Patterns of Holocene environmental change in the midwestern United States

Patterns of Holocene environmental change in the midwestern United States

QUATERNARY 31, 379-389 (1992) RESEARCH Patterns of Holocene Environmental Change United States RICHARD of Geology Departments and Botany, in ...

2MB Sizes 0 Downloads 19 Views


31, 379-389 (1992)



of Holocene

Environmental Change United States RICHARD

of Geology


and Botany,

in the Midwestern

G. BAKER Universiry

of Iowa,





LOUIS J. MAHER Ljepartment

of Geology,


of Wisconsin,





York State




The State












2419 Amoco


ZANT 1670 Broadway,




Received January 3, 1991 Four pollen sequences along a transect from north-central Iowa to southeast Wisconsin reveal the distribution of prairie and forest during the Holocene and test the use of pollen isopolls in locating the Holocene prairie-forest border. Prairie was dominant in central Iowa and climate was drier than present from about 8000 to 3000 yr B.P. During the driest part of this period in central Iowa ([email protected] yr B.P.), mesic forest prevailed in eastern Iowa and Wisconsin, suggesting conditions than at present. Prairie replaced the mesic forest about 5400 yr B.P. in eastern Iowa but did not extend much farther east; mesic forests were replaced in southern Wisconsin and northern Illinois about 5400 yr B.P. by xeric oak forests. This change from mesic to xeric conditions at 5400 yr B.P. was widespread and suggests that the intrusion of drier Pacific air was blocked by maritime tropical air from the Gulf of Mexico until the late Holocene in this area. 0 1992 University

of Washington.

INTRODUCTION Models of Holocene vegetational change in the midwestern United States first appeared more than 50 yr ago with Transeau’s (1935) description of the prairie peninsula (Fig. la). Borchert (1950) and Bryson (1966) pointed out that the modern distribution of prairie is related to climatic patterns, especially the frequency of dry Pacific air masses, and Webb has tried to characterize the modern prairie-forest boundary by pollen isopolls (Fig. lb) (Bernabo and Webb, 1977; Webb et al., 1983). Prairie was thought to have extended eastward into Indiana and Ohio during a warmer, drier middle Holocene episode (Wright, 1968):; when cooler, more humid

conditions returned in the late Holocene, the patches of prairie that characterize the prairie peninsula remained in these states. Fossil evidence seemed to support this picture in Minnesota, where sensitive ecotonal sites like Kirchner Marsh (Fig. 2) (Wright et al., 1963; Watts and Winter, 1966) and McAndrews’ (1966) Itasca transect in northwestern Minnesota showed clear evidence of an early to middle Holocene dry period (Holloway and Bryant, 1985; Webb et al., 1983). Desiccation began before 10,000 yr B.P. in South Dakota (Watts and Bright, 1968) and gradually extended eastward. Maximum aridity at Kirchner Marsh occurred about 7000 and 5000 yr B.P. During the late Holocene, cooler climate returned at virtually all these sites. Sites in 379 0033-5894192 $5.00 Copyright 0 1992 by the University of Washington. AU rights of reproduction in any form reserved.


380 a



FIG. 1. (a) Map showing location of prairie peninsula (generalized from Transeau, 1935): (b) Isopolls for prairie forbs 500 yr B.P.

western and central Iowa (Kim, 1986; Van Zant, 1979) also show early to middle Holocene aridity, including Clear Lake on the east side of the former Des Moines Glacier Lobe (Figs. 2 and 3). Webb et al. (1983) used pollen data then available to plot the changes in the position of the prairie-forest border throughout the Holocene, and Bartlein et al. (1984) applied transfer functions to generate Holocene paleoclimatic variables for the Midwest. These studies suggested that prairie expanded well into southern Wisconsin from 8000 to 6000 yr B.P. In contrast, fossil pollen sites along the axis of the prairie peninsula in Ohio and Indiana (Ogden, 1966; Williams, 1974) do not contain middle Holocene increases in percentages of such prairie-indicator taxa as Ambrosia (ragweed), Artemisia (wormwood), Chenopodiineae (goosefoot family and related taxa), Poaceae (grass family), and Asteraceae (sunflower family). Even Chatsworth Bog in central Illinois exhibited relatively insignificant increases in prairie elements (King, 1981). These data did not dispel the idea that the prairie peninsula pushed to its eastward extremity in the early Holocene (King, 1981). Studies of sites in Wisconsin began to suggest that the prairie expansion was less

