker et al., 1992, 1996; Chumbley et al., 1990).A sharp early to mid-Holocene climatic boundaryhas been documented between these regions(Baker et al., 1990, 1992), but its position is stillnot well established. Furthermore, the boundarywas proposed largely on the basis of evidencefrom plant remains in stream deposits ; this tech-nique appears to be powerful, but it needs furthertesting.

This paper tests (1) the hypothesis that a sharpmid-Holocene climatic and vegetational boundaryextended northward from Iowa into southeasternMinnesota, (2) the reliability of using fossils instream cutbanks by comparing these new siteswith the Roberts Creek sites in northeasternIowa and with high-resolution isotopic chronolo-gies and (3) the consistency of the relationshipbetween climate, vegetation, and stream activityat new sites.

To accomplish these goals, we (1) report on thedetailed paleoecology derived from plant macro-

fossils and pollen at sites from two nearly adja-cent valleys, Money Creek and Pine Creek (MOon Fig. 1), in southeastern Minnesota, (2) describethe alluvial stratigraphy of outcrops that illus-trates the pattern of Holocene alluvial cut-and-¢ll on many Midwestern £oodplains, and relatethat history to the climatic and vegetationalchanges, (3) compare these data with carbon iso-topic studies of speleothems from caves in south-eastern Minnesota and northeastern Iowa, and(4) use these data to de¢ne the position of themid-Holocene climatic boundary across the cen-tral Midwest.

We sought to test the hypothesis that a mid-Holocene climatic boundary extended northwardfrom Iowa into Minnesota by ¢nding a site be-tween Kirchner Marsh in southeast central Min-nesota (Fig. 1), where drying occurred in the earlyHolocene (Watts and Winter, 1966), and north-eastern Iowa, where dry conditions did not devel-op until mid-Holocene time (Baker et al., 1996).

Fig. 1. Map of central Midwest showing locations of caveand paleoecological sites used to constrain mid-Holoceneprairie-forest boundary. Money Creek and Pine Creek arewithin the dot for locality MO.


R.G. Baker et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 177 (2002) 103^122104

Organic-rich sediments were discovered in alluvi-um in two nearly adjacent stream valleys in south-eastern Minnesota in the area of the proposedboundary: Money Creek (43³53P49Q N. Lat.,91³40P24Q W. Long.) and Pine Creek (43³52P25QN. Lat., 91³28P20Q W. Long.), Winona Co., MN,USA (Fig. 1). We used the record of plant re-mains and cut-and-¢ll relationships exposed inthe banks of these small streams, as describedby Baker et al. (1996) to reconstruct the vegeta-tional and alluvial history. Each cutbank localityrepresents a rich fossil assemblage providing asnapshot of 6 100^200 yr of deposition. Thesesites di¡er from those at Roberts Creek in north-eastern Iowa (Chumbley et al., 1990; Baker et al.,1996), however, in that they are in a landscape ofvalleys deeply incised into bedrock, with 100 m ormore of relief. This landscape houses abundantmicrohabitats related to slope aspect, bedrock, lo-cal groundwater £ow, and other local features,that have been important in£uences on vegetationduring the Holocene.

The carbon and oxygen isotopic composition ofspeleothem calcite has been used for paleoclimatereconstruction (Dorale et al., 1992). The N18O ofspeleothem calcite has been interpreted as re£ect-ing the isotopic composition of precipitation andmean annual temperature (Hendy and Wilson,1968; Dorale et al., 1992). However, these rela-tionships can be complicated by evaporative en-richment of 18O in meteoric waters prior to theirin¢ltration into the subsurface (Denniston et al.,1999a). Speleothem N13C may also record changesin vegetation. The carbon in speleothem calcite isderived from dissolution of carbonate bedrockand decay of soil organic matter (atmosphericCO2 is negligible). Because N13C values of C4

plants (319x to 39x) are higher than thoseof C3 vegetation (332x to 320x), increases inC4/C3 ratios also increase N13C values of soil or-ganic matter (Boutton, 1991). This carbon iso-topic signal is transported vertically downwardwith in¢ltrating £uids and is eventually recordedby higher N13C of speleothem calcite. Because car-bon is added to solutions after it has penetratedbelow the land surface, it does not su¡er frompre-in¢ltration e¡ects and is therefore a reliabletool for reconstructing paleovegetation.

2. Study area

Money and Pine Creeks are tributaries of theRoot and Mississippi Rivers respectively, that£ow southeastward through the so-called `Drift-less Area' of southeastern Minnesota, an exten-sion of Iowa's Paleozoic Plateau landform region(Fig. 1; Prior, 1991). The region's landscape re-£ects a strong in£uence of underlying Paleozoicmarine carbonate and sandstone bedrock andwas formerly included in the `Driftless Area' ofthe Upper Mississippi Valley, a designation nowknown to be incorrect (Knox and Attig, 1988;Hobbs, 1995). Valleys are narrow and deeply en-trenched with steep, rocky valley walls. Uplandsare moderately to steeply sloping and a thin (1^10m thick) mantle of Pleistocene deposits coversthe bedrock surface (Hobbs, 1995). The groundwater table of the Cambrian sandstone aquifer isintersected by streams in the study area, andground water discharges to the streams from thebedrock units, fostering continual saturation andpreservation of organic plant remains in the basalparts of these valleys' alluvial ¢ll.

The area has a continental climate, with a meanannual temperature of 46³ F. (7.7³C.) and a meanannual precipitation of 33 inch (108 cm) at nearbyWinona, MN. Precipitation is highest in thesummer months. The pre-settlement vegetationin this area consisted of deciduous forest. Com-mon forest trees included Quercus macrocarpa(bur oak), Quercus alba (white oak), Quercusrubra (red oak), Acer saccharum (sugar maple),Tilia americana (basswood), Fraxinus pennsylva-nica (green ash), Ulmus americana (Americanelm), Ulmus rubra (red elm), Carya cordiformis(bitternut hickory), Juglans cinerea (butternut),Populus tremuloides (aspen) and Ostrya virginiana(hophornbeam) (Grimm, 1984; Strickland, 1998).The prairie-forest ecotone was an irregular tran-sition zone only a few km to the west. Beyondthat the prairie was dominated by tall C4 grassesincluding Andropogon gerardii (big bluestem),Schizachyrium scoparium (little bluestem), andSorghastrum nutans (Indian grass), along withmembers of the sun£ower (Asteraceae) family, in-cluding Helianthus spp. (sun£ower) and Heliopsishelianthoides (ox-eye), and legume (Fabaceae)


R.G. Baker et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 177 (2002) 103^122 105

family, including Amorpha canescens (leadplant)and Desmodium spp. (tick-trefoil). These taxalisted are ones that also occur in our fossil re-cords.

