Pollen Surface Michigan

8/3/2019 Pollen Surface Michigan

1/18

QUATERNARY RESEARCH 1, 45&467 (1971)

Pollen Grains in Lake Sediments: Pollen Percentages inSurface Sediments from Southern Michigan

111.B. DAVIS,~ L. B. BRUBAKER, AND J. M. BEISWENGERReceived July 28 , 1971

Pol len in surface sediments f rom f ive lakes in southern Michigan show s evidence ofdifferential depo sition. In mo st lakes, pollen from ragweed occu rs in a higher ratioto tree pollen in shallow-wa ter sedim ent than in deep wate r. Pine and certain herbswith smal l pol len grains fol low the same pattern. One very deep lake is an except ion,with the ratio of ragweed to tree pollen highest in deep-w ater sedim ent. P ollen fromlocal aquat ics and from wi l low, which grows along the lake shore, is also unevenlydistributed, occurring in highest frequen cies near the parent plants. Pollen from decidu-ous trees, however, occurs in s imi lar rat ios at al l sampl ing stat ions within each lake.Decidu ous pollen occu rs in uniform ratios, also in older sedim ent, d eposited in theear ly 19th century, w hen the landscape w as st i ll forested.

Percentages of deciduous tree pollen (as percent t ree pol len) were compared amonglakes. Single samples were taken for this purpose from the deepest part of each lakebasin. Oak pol len percentages are higher in three lakes in western Washten aw Coun tythan in three lakes in eastern Wa shtenaw Cou nty. Th is dif ference ref lects a s imi lardifference in present-da y vegetation : second-growth oak forests grow near the lakesin the western half of the county , while all but 5% o f the area in the eastern pa rt o fthe coun ty is farmland . (The difference in the ratio of farmland to forestland in the

, two parts of the coun ty is not reflected clearly in the ratio of herb pollen to tree pollen,because there is so much var iat ion within each lake.) In 140-year-old sediment, onthe other hand, t ree pollen percentages in the s ix samples are homogeneous as shownby a chi-square test . The hom ogeneity in sediment deposi ted be fore the forest wascleared is surpr is ing, because witness-tree data from presett lement t ime show that thefrequencies of tree spec ies in the two areas were quite differen t. Pollen dispersal atthat t ime m ust have been effect ive enough, to counteract di f ferences over distances ofa few ten s of k i lometers in the amou nts and k inds of pol len produced by the vegetat ion.

INTRODUCTION mately to the quant i t ies of parent plants.Fossil pollen grains are useful to paleo-ecology because they record ancient

vegetat ion. The pol len accumulat ing in lakesediment has its source in vegetation sur-rounding the lake; as a consequence thequantities of pollen must be related ulti-

Pollen has been important in paleoecologyprimarily because of its potential for provid-ing quantitative information. Often, how-ever, the quantitative reflection of vegetationprovided by pollen grains is distorted anddifficult to decipher. The distortion can becaused by differential pollen production, dif -

1 Contr ibution number 141 from the Great Lakes ferential transport in air, and differentia l re-Research Div is ion, Univers i ty of Michigan. sistance to decay (Faegri and Iversen,2Great Lakes Research Div is ion, U nivers i ty of 1964). Another factor, seldom discussed in

Michigan, Ann Arbor, M ichigan 48104. th e literature, is differential sedimentation of45 0


2/18

POLLEN PERCENTAGES IN SURFACE SEDIMEXTS 451pollen in lake water. Differential sedimenta-tion forms the subject of this paper.

Differentia l sedimentation means that onetype of pollen will be deposited in prefer-ence to others in certain parts of a lake. Theratio between this particular kind of pollenand others in the total assemblage will bedistorted by this process. The distortion willbe different in different kinds of sediment.The differences will confuse interpretation,when foss il pollen from one sediment faciesis compared with fossil pollen from another.In studies of modern pollen assemblages theeffect will be an increase in variability, ornoise, in samples collected from differentsedimentary situations. The noise may maskimportant differences in pollen assemblagesthat otherwise could be used to distinguishbetween different kinds of vegetation.