extensive and later than previously thought. Maher (1982) found that pollen of mesic forest elements at Devils Lake (Figs. 4 and 5) lasted throughout the period when conditions were driest at Kirchner Marsh but declined about 5500 yr B.P. Similar declines in pollen percentages of mesic trees occurred at Lake Mendota. Wisconsin (Winkler et al., 1986), along with evidence that lake levels were lowest (and climate driest) between 6000 and 3000 yr B.P. at two bog sites in central Wisconsin (Winkler, 1988), in eastern Wisconsin (Webb, 1987), and at the Lima Bog site (Figs. 4 and 5). At none of these eastern sites is there a significant increase in prairie elements in middle or later Holocene time. This paper expands on interpretations suggested in a short article on deposits at Roberts Creek in northeast Iowa (Chumbley et al., 1990). We illustrate the regional pattern in order to test previous numerical



cl fg

ConiferHardwood 0 Forest Mixed Prairie and cl Forest



Maximum Extent of Wisconsinan Glacier

FIG. 2. Map showing location of sites. boundary of Wisconsinan glacial maximum, and generalized vegetation.


Roberts Pollen







Creek Percentages

FIG. 3. Pollen diagrams from Clear Lake and Roberts Creek sites, Iowa. *, rejected date.

models that use modern pollen-vegetation and pollen-climate patterns to reconstruct Holocene vegetational change. The purposes of the paper are two-fold: First, we

wish to present the pattern of regional vegetational changes during the Holocene on pollen diagrams from four sites along a transect (Fig. 2) from prairie (Clear Lake)






Lima Pollen




FIG. 4. Pollen diagrams from Devils Lake and Lima Bog, Wisconsin. *, rejected date

to prairie-forest border (Roberts Creek) and eastward into oak savanna/oak forest (Devils Lake and Lima Bog). Because pollen-depositional settings are scarce in this area, we compare the pollen records from

three types of sites: floodplain ponds, a bog, and lakes. If all show consistent patterns, it is unlikely that the changes were caused by local site-specific conditions. A second purpose of the paper is to present






FIG. 5. (a) Early Holocene prairie-forest border (Modified from Webb et al., 1983). (b) Revised Holocene prairie-forest border (this paper). (c) Air masses important in controlling the early-middle Holocene prairie-forest border, showing the location of additional pollen sequences examined. Numbered sites are: 1, Indian Creek Nature Center (Baker et al., in press); 2, Mud Creek (Baker er al., 1990); 3, Chatsworth Bog (King, 1981); 4, Volo Bog (King, 1981); 5, Lake Mendota (Winkler et al., 1986); 6, W.ashbum Bog (Winkler, 1988); 7, Disterhaft Bog (Baker, unpublished data; West, l%l); 8, Iola Bog (Schweger, 1969); 9, Radtke Lake (Webb, 1987); 10, Gass Lake (Webb, 1987) and Seidel Lake (West, 1961); and 11, Kirchner Marsh (Wright et al., 1%3). (d) Arrows indicate air masses important in controlling the late Holocene prairie-forest border.

hypotheses to explain the timing of the Holocene dry period. Roberts Creek and Devils Lake have extremely good radiocarbon control. The

Roberts Creek sequence is derived from a series of small ponds along Roberts Creek in northeastern Iowa (Fig. 2). Each site represents only 100-200 yr of accumulation,