Money and Pine creeks are 4th-order valleyswith drainage areas of approximately 42 km2

and 60 km2 respectively at the study areas. Thevalley £oors of the creeks range in width from 0.2to 0.4 km, and both creeks meander within theirvalleys, though several segments of each havebeen arti¢cially straightened. The creeks £owthrough channels 2^4 m in width and 2^3 mdeep. Both streams have pool and ri¥e morphol-ogy, with low-gradient, mud bottom, quiet waterpools separated by short, steep-gradient, cobble-bottom, swift-water ri¥es. During low-£ow con-ditions in most summers water depth ranges from0.5 to 1.5 m in pools to 0.2^0.1 m in ri¥es. Highwater conditions are usually associated withspring snow melt and thunderstorms, while low-£ow conditions, when most of the stream's £ow issupplied by ground water discharge, usually occurduring the summer and early fall.

3. Methods

Samples of organic-rich sediments were col-lected in cutbanks along the upper reaches ofMoney Creek and Pine Creek by shovel, storedin sealable plastic bags, and brought to the lab.Sediments and soils were described along MoneyCreek using standard terminology and procedures(Soil Survey Sta¡, 1993). Pollen samples from thesame sediment used for macrofossils were pro-cessed using KOH, HCL, HF, and acetolysis so-lution as described in Faegri et al. (1989) withfurther screening after treatment with 0.1 N so-dium pyrophosphate followed by dilute (0.2%)bleach (NaOCl). Sediment samples of 200 mlwere washed through 0.5- and 0.1-mm sieves forplant macrofossils, picked by hand, and stored inglycerin with crystals of phenol. Because of thelarge volume of ¢ne-grained sediment, it was im-practical to pick through all the ¢ne residue; aminimum of three dishes (10 cm diameter) werepicked from each level. More material was pickedif new taxa continued to be found. As a result,

macrofossils between 0.5 and 1 mm, marked withan asterisk on the diagrams, are minimum values.Macrofossils are mostly seeds and fruits, unlessotherwise labeled. All grass macrofossils identi¢edwere £orets. Pollen and plant macrofossils wereidenti¢ed using reference collections at the Geo-science Department, University of Iowa. Macro-fossil specimens and pollen slides are preserved inthe Geological Repository there, and completemacrofossil data are on ¢le at the North Ameri-can Plant Macrofossil Database, at the NationalGeophysical Data Center, Boulder, CO, USA.Diagrams were plotted using Tiliagraph 2.0.b.5.Radiocarbon dates (Table 1) were obtained fromwood collected from each cutbank. Unless other-wise speci¢ed, all radiocarbon dates are in uncali-brated 14C yr B.P. (before present).

Stalagmites were sawed in half vertically, pol-ished, and sampled for carbon and oxygen iso-topes at intervals ranging from 1 to 3 mm usinga modi¢ed dental drill with a 500-Wm-diameterbit. Carbon and oxygen isotopic analyses wereperformed at the University of Michigan StableIsotope Laboratory using a MAT-251 gas-sourcemass spectrometer. Samples were converted toCO2 by reacting them with phosphoric acid at72³C. All values are reported in x with oxygenrelative to SMOW and carbon relative to PDB;analytical precision is greater than 0.1x for bothcarbon and oxygen.

Powders were similarly extracted for 238U^234U^230Th dating by thermal ionization massspectrometry (TIMS) at the University of Minne-sota for the Spring Valley stalagmite and at LosAlamos National Laboratories for the ColdWater Cave stalagmite (Dorale et al., 1992). An-alytical procedures are modi¢cations of those dis-cussed in Chen et al. (1986) and Edwards et al.(1987; 1993).

4. Results

4.1. Alluvial history

The valley £oors of both Money and Pinecreeks consist of four alluvial surfaces designatedT0^T3. T3 is the highest surface and is elevated



about 3 m above creek level at low £ow. T3 islevel and merges in a smooth, concave upwardpro¢le with small, loess-free colluvial slopes alongthe valley margins. This terrace is underlain byloamy alluvium that grades downward into siltyand gravelly alluvium. T2 is separated from T3 bya long gentle scarp and is approximately 0.2^0.5 mlower in elevation than T3. T2 is underlain byloamy grading downward to gravelly alluvium.T1, separated from T2 by a sort, steep, 0.3^0.8 mhigh scarp, is marked with shallow, ¢lled aban-doned meander scars and subtle undulations. T1roughly parallels the modern (unstraightened)channel and is underlain by loamy alluvium thatgrades downward to gravelly alluvium. T0, thelowest surface in the valley, is the modern channelbelt, separated from older surfaces by a short,steep one to two meter high scarp and character-ized by abandoned meanders that contain water,small natural levees, and partially masked to ac-tive point bars and chutes. T0 is £ooded annually.

The valley £oor surfaces are underlain by 3^4 mof DeForest Formation alluvium (Bettis et al.,1992; Baker et al., 1996; Bettis and Autin,

1997). The alluvial ¢ll consists of three lithologi-cally distinct units, the Gunder, Roberts Creek,and Camp Creek members of the formation(Fig. 2). Table 2 summarizes the lithologic (tex-ture, color, bedding structures, pedogenic altera-tions) properties of DeForest Formation units inMoney and Pine Creek valleys.

Gunder Member deposits, averaging 3^4 m inthickness, underlie the two highest valley surfaces(T3 and T2). Gunder Member deposits beneathT3 are siltier and thicker than those beneath T2,and point bar gravels beneath T2 are thicker thanthose beneath T3. Basal radiocarbon ages onwood from the Gunder Member in Money andPine creeks range from 11 450 to 8120 yr B.P.beneath T3, and 6060 to 5090 yr B.P. beneathT2 (Table 1). The Roberts Creek Member aver-ages 2^3 m in thickness and underlies T1. RobertsCreek Member point bar gravels are thinner thanthose of the Gunder Member beneath T2, andsimilar in thickness to those of the Gunder Mem-ber beneath T3. Radiocarbon ages from the Rob-erts Creek Member range from 3850 to 1780 B.P.in these valleys (Table 1). Camp Creek Member

Table 1Radiocarbon ages and estimated post-settlement age from Money Creek and Pine Creek

Site 14C ages Unit Lab no. Cal. age

Money Cr., Site 6 post-settlement CC 100Money Cr., Site 9 1780 þ 70 RC ISGS-3622 1700Money Cr., Site 9 1820 þ 50 RC NSRL-3510 (CAMS-36636) 1720Pine Cr., Site 2A 3060 þ 70 RC ISGS-3432 3260Pine Cr., Site IVA 3110 þ 70 RC Beta-84021 3350Money Cr., Site 9, N. end 3320 þ 70 RC ISGS-3756 3510Money Cr., Site 1 3580 þ 70 RC ISGS-3220 3860Money Cr., Site 4 3830 þ 70 RC ISGS-3247 4230Money Cr., Site 15 3850 þ 80 RC ISGS-3625 4240Money Cr., Site 7 4390 þ 140 RC ISGS-3438 4930Money Cr., Site 10 5090 þ 70 LG ISGS-3245 5790Money Cr., Site 5 5340 þ 70 LG ISGS-3624 6140Money Cr., Site 13 5410 þ 70 LG ISGS-3623 6250Pine Cr., Site I 5540 þ 60 LG Beta-84023 6310Pine Cr., Site 3 5920 þ 60 LG CAMS-33983 6740Money Cr., Site 14 5980 þ 110 LG ISGS-3621 6830Money Cr., Site 3 6060 þ 70 LG ISGS-3217 6890Money Cr., Site 8 8120 þ 90 EG ISGS-3248 8990Money Cr., Site 9 8270 þ 100 EG ISGS-3763 9250Money Cr., Site 9, S end 8340 þ 210 EG ISGS-3755Money Cr., Site 2, top 11 290 þ 100 EG ISGS-3222 13 200Money Cr., Site 2, bottom 11 450 þ 80 EG ISGS-3221 13 360