Recently attempts have been made tomeasure variability by comparing large suitesof samples collected within lake basins. Databy R. B. Davis ct al. (1969) and otherauthors, primarily Kabaliene (1970)) impli-cate sedimentary processes as the source ofobserved within-lake variations in the pollencontent of sediments. The patterns of distri-bution of pollen frequencies they observedwere related to water depth, implying thatlimnological processes were responsible forthe patterns of deposition. The depositionpatterns we found involve different kinds ofpollen and differ from their results in somerespects, because of differences in the geo-graphical region and method of study. Butwe have come to the same conclusion, thatvariations in pollen frequencies are corre-lated with water depth, and are caused bylimnological factors that bring about differ-ential deposition of pollen grains. They arecaused by processes internal to the lakesand independent of loca l patterns of ter-restrial vegetation.

The sets of samples we studied were col-lected within seven lake basins, most ofthem in Washtenaw County, in southernMichigan. Most samples were from the sedi-

ment surface, collected systematically alongtransects to include all sedimentary situa-tions. At a number of stations pollen wasalso studied in sediment deposited 140 yearsago, just before the forest was cleared fromthe region for farming.

Our results show large differences in pol-len assemblages within each of the lakes.The differences follow systematic patterns,repeatable to a large extent from lake tolake. Further studies, to be reported in laterpapers (M. I3. Davis, in press; Davis andBrubaker, unpublished) have revealed thatthe differences we observed result from pref-erential sedimentation of small pollen grains(and conifer grains with air sacs) in shal-low water, leading to higher ratios with re-spect to the rest of the pollen flora in sedi-ment there. The remaining large pollengrains (without air sacs) display uniforminput to sediment everywhere in the lakebasins. The assemblages of these types arevery nearly homogeneous in all parts ofeach of the lakes studied. Redeposition ofsediment moves the sediment and its en-closed pollen some months after initial sedi-mentation, redistributing the sediment andpollen and determining the final rates of ab-solute accumulation in different parts of thebasin.

The variations in pollen assemblages weobserved in Michigan lakes have importantimplications for studies of pollen in fossilsediment. The changes in pollen throughtime in fossil series can be evaluated only bycomparing them with the differences we findnow from lake to lake. Irregu larly depositedpollen types display high within-lake andbetween-lake variance : they cannot be inter-preted with precision in the fossil record.But uniformly deposited types-the largegrains which happen to include most of thedeciduous tree species in southern Michigan-can be compared usefully from lake tolake. Because there is so little backgroundvariance, they are sensitive to differences ininput from the vegetation, revealing the ex-


3/18

452 DAVIS, BRUBAKER, AND BEISWEXGERtent to which small differences in vegetationon the landscape can be reflected by pollenassemblages in lakes.

DESCRIPTION AND RECENTHISTORY OF STUDY AREAThe study area, Washtenaw County,

Michigan, is in a region of low-relief andgently rolling glaciated topography. Ninetyto ninety-five percent of the landscape isnow under cultiva tion as farmland. Beforethe initiation of farming 140 years ago,however, the area supported deciduous for-ests. A quantitative sample of the speciescomposition of this forest was provided bythe early land surveys (1819-1825), whichrecord witness trees at each quarter-section

corner and at points along section bounda-ries. At that time white and black oak(Quercus alba and Q. velutina) were themost common species, making up almost60% of all forest trees in WashtenawCounty. In the northwestern part of thecounty, where three of the lakes studied arelocated, the forests growing on the sandy up-land soils were composed almost exclusivelyof oak (85% of trees recorded) and hickory(Cnrya, 6% ) (Fig. 1) Oaks were also com-mon (47%) 35 km to the east, near Frainsand Murray Lakes. But on these clay-r ichsoils in the eastern part of the county theforests were more diverse, with elm( UZwzyzus)beech (Fugus), ash (Fraxinus) ,maple (Aces), aspen (P0p2&s j, ironwootl(Ostrya) and basswood (TiZia) each con-

CC

o Oak0 Hickory Washtenaw County,Michigan0. km

FIG. 1. Locations of oak and hickory trees recorded in the course of the land survey in the earlytntury (after Merk, 1951) (compare Fig. 2).