and the diagram is a compilation of individually dated sites along the creek (Chumbley, 1989; Chumbley et al., 1990). These pond deposits contain well-preserved pollen and plant macrofossils, and each site has one or more dates for wood, the material acknowledged as most reliable for radiocarbon dating. Devils Lake, Wisconsin (Fig. 1) is located in a paleo-channel of the Wisconsin River cut through Precambrian quartzite. The lack of access to any ancient carbonate sources or pre-Quaternary carbonaceous deposits make the sediments in this softwater lake noncalcareous and favorable for radiocarbon dating; 11 radiocarbon dates provide an excellent chronology (see also Maher, 1982). In contrast, the chronologic control at Clear Lake and Lima Bog is less reliable, especially near the base. Clear Lake is the easternmost large lake in Iowa (Fig. 2). Carbonates and reworked spores in the till at this site have apparently contributed ancient carbon to the lake sediment below a depth of 980 cm (Fig. 3) yielding anomalous dates. Lima Bog is located between drumlins in southeastern Wisconsin. The lowest three dates at this site are all older than the drift surrounding the site; the bog lies in a belt of lower Paleozoic carbonate rocks and the lower sediments contain numerous pre-Quaternary microfossils. The upper four dates are reasonable and are plotted on the diagram. Generalized pollen sequences from Devils Lake (Maher, 1982) and Roberts Creek (Chumbley, 1989; Chumbley et al., 1990) are available, whereas the Clear Lake and Lima Bog work is unpublished. Although the late-glacial and Holocene are briefly discussed in this paper, the main focus is on the period from about 10,000 to about 3500 yr B.P. at each site. METHODS

Pollen was prepared by usual methods (Faegri et ul., 1989), except that the Roberts Creek samples were also screened, floated in ZnCl, (specific gravity 1.7), and


treated with dilute bleach (Chumbley, 1989). A minimum of 300 grains were counted for each level. The diagrams have been simplified by deleting local taxa and others that do not change significantly during the Holocene. Pollen sums are based on 17 taxa (16 at Lima Bog). The diagrams were zoned by inspection for convenience of comparison. RESULTS

Clear Lake (Figs. 2 and 3) was located in prairie prior to settlement, although the lake probably had a border of trees in places. The late-glacial interval (zone 1) is characterized by peaks in Picea (spruce), Larix (larch), Fraxinus nigra (black ash), and Cyperaceae (sedge). Sometime shortly before 10,000 yr B.P. the late-glacial pollen types were replaced by pollen of deciduous-forest taxa including Quercus (oak), Ulmus (elm), Ostrya-Carpinus (hophornbean or hornbeam), Bet& (birch), and Acer saccharum (sugar maple), along with Abies (fir). This pollen assemblage comprises zone 2 and was dominant until about 8000 yr B.P. In zone 3 (about 8000 to 3800 yr B.P.) pollen of inferred prairie elements prevailed, including Ambrosia (ragweed), Artemisia (wormwood), Chenopodiineae (goosefoot group), Poaceae (grass family), and Cyperaceae (sedge family). Except for Ambrosia, these elements remained abundant pollen producers in zone 4, when pollen of Quercus returned to prominence. Zone 5 marks the period following settlement and is dominated by Ambrosia pollen. The Roberts Creek area is located along a diffuse part of the prairie-forest border (Fig. 2) and hence is most likely to record prairie invasion during any dry interval. The presettlement vegetation of the Roberts Creek basin was a mosaic of oak openings (mainly Quercus macrocarpa) and prairie (Chumbley, 1988). The late-glacial interval is dominated by Picea pollen, with peaks in Lark, Fruxinus nigra. and Cyperaceue as well (Fig. 3). Pollen of Quercus, Ulmus, Ostryu-Carpinus, Tilia (basswood),



and Acer stlccharum begins to increase shortly before 10,000 yr B.P. The latter 4 taxa are considered mesic deciduous forest (hereafter MDF) elements that prevailed along with Quercus from about 9000 to 5500 yr B.P., in c.ontrast to the Clear Lake sequence. Subsequently, arboreal pollen decreased to very low levels in zone 3 (5500 to 3000 yr B.P.), and Ambrosia and later Poaceae pollen became very abundant. The rise of Quercus pollen about 3000 yr B.P. marks the base of zone 4, which extends to the time of settlement. Postsettlement horizons are marked in zone 5 by a sudden increase in Ambrosia pollen, which marks postsettlement cultivation seen throughout most of eastern United States. The Devils Lake sequence (Fig. 4) shows the course of vegetational change in central Wisconsin. The area was mapped as oak savanna with MDF in nearby areas during presettlement time (Wisconsin Geological and Natural History Survey, 1965). Zone 1 at Devils Lake, dating from about 12,500 to 11,000 yr B.P., is dominated by Picea pollen, along with Larix, Fraxinus nip-a, and other typical late-glacial elements. Picea is replaced in zone 2 by Betula (birch) and Pinus (pine) which drop off sharply about 9700 yr B.P. :In the overlying zone 3, dating from 9700 to 5500 yr B.P., high values of pollen from :MDF elements prevail. Total AP is generally between 80 and 95%. Pollen of MDF elements decreases abruptly about 5500 yr B.P. at the base of zone 4, and Quercus pollen percentages rise sharply. Poaceae (grass), Ambrosia (ragweed) pollen, and other NAP (nonarboreal pollen) percentages increase only slightly here, in contrast to the sequence at Clear Lake and Roberts Creek. This dominance of Quercus pollen with only low peaks in NAP is present from about 5500 to about 3400 yr B.P. Zone 5 differs from zone 4 only in the slight increase in Betula pollen, and the slight decrease in Poaceae and Ambrosia pollen, Zone 6 records the postsettlement increase in Ambrosia pollen. The upland forest adjacent to Lima Bog