EG = early Gunder, LG = late Gunder, RC = Roberts Creek, CC = Camp Creek



deposits underlie T0 (the modern channel belt)and lap up onto T1 and T2. Elsewhere in theUpper Midwest the bulk of the Camp CreekMember is Historic in age (Bettis et al., 1992;Beach, 1994; Baker et al, 1996; Bettis and Autin,1997). Although no datable material or artifacts

were recovered from the Camp Creek Memberduring this study, its lithology, distribution rela-tive to other DeForest Formation units, and gen-eral lack of pedogenic alteration suggest it is ofHistoric age.

The overall sedimentology of the Gunder and

Fig. 2. Photograph of section at Site 9, Money Creek, showing alluvial units and radiocarbon ages.

Table 2Lithologic and chronologic characteristics of DeForest Formation alluvium in Money and Pine creeks.

Camp Creek MemberVery dark grayish brown to yellowish brown (10YR3/2^5/4) silt loam to loam with thin sandy loam beds in abandoned channelareas; sand and gravel in the active channel; non-e¡ervescent; horizontally strati¢ed where greater than 0.25 m in thickness;surface soils (where present) are Entisols (A^C pro¢les); unit often buries pre-settlement surface soil ; thickest in and adjacentto channel belt (T0), and at the base of steep slopes. Not directly dated in this study; elsewhere in the Upper Midwest, USAca. 150 B.P. to present.Roberts Creek MemberVery dark gray to grayish brown (2.5Y3/0-10YR3/1-3/2) silty clay loam, and loam grading downward to sand and gravel;non-e¡ervescent; thick sections are horizontally strati¢ed at depth; detrital organic matter in lower part; relatively thick Mollisol(A^C or A^Bw^C pro¢le) developed in upper part; B horizon (if present) is dark colored; strong brown and yellowish red(7.5YR5/8, and 5YR5/8) mottles may occur throughout unit; found beneath T1, usually parallelling modern channel belt,ca. 4390^1000 B.P.Gunder MemberBrown to yellowish brown to grayish brown (10YR4/3-5/4 to 2.5Y5/2) silt loam (early Gunder) or brown to dark grayish brown(10YR4/3-4/2) loam (late Gunder) grading to sand and gravel at depth; non-e¡ervescent; lower part may be horizontally strati-¢ed; detrital organic matter often present in lower, unoxidized part of unit; thick Mollisol or Al¢sol (A^Bt^C, or A^E^Bt^Cpro¢le in early Gunder or A^Bw^C in late Gunder) developed in upper part; B horizon is brown to yellowish brown; C horizonusually contains strong brown (7.5YR5/8), yellowish red (5YR5/8), or dark brown (10YR3/3) mottles; comprises alluvial ¢llbeneath T3 (early Gunder) and T2 (late Gunder). Early Gunder 11,450 to 8120 B.P.; late Gunder 6060^5090 B.P.



Roberts Creek Members is similar. Both units arecharacterized by a 0.8^1.3-m-thick basal cobblegravel (lag) composed of clasts of local carbonateand sandstone that overlies bedrock. This channelbottom lag gravel maintains a uniform cross-val-ley elevation beneath all valley surfaces, indicatingthat channel bottom cutting and ¢lling have beeninsigni¢cant during the last ca. 11 500 yr. Thechannel bottom lag is overlain by about 0.3^0.7m of sandy and/or pebbly sand in-channel depos-its that grade upward to ¢ne-grained alluvium, orby 0.6^1.6-m-thick sandy gravel point bar depos-its overlain by ¢ne-grained alluvium. Thin hori-zontal strati¢cation is often present in the transi-tion zone from the sandy to the ¢ne-graineddeposits.

In nearby southwestern Wisconsin Knox (1985,1993) also found insigni¢cant Holocene channelbottom cutting and ¢lling, and used the modernchannel bed level as a reference against which past£ood depths could be compared. He found thatpoint bar deposits dating between 6000 and 4500yr B.P. were signi¢cantly thicker than those pre-ceding and post-dating that period, and con-cluded that the larger bars were formed during aperiod of relatively large £oods (Knox, 1985). In alater paper, he computed paleo£ood depths usingcobbles or boulders found in overbank £ood de-posits (£ood ramps) and found, in contrast to his1983 study, that the period between 3300 and5000 yr B.P. was characterized by relatively small£oods (Knox, 1993). Point bar gravels associatedwith Gunder Member alluvium beneath T2 (ca.6000^5000 yr B.P.; 6900^5800 yr ago) are thickerthan older or younger point bar deposits in bothMoney and Pine creeks. In addition, more £oodramp deposits, and larger average clast size in the£ood ramp deposits, are also associated with theGunder Member alluvium beneath T2. We con-clude that the larger point bars and more frequentand larger-clast £ood ramps associated with allu-vium dating from ca. 6000^5000 yr B.P. indicate aperiod of larger £oods. Larger £oods would alsopromote increased channel migration, channelbelt widening, and accelerated removal of olderalluvium. This would create many £oodplain en-vironments favorable for pioneer species. Pointbar and £ood ramp deposits dating to the periods

11 450^8200 and after 4000 yr B.P. suggest small-er, less competent £oods during these periods.The Historic period witnessed a return to larger£oods, which deposited alluvium over all but thehighest valley surfaces.

Plant macrofossils are preserved primarily inthe sandy in-channel deposits and in the overlyingstrati¢ed transitional deposits. The thickest plantmacrofossil-bearing deposits (approximately 0.5m thick) are associated with thick transitionzone deposits. These deposits accumulated inpools, especially parts of pools opposite thedownstream end of point bars. Fine-grained de-posits and organic remains accumulate in poolsslowly under relatively quiet water conditions dur-ing average and low-£ow conditions. The ¢nesand organics are rapidly buried during higher£ow conditions as sediment-laden runo¡ entersthe pool and drops its sediment load (includingmore organic remains) in response to decreases invelocity, turbulence, and gradient in the poolarea. This is especially prevalent in the £ow sep-aration zone on the downstream end of pointbars. Thick transition zone deposits have beenrecorded in northeastern Iowa fossil localities im-mediately upstream of fossil beaver dams (Bakeret al., 1996). Though no fossil beaver dams depos-its were encountered in Money or Pine Creek, it islikely that some of the thicker fossil deposits maybe associated with the pool area behind beaverdams. Gradual channel migration across the£oodplain, and meander abandonment by chuteand neck cuto¡s remove fossil localities from theactive channel area, and once isolated the local-ities are further buried by overbank deposits dur-ing £oods.