19th


4/18

POLLEN PERCENTAG ES IN SURFACE SEDIMENTS 453tributing 4-9s of the forest (Fig. 2) second-growth forests, occur today as scat-(Perk, 1951). Note that pine (Pinus) is tered woodlots, totaling about 5% of thenot shown on the maps; it is almost com- area in eastern Washtenaw County (Unitedpletely absent from the forests of southern States Geological Survey 1:250,000 map,Michigan. Detroit Sheet).Farmers settling in the region clear-cut Five or ten years after deforestation inthe species-rich forests near Frains and eastern Washtenaw County, the forests inMurray Lakes, starting around 1830 (Anon- the northwest, near Blind, Pickerel andymous, 1881) . The cleared land has been Sayles Lakes (Fig. 2) were cleared forfarmed and pastured continuously ever farmland. The soi ls there are sandy, andsince. Remnants of the original forest, and these farms were less successful than those

Sayies LakeBlind Lake\,

Whitmore LakeA -:erel Lake /

4220,00

0

42IO-0

0 000 . 8340

* Elm0 Beech0 Ash@ Sugar maplea lronwood

FrainsLakeHurrayII Lake

Washtenaw County,Michigan Km

FIG. 2. Elm, beech, ash, sugar m aple and i ronwood trees recorded in the course of the land survey inthe ear ly 19th century (af ter Merk, 1951) (compare Fig. 1) . The locat ions of s ix lakes inc luded in thepresent study are indicated.


5/18

454 DAVIS, BRUBAKER, AND BEISWENGERto the east. Many were abandoned early inthis century. As a result second-growth oakforests are common in the region (about9% of the area of the western half of thecounty, according to USGS 1:250,000 map,Detroit and Grand Rapids Sheets). Forestsnow surround the shores of Blind, Pickerel,and Sayles Lakes.

Our most intensive studies were done atFrains Lake, a small (6.6 ha) lake nearAnn Arbor. The lake has a single, symmetri-cal basin, 9.5 m deep in the center. Waterenters the basin as runoff from the sur-rounding gentle, grass covered slopes. Watermay also enter the lake through a marshyarea connecting with nearby Murray Lake.At times of high water in the springtime,water overflows northward through a shal-low ditch. At other times of year there is nooutflow. A few willow (S&r) trees borderthe north shore of the lake, but around mostof the shore there is only a narrow borderof shrubs with grassy meadows behind them(Fig. 3).

Several lakes were studied for compari-son. The locations of all but one are indi-cated in Fig. 2; their important characteris-tics are summarized in Table 1. Surfacesediments were collected along transectsacross Frains (Fig. 4), Blind (Fig. S),Sayles, Pickerel and Murray Lakes, and inthe deepest part of Sodon and WhitmoreLakes.

METHODSThe samples used for surface sediment

study were cores through the mud-water in-terface, collected in plastic tubes with a pis-ton sampler (Rowley and Dahl, 1956), orwith a modified free-fall valve corer (Phle-ger, 1951). The latter uses a plas tic tubeheld within a weighted brass tube equippedwith a check valve at the upper end. Thesampler is lowered into the water on a ropeand allowed to fall the last 6 ft. into thesediment. Cores were frozen in their plas tic

tubes in a upright position to prevent mix-ing of the sediment. Later they were ex-truded from the tubes and chopped into seg-ments, each of which was analyzedseparately. The uppermost 2 cm of sedimentis considered the surface sample. The sedi-mentation rates in these lakes average 2-3mm per year (RI. B. Davis, Brubaker andBeiswinger, unpublished data) ; surface sam-ples represent accumulation during the last10 years, approximately. At Blind and Pick-erel Lakes the upper 5 cm was used. Herethe sedimentation rate is slower (1-2 mm)and the samples represent sediment depos-ited over several decades. In all cases it canbe assumed that burrowing organisms havemixed the sediment to a greater or lesserdegree with older material (Davis, 1967).

Chemical treatment of samples used forpollen analysis included HF, treatment for 6min. with hot 10% KOH, and acetolys is for1 min. (Faegri and Iversen, 1964). Marlysediment was treated with 10% HCl aswell. P,romoform flotation was used for afew samples containing coarse sand. Theprepared residue was mounted in siliconefluid ; counting continued until at least 150tree pollen grains were tabulated. Counts forall relevant samples are given in Tables 2-6.3

PRESETTLEMENT SEDIMENT,4 distinctive time horizon can be identi-fied in the cores by pollen analysis. Forest

clearance during the 19th century changedthe percentages of pollen produced by thevegetation. Tree pollen declined in fre-quency, and pollen from ragweed (A~nbro-sia) and other herbaceous spec ies increased(Fries, 1962 ; McAndrews, 1%6). The ho-rizon is identified easily in sediment from

s For Tables 2, 3; 4, 5 and 6, order NAP SDocument 01679 from ASIS National AuxiliaryPublications Service, % CCM InformationSciences, Inc., 22 W. 34th Street, New York,New York, 10001, remittin g $2.00 for microficheor $5.00 for photocopies.