consists mainly of Quercus, Carya (hickory), Juglans (walnut/butternut), and Ulmus. Zone 1 on the pollen diagram is similar to that at the other sites, with high percentages of Picea, along with lower peaks of Fraxinus nigra and Larix (Fig. 4). Successive peaks of Abies, Betula, and Pinus lead into zone 2, which is dominated by MDF elements. This zone terminates about 5200 yr B.P. as linearly interpolated between radiocarbon dates. Quercus pollen prevails to the top of the Lima Bog sequence. The extraordinary rise in Poaceae pollen in zone 3 is almost certainly caused by the establishment of a local wetland grass, possibly Zizania aquatica (wild rice). For unknown reasons, the postsettlement Ambrosia horizon is absent at Lima Bog. DISCUSSION

All sites have late-glacial records that suggest a Picea-Larix forest, with associated Fraxinus nigra and relatively abundant sedges. This conifer-dominated forest gave way to a rapid succession of trees generally associated with conifer-hardwood forests. These vary somewhat from one site to the next but include Abies, Betula, Abies, and Fraxinus nigra. All sites follow with a pollen assemblage that indicates a mesic deciduous forest with associated Quercus, beginning between 10,000 and 9000 yr B.P. At Clear Lake the MDF elements begin to decrease before about 8500 yr B.P. and are at minimum levels between 8000 and 3000 yr B.P. They are replaced by maxima of virtually all pollen types commonly associated with prairie (Artemisia, Ambrosia, other Asteraceae, Chenopodiineae, Poaceae, and trace amounts of Amorpha canescens and Petalostemum). The return of Quercus about 3000 yr B.P. signals the presence of at least scattered oak trees, but the continued presence of significant NAP suggests that the vegetation was oak savanna. This sequence of changes is similar to those at other sites in central and western Iowa (Kim, 1986; Van Zant, 1979), as well



as those in southern Minnesota (e.g., Wright et al., 1963). At Roberts Creek, MDF elements prevailed from prior to 9000 to 5500 yr B.P. Acer saccharum and Tilia became more abundant toward the end of that interval, but a mesic forest and a humid climate is indicated throughout. However, from 5500 to about 3000, trees were scarce in the Roberts Creek area, and prairie became dominant. Zone 3 at Devils Lake and zone 2 at Lima Bog contain the maximum percentages of MDF elements (and maximum pollen accumulation rates; L. J. Maher, unpublished data; K. L. Van Zant, unpublished data). The vegetation between about 10,000 and 5500 yr B.P. was apparently MDF, and the higher AP (arboreal pollen) percentages suggest that there were fewer openings than at Roberts Creek. This kind of forest is considerably more mesic than the modern Quercus-dominated forest and indicates a more humid climate than at present. In Devils Lake zone 4, from 5500 to 3400 yr B.P., all MDF elements drop to very low levels, and a Quercus forest developed that was probably similar to that at present. The increase in Poaceae and Ambrosia pollen during this period may indicate some moreopen areas in the forest, but some of this pollen is probably also derived from longdistance transport from prairies farther west (Maher, 1982). A similar change from MDF to oak forest is estimated to have occurred at 5200 yr B.P. in the Lima Bog area and lasted until sediment deposition ceased. The climate for this period was warmer and drier than that of zone 2 below. This timing of vegetational and climatic change characterizes nearly all sites in southern Wisconsin and northern Illinois (Fig. 5) (King, 1981; L. J. Maher. unpublished data; Winkler, 1988; Winkler et al., 1986). Zone 5 at Devils Lake, from 3400 yr B.P. to settlement time in the 18OOs, differs from zone 4 only in a slight rise in Betula pollen. This Betula rise becomes more prominent