The sedimentology of the Camp Creek Memberdi¡ers slightly from that of the prehistoric alluvi-um in the basin. Most of the Camp Creek Mem-ber in the Upper Midwest accumulated followingextreme agricultural disturbance and erosion ofuplands and valley wall slopes in the late 1800'sand up to about 1930 (Knox, 1987; Beach, 1994;Baker et al., 1996; Bettis and Autin, 1997). CampCreek Member alluvium mantles all valley surfa-ces except for T3, increasing in average thicknessfrom 0.1^0.54 m where it buries the GunderMember beneath T2, to 1.0 m where it buries



the Roberts Creek Member beneath T1. Overlap-ping of Camp Creek Member alluvium onto olderHolocene valley surfaces has reduced the localrelief on the valley £oors and obscured scarpsthat formerly marked contacts between terracelevels. In overbank settings Camp Creek Memberalluvium accumulated during £oods and does notcontain macrobotanical remains. Macrobotanicalremains from the Camp Creek Member were dis-covered at only one site, an abandoned channel,during this study. They have been discovered else-where only in abandoned-channel environmentsthat were short lived and characterized by veryrapid alluviation (Baker et al., 1996).

Gunder and Roberts Creek Member depositsboth encompass alluvium associated with manychannel positions over several thousand years.Through time channel activity has continuouslyremoved older alluvium and associated biotic re-mains. The remaining alluvial ¢ll of the valley istherefore a complex mosaic of alluvium of variousages. Apparent gaps in the depositional record inthis study probably represent sampling and/orpreservation bias in a small area rather than re-gional patterns. The stratigraphic framework ofthe DeForest Formation permits ¢eld recognitionof three broad age groups of alluvium in the val-ley; that deposited between about 11 500 and5000 yr B.P. (Gunder Member), from 4500 toabout 150 yr B.P. (Roberts Creek Member), andhistoric alluvium (Camp Creek Member). Di¡er-ences in the morphology of soils developed intothe Gunder Member and textural variations ofthe alluvium are used to distinguish early andlate Gunder Member localities (Table 2). GunderMember alluvium with basal ages ranging fromabout 11 500^8000 yr B.P. (early Gunder) is siltloam to light loam texture and has surface soilswith Bt horizons developed into it. Soils devel-oped in Gunder Member alluvium dating fromca. 6000^5000 yr B.P. lack Bt horizons and thealluvium is loam texture. Roberts Creek Memberalluvium is signi¢cantly darker colored and con-tains more organic matter than the Gunder Mem-ber. Soils developed in the upper part of the Rob-erts Creek Member have A^Bw pro¢les withdarker colored Bw horizons than soils formed in

late Gunder Member alluvium. These lithologicand pedologic distinctions allow us to estimatethe age of fossil localities in the ¢eld and samplethe valleys' alluvium such that localities spanningthe Holocene are represented.

Post-depositional alterations have di¡erentiallya¡ected the Gunder and Roberts Creek membersand associated biotic remains. Knox (1983) dis-cussed the Holocene climatic history of the UpperMidwest as it relates to the behavior of the £uvialsystem and Bettis (1992) discussed the role oflong-term water table relationships on the fossilrecord in Holocene alluvium. As a result of latemiddle Holocene (ca. 5500 yr B.P.), reductions inlong-term precipitation, and concomitant lower-ing of the water table, Gunder Member depositshave been oxidized to a greater degree than Rob-erts Creek Member deposits (Bettis et al., 1992;Baker et al, 1992; 1996). Pollen and plant macro-fossils are absent from the oxidized portion of theGunder Member because of post-depositionaldegradation, are poorly preserved in the mottledand reduced zone that marks the zone of long-term water table £uctuation, and are well pre-served only in the lower unoxidized portion ofthe unit that has remained saturated since depo-sition (Bettis et al., 1992; Baker et al., 1996). Or-ganic remains are preserved through a greaterthickness of the Roberts Creek Member becausethe water table has remained high since the unitwas deposited, the lower half to third of the unithas remained saturated, and organic matter deg-radation has therefor been minimal. In both Mon-ey and Pine creeks a prominent zone of iron stain-ing in Gunder Member alluvium separatesoverlying oxidized alluvium from underlying mot-tled and sometimes reduced alluvium (Fig. 2). Theiron-stained zone marks the average position ofthe water table subsequent to accumulation ofthe late Gunder Member, and probably marksthe late Holocene water table level. The iron ac-cumulation zone is not as evident in RobertsCreek Member exposures, but where it can betraced from adjacent Gunder Member exposures,such as that shown in Fig. 2, it generally marksthe upper limit of the zone of pollen and plantmacrofossil preservation.



Fig. 3. Pollen diagram from Money and Pine Creeks. Sites are not in stratigraphic position, but have been arranged by individualradiocarbon dates from each cutbank section.

Fig. 4. Plant macrofossils of boreal plants and deciduous trees from Money and Pine Creeks.



4.2. Vegetation history

Although the coverage of radiocarbon-datedsites is sparse between 11 450 and 6000 yr B.P.(early Holocene), and between 1850 yr B.P. (lateHolocene) and post-settlement time, coverage issu¤cient to answer key questions about regionalHolocene vegetational patterns. The early Holo-cene was a time of substantial change from forestto prairie at Kirchner Marsh (Fig. 1) and at othersites north and west. In contrast, at RobertsCreek and sites south and east, this interval wasdominated by deciduous forest, with relatively mi-nor changes within forest types. The last ca. 2000yr were also fairly stable throughout the Midwest.The dated sites are su¤cient to determine whethersoutheastern Minnesota followed the pattern ofKirchner Marsh and sites north and west, orthat of Roberts Creek and sites south and east(Fig. 1).

4.3. Pollen

The pollen diagram for Money and Pine Creeksis divided into four zones by visual inspection.Zone 1 (11 450^10 000 yr B.P.), represented byonly one site, is dominated by Picea (spruce)and Larix (larch) pollen (Fig. 3). A second site(not shown) was also dominated by these taxa,but extensive rootlet contamination prevented usfrom obtaining a radiocarbon age. The upperzone boundary marking the end of spruce domi-nation is arbitrarily chosen because of the lack ofsites showing the transition; approximately repre-sents the beginning of the next regional pollenzone.

Zone 2 (10 000^5400 yr B.P.) at Money Creek isdominated by Quercus (oak) and Ulmus (elm)with relatively high values (for these taxa) ofAcer saccharum, Juglans cinerea, Juglans nigra,(black walnut) and Tilia americana (Fig. 3). How-

Fig. 5. Fossil fruits or seeds of forest herbs from Money and Pine Creeks.



ever, no fossiliferous sites were found with agesbetween 11 290 and 8270 yr B.P. (Fig. 3), and thisinterval is so labeled.