6/18

POLLEN PERCENTAG ES IN SURFACE SEDIMENTS

FIG. 3. Frains Lake, showing depth contours ( in meters) and the areas occupied by trees, shrubs andwate r plants near the lake. Square and rectangular areas are buildings, the straight l ines roads. Land con -tours ( in feet) are shown by dotted lines. The elevation of the lake surface is 820 f t (af ter USGS 7%min. topographic sheet, Denton Quadrangle). Bathym etr ic map modif ied from the Michigan Conservat ionDepar tment map.southern Michigan (Fig. 6). Ragweed pol- pollen percentages increase. Pollen from oak,len increases from less than 1% to beech, ash and several additional tree gen-15-4076 ; grass ( Grumineae) , sorrel era, not shown on the diagram, decrease in(Rumex) and pigweed (Ckenopodiaceae) percentages. Pollen from maize (Zea) ,


7/18

456 DAVIS , BRTJBAKER, AND BEISWENGERTABLE 1

MORPHOMETRIC FEATURES OF LAKES STUDIED IN SOUTHERN MICHIGAN

Lakename

countyMaximum Num- Depth to

Area depth ber of thermocli ne Lake Surround ing(ha) (m, basins in summer type vegetation(m l

Sayles Livingston 6 3 1 None Holomictic

Sodon Oakland 2 20 1 7 MeromicticMurray Washtenaw 7 12 1 3-6.5 DimicticFrains Washtenaw 6.6 10 1 4 DimicticPickerel Washtenaw 6.8 15.5 1 5-8 DimicticBlind Washtenaw 25 24 2 6 Dimictic or

monomicticWhitmore Washtenaw 274 21 3 8-13 Dimictic

Forest, somemeadow

Forest,houselots

Forest,meadow

MeadowForestForest

Forest,meadow,houselots,resort

buckwheat (Fugopyruwt) and other cul ti-vated plants and from narrow-leaved plan-tain (Plantago lanceolata), a weed speciesintroduced from Europe, occur at the leve lsin which ragweed pollen i s abundant, andnot below, leaving little doubt that thechanges do in fact represent the clearanceof forest for agr iculture. In the diagramsfrom Blind Lake, in western WashtenawCounty (Fig. 6), there is a reversal verynear the surface, with oak pollen percent-ages increasing again, while weed pollen de-

clines. This trend records farm abandon-ment since the early 1900s in the immediatevicin ity of the lake. Diagrams from FrainsLake, 20 miles to the east in WashtenawCounty, where the landscape has remainedunder cultivation, show high percentages ofragweed continuing to the surface.

Presettlement samples, identified incores by pollen analysis, were used to studypollen deposition in the early 19th century,when the landscape was sti ll forested. InFig. 6, the samples identified as presettle-

FIG. 4. Sampl ing locations in Frains Lake.


8/18

POLLEN PERCENTAG ES IN SURFACE SEDIMEN TS 457

FIG. 5. Bathyme tr ic map of Bl ind Lake, showing sampl ing locat ions.merit are those immediately underlying the we are defining it here, is from plants grow-level indicated by the dashed line. ing so close to the lake that pollen can fall

REGIONAL AND LOCAL vertically onto the water surface. (Note:POLLEN This definition differs slightly from thatIN FRAINS LAKE used by Janssen, 1966.) The input of loca lpollen to the lake water will be irregular,A useful distinction can be made between because the greatest number of grains will

local and regional pollen. Local pollen, as fall onto the lake surface near the parentCORE 5068

i CORE 5071

FIG. 6. Pol len diagrams from two short cores from Bl ind Lake. Percentages of the major pol len typesare plot ted against depth in the sediment. T he dashed line indicates the sett lement hor izon (1830 A. D.).Core 5068 is f rom stat ion H, and core 5071 from stat ion K in Fig. 5.