from Devils Lake northward, where it apparently signals the southward migration of conifer-hardwood forest elements, but it is barely discernible at Lima Bog. The climatic signal suggested between 5500 and about 3500 is the same at the three eastern sites: warmer and drier than previous conditions, despite the difference in depositional environments of the sites. At Roberts Creek along the prairie-forest border, prairie supplanted the forest. At Devils Lake and Lima Bog, the forest remained, but mesic elements were replaced by the more xeric Quercus forest. It is unlikely that these changes were caused by local site-specific factors. No hint of these changes is present at Clear Lake. west of the prairie-forest border. The significance of this interpretation of Holocene vegetational change is its consistent trend and its contrast with previous interpretations. From the northern Great Plains east as far as south-central Minnesota and central Iowa the sequence of vegetational changes resembles those at Clear Lake; prairie elements appeared as early as about 9000 yr B.P. and were at their peak from 7500 to about 4000 yr B.P. During this time of maximum aridity in these areas, moisture levels were at their peak and mesic forest was present along Roberts Creek and at the Wisconsin sites (Fig. 5). Prairie invaded eastern Iowa only after 5500 yr B.P., at the same time that xeric forest replaced mesic forest in southern Wisconsin. Bartlein et al. (1984, Fig. 7) show a short period of wetter climate at 6500 and 6000 yr B.P. in southern Wisconsin and northern Illinois. Thus, in the heart of the prairie peninsula the expansion of prairie was several thousand years later than previously thought, and an apparently sharp vegetational gradient existed in eastern Iowa (Fig. 5). This vegetational boundary between prairie and forest is thought to have coincided with a climatic boundary throughout the Holocene (Fig. 5a), as it does at present (Fig. lb). Three air masses are dominant in



the Midwest at present (Bryson, 1966): warm, dry Pacific air is predominant and largely controls the extent of prairie; maritime tropical air from the Gulf of Mexico provides summer humidity and dominates where deciduous forest is present; and continental arctic air from the north supports conifer-hardwood forest (Figs. 5c and 5d). From the northern Great Plains to central Iowa, Minnesota, and northern Wisconsin dominance by Pacific air caused climatic warming and drying to begin in the early Holocene (Fig. 5~). However, maritime tropical air from the Gulf of Mexico blocked the eastward advance of Pacific air in southern Wisconsin and eastern Iowa from early Holocene until 5500 yr B.P. (Fig. 5c), allowing deciduous forest to prevail. Between 5500 and 3000 yr B.P. the blocking by maritime air relaxed, and dry Pacific air extended eastward more frequently. After about 3000 yr B.P., arctic air flow increased, and some forest elements moved southward throughout the Midwest. An alternate hypothesis is that climate did become drier in eastern Iowa and southern Wisconsin about 7500 yr B.P. as it did to the west and north. Prairie was present on the uplands, but Roberts Creek was able to retain mesic forest along the floodplain until some vegetational threshold was reached, and prairie replaced the floodplain forest. The Wisconsin sites might have also reached some threshold before they changed from mesic to xeric forest. It seems unlikely that threshold effects would occur simultaneously on three different kinds of sites in different vegetation types over such a broad area; thus, the first hypothesis is presently favored. Fires may have played an important role in the transition from forest to prairie at Roberts Creek, and from mesic to dry forest in Wisconsin. Fires would be more likely under a drier climatic regime and would favor both prairie and xeric forest over mesic forest. The climatic interpretation has ramifications for other types of Quaternary studies.