The boundary between Zones 2 and 3 at ca.5400 yr B.P. is sharply de¢ned by precipitousdrops in the pollen of deciduous trees and in totalarboreal pollen along with the rise in non-arbor-eal pollen, including Artemisia (wormwood) Am-brosia (ragweed), Poaceae (grass), and Asteraceae(composites) (subfamily Asteroideae) (Fig. 3).Zone 3 (5500^3400 yr B.P.) is characterized bymaxima of these prairie elements. Arboreal taxa,mainly Quercus, recover somewhat in Zone 4,where arboreal and non-arboreal pollen percen-tages are about equal.

4.4. Plant macrofossils

Plant macrofossils are abundant in these sedi-

ments, and they reveal a detailed picture of theHolocene environments in the area. The mostabundant single taxon is Picea, whose needlesreach ca. 2500/l in Zone 1. The most abundantand diverse group are the wetland taxa, a patternthat seems predominant in midwestern streamcutbank sites. These taxa dominate on the near-stream £oodplain environment throughout theHolocene. However, su¤cient numbers of uplandtaxa are present to reconstruct some upland hab-itats.

Zone 1 is strongly dominated by macrofossilsof boreal and subarctic plants including Betulaglandulosa (dwarf birch), Selaginella selaginoides(spikemoss), Picea mariana (black spruce; coneswere found), and Larix laricina (Fig. 4). Tilia, adeciduous forest element, occurs in this zone,along with such forest £oor taxa as Claytonia cf.virginica (spring beauty), and Cryptotaenia cana-

Fig. 6. Macrofossils of prairie plants from Money and Pine Creeks. All are fruits or seeds unless otherwise indicated.



Fig. 7. Fossil fruits or seeds of Holocene weeds from Money and Pine Creeks.

Fig. 8. Fruits or seeds of weeds, mainly introduced taxa, present only in post-settlement deposits from Money and Pine Creeks.



densis (honewort) (Figs. 4 and 5). Several weedy,wetland and aquatic taxa are also represented, butthe diversity of this zone is low.

Zone 2 is characterized by deciduous foresttaxa. Quercus cf. macrocarpa, Tilia americana, Ul-mus americana, Carpinus caroliniana, and Ostryavirginiana are all well-represented in this zone(Fig. 4). In addition, a few tattered specimens ofPicea and Larix needles and Selaginella selagi-noides megaspores are present; these taxa arecommonly found in mid-Holocene stream andeven lake deposits in the Midwest, and they arealmost certainly redeposited. Forest herbs, espe-cially Aralia racemosa, (spikenard), Laportea can-adensis (wood nettle), and Eupatorium rugosumtype (white snakeroot), are abundant in the earlyHolocene (Fig. 5), along with the forest-edgeplants Crataegus succulenta type (hawthorn), andRubus (raspberry, blackberry). Monarda ¢stulosa(wild bergamot), a somewhat weedy prairie spe-

cies, appears in the early Holocene samples, alongwith Heliopsis helianthoides (oxeye, which cangrow in open woodlands) and Helianthus grosse-serratus type (sawtooth sun£ower, but type in-cludes H. maximiliani and possibly others thatmay or may not be prairie species) (Fig. 6). Twoof the dominant prairie grasses, Schizachyriumscoparium (Andropogon scoparius, little bluestem) and Sorghastrum nutans (Indian grass) ap-pear only after 6000 yr B.P. at the top of Zone 2.Several weedy species are abundant in Zone 2 aswell (Fig. 7), although many reach their peaks inpost-settlement horizons (Fig. 8). Verbena hastata(common vervain), Verbena urticifolia (white ver-vain), Urtica dioica (nettle), and Chenopodium ber-landieri type (goosefoot) are abundant ruderaltaxa in Zone 2 (Fig. 7). A suite of wetland plantsincluding Boehmeria cylindrica (false nettle),Caltha palustris (marsh-marigold), Pilea pumila(clearweed), and Sagittaria latifolia (arrowhead),

Fig. 9. Fruits or seeds of wetland taxa with maximum abundance in the early Holocene from Money and Pine Creeks.



characterizes the early Holocene and becomes lessimportant or disappears in the middle and lateHolocene (Fig. 9). Aquatic plants are also abun-dant in this zone.

Prairie species, including Monarda ¢stulosa,Sorghastrum nutans, Amorpha canescens (leadplant), Andropogon gerardii (big bluestem), Des-modium canadense (tick trefoil), Pycnanthemum(mountain mint), and Rudbeckia hirta (black-eyed susan), are concentrated in Zone 3 (Fig. 6).Macrofossils of deciduous trees and most forestherbs decline or disappear in this zone (Figs. 4and 5). Several weedy taxa continue to be present,and others are limited or nearly limited to Zone 3(Fig. 7). Wetland and wet prairie species includingLycopus americanus (water-horehound), Glyceriastriata (slender manna grass), Epilobium colora-tum (willow herb), Helenium autumnale (sneeze-weed), Scirpus validus type (soft-stem bulrush),

Scirpus atrovirens (bulrush) and others are prom-inent in this interval (Figs. 10 and 11). Aquaticplants are common at the very beginning of thiszone, but are rare thereafter.

Zone 4 is sparsely represented in the record,with samples only at 3110, 3060, and 1800 yrB.P., along with two modern and one pre-mod-ern but post-settlement sample. Nevertheless,trees clearly return to the record about 3000 yrB.P., including Tilia, Ulmus americana and Car-pinus caroliniana, and Betula lutea (yellow birch)and Betula papyrifera (paper birch) make their¢rst appearance (Fig. 4). The latter two are gen-erally found in the conifer^hardwood forest inthe northern Midwest, and they have been ad-vancing southward during the late Holocene else-where in the Midwest (Webb et al., 1983). Forestherbs are also present mainly at the base of thiszone, with ¢rst and only occurrences of Aralia

Fig. 10. Fruits or seeds, unless otherwise indicated, of wetland taxa with maximum abundance in the middle Holocene fromMoney and Pine Creeks.



nudicaulis (wild sarsaparilla), Eupatorium altissi-mum (tall eupatorium), and Triosteum perfolia-tum/auranticum (horse-gentian), and shrubs arealso mostly limited to the base of this zone(Fig. 5). Prairie plants continue to be well repre-sented, and Hypoxis hirsuta (yellow stargrass)and Ratibida pinnata (gray-headed cone-£ower)¢rst occur in Zone 4 (Fig. 6). Weeds are sparsewithin the Zone, but they are the most abundantand diverse group in post-settlement and modernsamples at the top of the Zone (Figs. 7 and 8).Wetland plants continue to be abundant in thisinterval, and Helenium autumnale, Typha, (cat-tail), Eupatorium maculatum (Joe-pye-weed), andSium suave (water parsnip) reach high values at1800 yr B.P. (Fig. 11). Aquatic plants are rare inZone 4 and absent in modern and post-settle-ment sediments. Their absence is surprising, be-cause a lush but almost wholly non-nativeaquatic £ora including Potamogeton crispus(pondweed) and Veronica anagallis-aquatica

(water speedwell), grows in parts of the modernMoney Creek.