9/18

458 DAVIS , BRUBAKER, AND BEISWENGERplants. At Frains Lake a few trees thatgrow along the shores of the lake are in thelocal category. The most important is wil-low; large willow trees (S. nigra and S. ba-bylonica) and some willow shrubs growalong the northern and eastern shores (Fig.3). There are also a few maple and poplartrees near the shore, but since these treesare abundant at a greater distance from thelake, it seems like ly that only a small por-portion of the maple and poplar pollen en-tering the lake is local in origin. The mostabundant loca l pollen is from aquatic andemergent vegetation. These plants are irreg-ularly distributed in the shallow waters ofthe lake (Fig. 3)) and the pollen they pro-duce will fall onto the water surface in anirregular pattern.

Regional pollen, on the other hand, is car-ried some distance-several hundred metersor even several kilometers-before itreaches the lake. Consequently we may ex-pect a more or less uniform fallout onto thelake water surface. Oak pollen, for example,makes up between 20% and 40% of all thepollen in the lake sediment. Yet on ly threetrees grow near the lake, a large tree andtwo saplings about 25 m from the westernshore (F ig. 3). These three trees can hardlybe the source of the billions of grains (ca. 3X 1012) (M. B. Davis and Brubaker, unpub-lished) that enter the lake each year. Ac -

cording to Poh ls (1937) estimate of thepollen productivity of European oaks,30,000 IO-year old branching systems are nec-essary to produce this many pollen grainseach year. The majority of the oak pollengrains must, therefore, come from the oakwoodlots and forests that are scatteredthroughout Washtenaw County. The nearestof these is one-half km distant from thelake. There is no reason to suppose that oakpollen from these distant sources would fallonto one part of the lake in preference toanother.

Much of the herb pollen entering the lakeis also regional in origin. The most common

type is ragweed. It grows abundantly as aweed in fields and along roadsides and inother areas of broken ground and does notoccur in the grassy meadows that surroundFrains Lake. Grass and weed pollen canprobably be considered regional in origin,too, while sedge (Cyperaceae) pollen mustbe considered local since some at least maycome from sedges growing in the shallowwater along the lake margin.

RESULTS-FRAINS LAKEThe distributions of pollen percentages in

the lake sediments fall into three generalpatterns.(1) Regional types with little or no vari-

ntion within the lake. This group includesmost tree pollen, except pine, which occursrarely, and the local types like willow. Theratio of oak to total tree pollen (excludingwillow) is almost exactly sim ilar at all 28sampling stations, in both shallow and deepwater (F ig. 7). Presettlement samples areequally invariant. Oak is the only kind oftree pollen that occurs in high frequency-the others, prim arily elm, ash, beech, horn-beam and maple, occur in percentages of 5and 10%. Fig. 8 shows that these percent-age values are also similar in samples col-lected at a varie ty of depths within the lake.The frequencies of deciduous tree pollenwere tested for homogeneity by means of achi-square test. Al l deciduous tree pollenexcept willow at all sampling stations atFrains Lake was included in the test. Theresult (x2 = 269.00, df = 168, p < ,001)demonstrates that the tree pollen fre-quencies are nonhomogeneous. Oak pollenfrequencies appear by subjective judgmentto be greater in deep-water samples, whilemaple and butternut seem more abundant inshallow-water samples. These differenceswere tested for significance. The means andvariances of the frequencies of each decid-uous tree pollen type in the 14 shallow-water samples were compared to those in


10/18

POLLEN PERCENTAGES IN SURFACE SEDIMEN TS 459

1.0OAK t/TOTAL

TREE -5 ;.*

t. 8 .4 %Q -0

POLLEN

0 5 lb

.3-- a

0 .5 IIOWATER DEPTH (M. )FIG. 7. Pol len rat ios in surface sam ples from Frains Lake, plotted against the water d epth at the sam-pling stations.

the 14 deep-water samples by means of t frequencies of deciduous tree pollen in thetests and f tests, respectively ; no signif icant sediment are not completely homogeneous,differences could be demonstrated. One they show no clear pattern of differencemust therefore conclude that although the from water depth in one lake to another.

TREE POLLE

PERCENTAQE SUM INCLUDES TREE POLLEN ONLYFIG. 8. Pollen percentag es for deciduous tree pollen (as percen t total deciduous tree pollen) at FrainsLake, plot ted against the water depth at the sampl ing stat ion.