For example, Knox (1985) correlated changes in stream behavior in southwestern Wisconsin with paleoclimatic records deduced from pollen studies available at that time. These studies led him to conclude that “The early Holocene warming and drying corresponds with an apparent trend of decreasing flood magnitudes when, during the time of maximum dryness between 8000 and 7000 yr B.P., magnitudes of 1.58yr floods became 20 to 30% smaller than their modern counterparts,” and that “maximum warmth and dryness occurred in western Wisconsin about 7200 yr B.P., when grass apparently replaced forest vegetation. . . .” However, the present study indicates that this peak dry period did not occur in Wisconsin until 1700 years later. These changes in stream behavior may be due either to other less-dramatic climatic changes, or to nonclimatic events. Another effect of the later warming may have been on wetland evolution. Winkler (1988) suggested that warmer climate (rather than hydrarch succession or basin filling) was responsible for the change in wetlands from open, deep-water ponds to shallow ponds in central Wisconsin after about 6500 yr B.P. Archaeological studies also use paleoclimatic information deduced from pollen studies to derive subsistence patterns of Holocene cultures. Archaeologists have tended to accept King’s (1981) interpretation that the Chatsworth Bog pollen protile indicates prairie expansion into eastern Illinois about 8300 yr B.P. Previous work in southern Wisconsin and eastern Iowa have extended that interpretation to suggest that early Holocene prairie expansion was the rule for the entire Midwest (Stoltman and Barreis, 1983). Recent work from Illinois (Esarey, 1987) does recognize that the “Hypsithermal” began later in northern than in central Illinois. Unfortunately, few stratified upland sites have been found along the axis of the prairie peninsula east of Iowa, and the most extensive work has been done in large valleys as at the Koster



site (Brown and Vierra, 1983) that were protected from aridity by less-variable floodplain water tables. Sites in upland areas of the Prairie Peninsula should be sought to test this new hypothesis. ACKNOWLEDGMENTS Dr. George Malanson kindly read the manuscript. Discussions with Dr. William Green, Mr. A. E. Bettis III, and Mr. David Asch were also beneficial. Parts of this project were supported by National Science Foundation Grant ATM-88-06482 to R. G. Baker; Iowa Science Foundation Grant to R. G. Baker and C. A. Chumbley, and Sigma Xi grant to C. A. Chumbley.

REFERENCES Baker, R. G., Chumbley, C. A., Witinok, P. M.. and Kim, H. K. (1990). Holocene vegetational changes in eastern Iowa. Journal of rhe Iowa Academy of Science

91, 161-177.

Bartlein. P. J., Webb, T.. HI, and Fleri, E. (1984). Holocene climatic change in the northern Midwest: Pollen-derived estimates. Quaternary Research 22, 361-374. Bemabo, J. C., and Webb, T., III. (1977). Changing patterns in the Holocene pollen record from northeastern North America: a mapped summary. Quaternary


8, 64-96.

Borchert, J. R. (1950). The climate of the central North American grassland. Associalion of American Geographers Annals 40, l-29. Brown, J. A., and Vierra, R. K. (1983). What happened in the Middle Archaic? Introduction to an ecological approach to Koster Site Archaeology. In Archaic Hunters and Gatherers in the American Midwest (J. L. Phillips and J. A. Brown, Eds.). pp. 165-188. Academic Press, New York. Bryson, R. A. (1966). Air Masses. streamlines, and the boreal forest. Geographical Bulletin 8, 228-269. Chumbley, C. A. (1988). Late glacial and Holocene vegetation. In “Archaeological and Paleoenvironmental Studies in the Turkey River Valley, Northeast Iowa.” (W. Green, Ed.), pp. 74-109. Research Papers 13, Office of the State Archaeologist, Iowa City, IA. Chumbley, C. A. (1989). “Late-glacial and Holocene Vegetation of the Roberts Creek Basin, Northeast Iowa” Ph.D. dissertation, University of Iowa. Chumbley, C. A., Baker, R. G., and Bettis, E. A., III. (1990). Midwestern Holocene paleoenvironments revealed by floodplain deposits in northeastern Iowa. Science 249, 212-274. Esarey, D. (1987). The Town Branch Sites: Archaic