4.5. Speleothems

The paleoclimatic signi¢cance of Cold WaterCave stalagmite 1S is discussed in detail in Doraleet al. (1992) and Denniston et al. (1999a). In gen-eral, a cool and/or moist period from 7770 to5900 cal. yr B.P. is characterized by low speleo-them N13C and N18O values. N13C values increasesteadily beginning at 5900 cal. yr B.P. whereasN18O begins increasing at 5700 cal. yr B.P. Theseshifts were interpreted by Dorale et al. (1992) tore£ect rising middle Holocene temperatures andthe replacement of deciduous forest by prairie.N13C and N18O values begin decreasing ca. 3300cal. yr B.P. as trees reinvaded the prairie asoak-dominated savanna began to appear (seeFig. 12 and Table 1 for conversion of 14C to cal-endar yr). By 2000 cal. yr B.P., soil organic matter

Fig. 11. Fruits or seeds, unless otherwise indicated, of other wetland taxa from Money and Pine Creeks.





was once again dominated by low N13C values,indicating that the savanna was well established.

Spring Valley stalagmite SV-1 records nearlyidentical speleothem carbon and oxygen isotopiccompositions throughout the same interval. Theprimary di¡erence from Cold Water Cave is thetiming of the onset of higher N13C values. Threestalagmites from this cave (Denniston et al.,1999a) exhibit a synchronous increase in N13C val-ues beginning about 5900 cal. yr cal. B.P. Thesechanges re£ect the increased contribution of C4

prairie grasses to the soil organic material atSpring Valley, the shift toward elevated carbonand oxygen isotopic compositions and increasedC4 contributions occur at about 8000 cal. yr B.P.,2000 yr earlier than at Cold Water Cave. At ap-proximately 3300 cal. yr at Cold Water Cave and3000 cal. yr B.P. at Spring Valley, N13C valuesbegin to decline, indicating partial replacementof C4 grasses by C3 plants.

5. Discussion

5.1. Late-glacial and early Holocene vegetation

The late-glacial environment in southeasternMinnesota, represented in Zone 1 at MoneyCreek, was dominated by Picea and Larix grow-ing in a boreal wetland. Sediment at Site 2 con-tained numerous needles of these taxa, but alsocones of Picea mariana, and macrofossils of Betu-la glandulosa and Selaginella selaginoides. Thesetaxa indicate cold boreal to subarctic conditionsin a fen or bog 11 50 yr B.P., an interpretation inagreement with those at many other midwesternsites. This interval is not represented in the spe-leothem records.

The boundary between Zones 1 and 2 is unclearbecause of the lack of samples between 11 290 and8270 yr B.P., but it undoubtedly falls within thatinterval. Pollen of Picea and Larix declined and

were replaced by a succession of mainly deciduoustrees about 10 200 yr B.P. at Kirchner Marsh,about 10 000 yr B.P. at Clear Lake, and between10 000 and 9000 yr B.P. at Roberts Creek. Thesedates suggest that the zone boundary is likely ca.10 000 yr B.P.

In Zone 2 deciduous tree pollen and plant mac-rofossils are abundant until after 5500 yr B.P.,and pollen of prairie taxa is scarce. The presenceof several taxa of deciduous trees in the macro-fossil record indicates that Zone 2 clearly repre-sents deciduous forest that was present from atleast 8270 yr B.P. until about 6000 yr B.P. Someof the weedier, more aggressive prairie taxa (Mo-narda ¢stulosa (wild bergamot), Helianthus, Heli-opsis helianthoides) appeared as early as 8270 yrB.P. Weeds in this Zone include Verbena hastata,V. urticifolia, Urtica dioica, Polygonum scandens(false buckwheat), and Acalypha (three-seededmercury). These prairie taxa and weeds all cangrow in open woods, along woodland margins,or in £oodplain openings. The high arboreal pol-len and presence of trees in the macrofossil recordindicates that forests were still widespread untilshortly after 6000 yr B.P.

The transition between Zones 2 and 3 between6000 and 5500 yr B.P. marks the regional declineof the forest. The drop o¡ of Ulmus pollen per-centages occurred about 6000 yr B.P., (just whenQuercus pollen was at its peak) even though Ul-mus bud scales remained abundant until 5500 yrB.P. This pattern may indicate that as climatebecame drier at 6000, elm was no longer presentin the upland forest (hence its decline on the pol-len diagram), but it survived and produced budscales along the £oodplain. The increase in oakmay be explained by expansion of Quercus macro-carpa, the most drought-tolerant oak, in uplandforests following the decrease in elm. Some prairieplant macrofossils became prominent at 6000 yrB.P., including two of the dominant C4 prairiegrasses (Schizachyrium and Sorghastrum). These

Fig. 12. Carbon and oxygen isotopes from stalagmites from Cold Water Cave and Spring Valley Cave, compared with the tree-herb percentages at Kirchner, Money Creek, and Marsh Roberts Creek. Intervals on isotopic curves dated by U^Th series analy-sis are shown with error bars. Calibrated radiocarbon ages are shown between the Kirchner Marsh and Money Creek pollencurves to aid in comparison with the timing of the U^Th ages of the speleothems.



species were apparently able to become estab-lished locally on dry sites, but were not su¤cientlywidespread to be prominent on the pollen dia-gram. After 5500, the deciduous forest died backover much of the area.

5.2. Late Holocene vegetation

Zone 3 has the best coverage of sites, and itclearly shows the magnitude of changes that oc-curred after 5500 yr B.P. Although the diversityand abundance of prairie species is not as great asoccurs further west (Baker, 2000), they nonethe-less indicate a diverse prairie environment (similarto modern prairie remnants) in the late^middleHolocene. The large drop in tree pollen and theincreases in the pollen of `prairie' generalists, Ar-temisia (at one site), Ambrosia, and Poaceae atabout 5500 yr B.P. (Fig. 3) indicate that prairiereplaced forest in that interval at a regional scale.The dominance of macrofossils of prairie species,the near absence of macrofossils of trees and for-est herbs in Zone 3 (Fig. 4^6), and the N13C valuesin speleothems from the two caves support thisinference. This trend is remarkably similar in tim-ing and magnitude to the sharp decline in arbor-eal pollen and associated changes in plant macro-fossils at Roberts Creek (Fig. 12).

The sparse coverage of the last 3500 yr B.P.prevents a detailed interpretation, but the increasein arboreal pollen is a regional event between3500 and 3000 yr B.P. Further north in centralMinnesota, Grimm (1984) has demonstrated that¢re played an important role in the changing po-sitions of the prairie-forest border during the lateHolocene; ¢re was probably important in south-eastern Minnesota as well, but our data neithercon¢rm nor refute this hypothesis. The enormousincreases in macrofossils of weeds in the historicalsediments (Figs. 7 and 8), including a number ofalien species, demonstrates that the e¡ects ofEuro^American settlement on the landscapewere extremely important, as they were at RobertsCreek (Baker et al., 1993).

5.3. Speleothems and alluvium

The narrow mid-Holocene ecotone between

prairie and forest ¢rst suggested in Iowa by Bakeret al. (1990, 1992) is supported by these fossil andspeleothem data and extended into southeasternMinnesota. Sharp rises in carbon isotopic valuesat Spring Valley 8000 calendar yr B.P. indicateprairie environments. This chronology (when ra-diocarbon yr are corrected to calendar years) iscomparable in timing to the changes from forestto prairie at Kirchner Marsh (Fig. 12) and inClear Lake (Baker et al., 1992), except that thedate of prairie expansion at these sites is 9000 cal.yr B.P., rather than 8000 cal. yr B.P. as shown inthe speleothems. We are uncertain of the cause ofthis di¡erence. It is possible that this sharp eco-tone curved between these sites and re£ects realdi¡erences in timing. Similar changes occur atCold Water Cave at ca. 5900 calendar yr(w5500 radiocarbon yr) B.P., which correlateswith the later change from forest to prairie atRoberts Creek (Fig. 12). Denniston et al.(1999b) argue that this early to mid-Holoceneecotone was maintained by di¡erences in moisturealong the boundary. Furthermore, stalagmitegrowth rates and N18O changes (Denniston etal., 1999a) suggest that there was a pronouncedchange from summer dominated to winter domi-nated precipitation, rather than a decrease in an-nual precipitation.

The texture of overbank alluvium changedfrom silty to loamy between about 7000 and6000 yr B.P. in these valleys. This change mayre£ect either an increasing contribution ofsandy/loamy sources as the loess cover was pro-gressively removed from uplands and slopes, or ashift to more competent £ows that depositedcoarser overbank alluvium. The fact that the larg-est point bars in these valleys formed between6000 and 5000 yr B.P., and that £ood ramps arealso more frequent in alluvium of this age, indi-cates that a late middle Holocene (ca. 6000^5000yr B.P.) period of large £oods occurred in thesevalleys. If the cool-season precipitation peak fellin the spring, as suggested by the speleothem rec-ord, then most plants would be dormant and run-o¡ would be greater. Runo¡ is also considered tobe greater in prairie than in forested environ-ments. This is the time when prairie became es-tablished (Fig. 6) along the creek, and there was



also a greater diversity in weeds, which also sug-gest more disturbed, open ground (Fig. 7), aswould be expected with larger £oods, more activechannel migration, and channel widening. Thetrends in wetland plants also strongly re£ectchanges that were occurring. It is unclear fromthe associations why they changed or what thecauses of the changes in wetland habitats were,but presumably they re£ect such things as fre-quency and magnitude of £ooding, changes inthe local water table, and other factors.

5.4. Correlation with nearby sites

The entire Money Creek and Pine Creek se-quence correlates almost exactly with the RobertsCreek sequence in northeastern Iowa, despite thefact that the Minnesota sites are in deeply dis-sected terrain, whereas the northeastern Iowa sitesare in a relatively low relief area. Both had late-glacial spruce-larch forests followed by rich decid-uous forest. Both sites had prairie establishmentdelayed until 5500 yr B.P., followed by return ofpredominantly oak (with continuing prairie taxa).This assemblage is interpreted to represent savan-na that became established between 3500 and3000 yr B.P., and remained until settlementtime. Kirchner Marsh and Clear Lake have a sim-ilar early history, but they are on the west side ofthe proposed mid-Holocene ecotone, where prai-rie began replacing deciduous forest earlier, ca.8000 yr B.P. The return of oaks about 3500^3000 yr B.P. is similar at all sites in the area.

6. Conclusions:

1. Evidence from pollen, plant macrofossils,and speleothem isotopes strongly support the ex-istence of a sharp ecological boundary betweenprairie on the west and forest on the east, extend-ing northward from eastern Iowa into southeast-ern Minnesota, between about 8000 and 6000 yrB.P.

2. Combined pollen and plant macrofossilsfrom stream cutbank exposures provide a depend-able, detailed record of vegetational change that isconsistent between cutbank sites in eastern Iowa

and southeastern Minnesota. These deposits canprovide species-level paleoenvironmental evidencein areas where no other fossil deposits exist.

3. Speleothem isotopes provide a high-resolu-tion paleoclimatic and paleoecologic record thatstands by itself, but also complements the evi-dence from stream alluvium and its associatedplant remains.

4. Inferred changes in the seasonality of precip-itation and the vegetation coincide with a changein stream behavior. In the mid-Holocene, pre-dominantly cool-season precipitation, warm sea-son droughts, and a change from forest to prairie(which all increase runo¡) were accompanied bylarger £oods.

Acknowledgements

We thank Dr. Joe Mason for bringing the PineCreek sites to our attention and Cara Day for helpin sieving and picking macrofossils. Harry andSteve Walsky kindly allowed us to collect sampleson their property. Thanks to Ken and WandaFlatland at Cold Water Cave and John Ackermanat Spring Valley Cave. This work was supportedby NSF Grant EAR-93-16391 and NOAA GrantNA366PO238.

References

Baker, R.G., 2000. Holocene environments reconstructed fromplant macrofossils in stream deposits from southeastern Ne-braska. The Holocene 10, 357^365.

Baker, R.G., Chumbley, C.A., Witinok, P.M., Kim, H.K.,1990. Holocene vegetational changes in eastern Iowa.J. Iowa Acad. Sci. 97, 167^177.

Baker, R.G., Maher, L.J., Chumbley, C.A., Van Zant, K.L.,1992. Patterns of Holocene environmental change in theMidwest. Q. Res. 37, 379^389.

Baker, R.G., Schwert, D.P., Bettis, A.E., III, 1993. The impactof EuroAmerican settlement on landscapes, vegetation andwater quality in northeastern Iowa. Holocene 3, 314^323.

Baker, R.G., Bettis, E.A., III, Schwert, D.P., Horton, D.G.,Chumbley, C.A., Gonzalez, L.A., Reagan, M.K., 1996. Ho-locene paleoenvironments of northeast Iowa. Ecol. Monogr.66, 203^234.

Baker, R.G., Gonzalez, L.A., Raymo, M., Bettis, E.A., III,Reagan, M.K., Dorale, J.A., 1998. Comparison of multiple



proxy records of Holocene environments in midwesternUSA. Geology 26, 1131^1134.

Barnosky, C.W., Grimm, E.C., Wright, H.E., Jr., 1987. To-wards a postglacial history of the northern Great Plains: Areview of the paleoecologic problems. Ann. Carnegie Mus.56, 259^273.

Beach, T., 1994. The fate of eroded soil: Sediment sinks andsediment budgets of agrarian landscapes in southern Minne-sota, 1851^1988. Ann. Assoc. Am. Geogr. 84, 5^28.