encampments on the Bushnell Prairie of west central Illinois. The Wisconsin Archeologist 68, 109-124. Faegri. K.. Kaland, P. E., and Krzywinski, K. (1989). “Textbook of Pollen Analysis.” Wiley. New York. Holloway, R. G.. and Bryant, V. M., Jr. (1985). LateQuaternary pollen records and vegetational history of the Great Lakes Region: United States and Canada. In “Pollen Records of Late-Quatemary North American Sediments” (V. M. Bryant, Jr.. and R. G. Holloway. Eds.), pp. 205-245. American Association of Stratigraphic Palynologists Foundation, Dallas, TX. Kim, H. K. (1986). “Late-Glacial and Holocene Environment in Central Iowa: Comparative Study of Pollen from Four Sites.” Unpublished Ph.D. dissertation, University of Iowa. King, J. E. (1981). Late Quatemary vegetation history of Illinois. Ecologica/ Monographs 51, 43-62. Knox, J. C. (1985). Responses of floods to Holocene climatic change in the Upper Mississippi Valley. Quaternary


23, 287-300.

Maher, L. J. (1982). The palynology of Devils Lake, Sauk County, Wisconsin. In “Quaternary History of the Driftless Area,” pp. 119-135. Field Trip Guide Book No. 5. University of Wisconsin Extension, Geological and Natural History Survey, Madison, WI. McAndrews, J. H. (1966). Postglacial history of prairie, savanna, and forest in northwestern Minnesota. Memoirs of the Torrey Botanical Club 22, l-72. Ogden, J. G., III. (1966). Forest History of Ohio. I. Radiocarbon dates and pollen stratigraphy of Silver Lake, Logan County, Ohio. The Ohio Journal of Science

66, 387-400.

Schweger. C. E. (1969). Pollen analysis of lola Bog and paleoecology of the Two Creeks Forest Bed, Wisconsin. Ecology 50, 859-868. Stoltman. J. B.. and Barreis, D. A. (1983). The Evolution of Human Ecosystems in the Eastern United States. In “Late Quatemary Environments of the United States. Vol. 2, The Holocene” (H. E. Wright, Jr., Ed.). pp. 252-268. University of Minnesota Press, Minneapolis, MN. Transeau, E. N. (1935). The Prairie Peninsula. Ecology 16, 423437. Van Zant, K. L. (1979). Late-glacial and postglacial pollen and plant macrofossils from Lake West Okoboji, northwestern Iowa. Quaternary Research 12, 358-380. Watts, W. A., and Bright, R. C. (1968). Pollen, seed, and Mollusk analysis of a sediment core from Pickerel Lake. Northeastern South Dakota. Geological Society

of America


19, 855-876.

Watts, W. A., and Winter, T. C. (1966). Plant macrofossils from Kirchner Marsh, Minnesota-A paleoecological study. Geological Society of America Bulletin 77, 1339-1360. Webb, S. L. (1987). Beech range extension and vege-



tation history: Pollen stratigraphy of two Wisconsin bog sites. Ecology 68, 1993-2005. Webb, T., III, Cushing, E. J., and Wright, H. E., Jr. (1983). Holocene changes in vegetation of the Midwest. In “Late Quaternary Environments of the United States. Vol. 2, The Holocene” (H. E. Wright, Jr., Ed.), pp. 142-165. University of Minnesota Press, Minneapolis, MN. West, R. G. (1961). Late- and postglacial vegetational history in Wisconsin, particularly changes associated with the Valders readvance. American Journal of Science 259, 766433. Williams, A. S. (1974). “Late-Glacial-Postglacial Vegetational History of the Pretty Lake Region, Northeastern Indiana.” U.S. Geological Survey Professional Paper 686-B. Winkler, M. G. (1988). Effect of climate on develop-



ment of two Sphagnum bogs in south-central Wisconsin. Ecology 69, 1032-1043. Winkler, M. G., Swain, A. M., and Kutzbach, J. E. (1986). Middle Holocene dry period in the northern midwestem United States: Lake levels and pollen stratigraphy. Quaternary Research 25, 235-250. Wisconsin Geological and Natural History Survey, (1965), Early Vegetation of Wisconsin. Wright, H. E., Jr. (1968). History of the Prairie Peninsula. In “The Quatemary of Illinois” (R. E. Bergstrom, Ed.), pp. 78-88. University of Illinois, College of Agriculture Special Publication 14. Wright, H. E., Winter, T. C. and Patten, H. L. (1963). Two pollen diagrams from southeastern Minnesota: problems in the regional late-glacial and postglacial vegetational history. Geological Society of America Bulletin 74, 1371-1396.