Bettis, E.A. III, 1992. Soil Morphologic properties and weath-ering zone characteristics as age indicators in Holocene allu-vium in the Upper Midwest. In: V.T. Holliday, (Ed.), Soilsin Archaeology. Smithsonian Institution Press, Washington,DC, pp. 119^144.

Bettis, E.A., III, Autin, W.J., 1997. Complex response of amidcontinent North America drainage system to late Wis-consinan sedimentation. J. Sediment. Res. 67, 740^748.

Bettis, E.A. III, Baker, R.G., Green, W.R., Whelan, M.K.,Benn, D.W., 1992. Late Wisconsinan and Holocene alluvialstratigraphy, paleoecology, and archaeological geology ofeast-central Iowa. Iowa Department of Natural ResourcesGuidebook Series No. 12. Iowa Department of Natural Re-sources, Iowa City, IA.

Boutton, T.W., 1991. Stable carbon isotope ratios of naturalmaterials. II. Atmospheric, terrestrial, marine, and fresh-water environments. In: Coleman, D.C., Fry, B. (Eds.), Car-bon Isotope Techniques. Academic Press. San Diego, CA,pp. 155^171.

Bradbury, J.P., Dean, W.E., 1993. Elk Lake, Minnesota: Evi-dence for rapid climate change in the North-Central UnitedStates. Geological Society of America Special Paper 276,Geological Society of America, Boulder, CO.

Chen, J., Edwards, L., Wasserburg, W., 1986. 238U, 234U, and232Th in seawater. Earth Planet. Sci. Lett. 80, 241^251.

Chumbley, C.A., Baker, R.G., Bettis, E.A., III, 1990. Mid-western Holocene paleoenvironments revealed by £oodplaindeposits in northeastern Iowa. Science 249, 272^274.

Cushing, E.J., 1965. Problems in the Quaternary phytogeogra-phy of the Great Lakes Region. In: Wright, H.E. Jr., Frey,D.G. (Eds.), The Quaternary of the United States. PrincetonUniversity Press, Princeton, NJ, pp. 403^416.

Denniston, R.F., Gonzalez, L.A., Asmerom, Y., Baker, R.G.,Reagan, M.K., Bettis, E.A., III, 1999a. Evidence for in-creased cool season moisture during the middle Holocene.Geology 27, 815^818.

Denniston, R.F., Gonzalez, L.A., Baker, R.G., Asmerom, Y.,Reagan, M.K., Edwards, R.L., Alexander, C.E., 1999b. Spe-leothem evidence for Holocene £uctuations of the prairie-forest ecotone, north-central USA. The Holocene 9, 671^676.

Dorale, J.A., Gonzalez, L.A., Reagan, M.K., Pickett, D.A.,Murrell, M.T., Baker, R.G., 1992. A high-resolution recordof Holocene climate change in speleothem calcite from ColdWater Cave, northeast Iowa. Science 258, 1626^1630.

Edwards, R.L., Beck, J.W., Burr, C.S., Donahue, D.J., Chap-pell, J.M.A., Bloom, A.L., Dru¡el, E.R.M., Taylor, F.W.,1993. A large drop in atmospheric 14C/12C and reduced

melting in the Younger Dryas documented with 230Th agesof corals. Science 260, 962^968.

Edwards, R.L., Chen, J.H., Wasserburg, G.J., 1987. 238U^234U^230Th systematics and the precise measurement oftime over the last 500 000 years. Earth Planet. Sci. Lett.81, 175^192.

Faegri, K, Kaland, P.E., Krzywinski, K., 1989. Textbook ofPollen Analysis. John Wiley and Sons, New York.

Gleason, H.A., 1922. The vegetational history of the MiddleWest. Assoc. Am. Geogr. Ann. 12, 39^85.

Grimm, E.C., 1984. Fire and other factors controlling the BigWoods in the mid-nineteenth century. Ecol. Monogr. 54,291^311.

Hendy, C.H., Wilson, A.T., 1968. Palaeoclimatic data fromspeleothems. Nature 216, 48^51.

Hobbs, H.C., 1995. Sur¢cial geology (Plate 3). In: Lively,R.S., Balaban, N.H., (Eds.) Geologic atlas, FillmoreCounty, Minnesota. Minnesota Geological Survey CountyAtlas Series C-12, Pt. A, scale 1:100, 000. Minnesota Geo-logical Survey, Minneapolis, MN.

Knox, J.C., 1983. Response of river systems to to Holoceneclimates. In: H.E. Wright Jr., (Ed.) Late-Quaternary Envi-ronments of the United States. Vol. 2: The Holocene. Uni-versity of Minnesota Press, Minneapolis, MN, pp. 26^41.

Knox, J.C., 1985. Responses of £oods to Holocene climaticchange in the Upper Mississippi Valley. Q. Res. 23, 287^300.

Knox, J.C., 1987. Historical valley £oor sedimentation in theUpper Mississippi River Valley. Ann. Assoc. Am. Geogr.77, 224^244.

Knox, J.C., 1993. Large increases in £ood magnitude in re-sponse to modest changes in climate. Nature 361, 430^432.

Knox, J.C., Attig, J.W., 1988. Geology of the Pre-Illinoiansediment in the Bridgeport Terrace, Lower Wisconsin RiverValley, Wisconsin. J. Geol. 96, 505^513.

McAndrews, J.H., 1966. Postglacial history of prairie, savan-na, and forest in northwestern Minnesota. Torrey BotanicalClub Memoir 22, 72 pp.

Prior, J.C., 1991. Landforms of Iowa. University of IowaPress, Iowa City, IA.

Soil Survey Sta¡, 1993. Soil Survey Manual, U.S. Departmentof Agriculture Handbook No. 18, Washington, DC.

Strickland, L.E., 1998. Taphonomy of plant remains in PrairieCreek, Nerstrand Big Woods State Park, Rice County, Min-nesota: A modern analog for paleoenvironmental interpre-tation. M.S. Thesis, University of Iowa, Iowa City, IA.

Watts, W.A., Winter, T.C., 1966. Plant macrofossils fromKirchner Marsh, Minnesota ^ A paleoecological study.Geol. Soc. Am. Bull. 77, 1339^1360.

Webb, T. III, Cushing, E.J., Wright, H.E. Jr., 1983. Holocenechanges in the vegetation of the midwest, In: Wright, H.E.Jr. (Ed.), Late Quaternary Environments of the UnitedStates, Vol. 2, The Holocene. University of Minnesota Press,Minneapolis, MN, pp. 143^165.

Wright, H.E. Jr., 1968. History of the Prairie Peninsula. In:Bergstrom, R.E. (Ed.), The Quaternary of Illinois. Univer-sity of Illinois College of Agriculture Special Publication no.14, Urbana, IL, pp. 78^88.



Documents

Holocene paleoenvironments in southeastern Minnesota ...people.ku.edu/~lgonzlez/NewFiles/Publications/Bakeretal02.pdf · southeastern Minnesota, USA. Early Holocene forests were replaced