11/18

460 DAVIS , BRUBAKER, AND BEISWENGERThe pollen counts themselves are given inTables 2-6, to which the reader is re-ferred for detailed information.We find that chi-square is an unsatisfac-tory measure to use for expressing variabil-ity of pollen assemblages. We have used itfor lack of a better statistic, but feel calledupon to warn the reader of the misleadingconclusions that can be drawn from chi-square calculations. One of the drawbacks ofchi-square homogeneity tests is that thenumber of degrees of freedom wi ll dependstrong ly on the number of grains counted.Each data cel l must contain 5 counts; col-lapsing of categories is necessary wherethere are fewer. The result is that authorscounting 1000 or more grains can use a x2test with many degrees of freedom, demon-strating significant differences between sam-ples. Another author, counting only 250grains, could use a x2 test on the same ma-terial, in necessity a less sensitive test withfewer degrees of freedom, and show thatthere are no signi ficant differences amongthe samples. This is one reason for theseemingly contradictory statements in the lit-erature about the homogeneity of pollen incontemporaneous sediments. In the presentinstance we have demonstrated inhomogene-ities, but we want to emphasize that the de-ciduous tree pollen frequencies are never-theless close ly similar from one sample tothe next.

Because oak pollen is abundant, andshows no significant differences in its fre-quency relative to total trees from one sta-tion to another (Fig. 7)) we have found itconvenient to use oak pollen as a standardto which other pollen types can be com-pared. In the following discussion relativepollen frequencies are often expressed as theratios of individual types to oak pollen. Itmust be understood, however, that these ra-tios give no information about the absoluteamounts of pollen. We are not assumingthat the oak numbers are constant-onlythat they are a convenient standard for com-

parison. To understand the processes affect-ing pollen deposition, one must know whichtypes are varying independently of others ;this can be determined most easily by meas-uring absolute deposition (M. B. Davis andBrubaker, unpublished). The relative numb-ers deposited are nevertheless of importancebecause they are so frequently used in paleo-ecology.

(2) Regional types that vary within thelake. This group includes pine and all ter-restrial herbs. Pine trees are very rare insouthern Michigan and the pollen frequencyis low- about 2-3s of total pollen. Never-theless there is a tendency for the percent-ages to be higher in shallow water than indeep water (Fig. 7). Th is corresponds toobservations made by others in NorthAmerica and in Europe (R. B. Davis et al.,1969 ; Kabaliene, 1970). The ratio of pine tooak pollen, plotted against water depth at thesampling stations, is shown in Fig. 7. Therewere few pine grains in the samples; count-ing errors are partly responsible for thewide scatter of points.

Terrestrial herb pollen also occurs inhigher percentages in shallow water than indeep water. This effect does not arise fromthe loca l distribution of these herbs becausethey are not abundant near the lake. Rag-weed, for example, which contributes abouttwo-thirds of the herb total, occurs in high-est percentages in shallow water (Fig. 9).Terrestrial herbs exclus ive of ragweed andsedge (which is local) are also shown inFig. 9. The ratio of grass pollen to oak pol-len (Fig. 9) also shows highest va lues inshallow-water samples.(3) Local pollen types that occur inhighest frequency in sediment near the par-ent plants. The paucity of local vegetation inFrains Lake makes this a small group, in-cluding only willow pollen from the shrubsand trees that grow along the shore, andpollen from aquatic and emergent plantsthat grow in the lake itself. Figure 10 showsthe wil low : oak ratio plotted on a map of


12/18

POLLEN PERCE

RAGWEED 2/ OAK

H E R B S/OAK

JTAGES IN SURFACE SEDIMENTS 461

G R A S S/OAK

0 5 IOWATER DEPTH (M.1

FIG. 9. Pol len rat ios in surface samples from Frains Lake, plot ted against the water depth at the sam-pling s tation s. Herbs refers to total terrestrial herbs less ragweed and Cype raceae .Frains Lake. The ratio varied from 0.79(the largest circles) to 0.15 (the smallestcircles). The highest willow : oak ratios arefound very near the willow trees along theshores, or near to concentrations of willowshrubs at the eastern end of the lake (com-pare Fig. 3).

The ratio of aquatic plant pollen to oak ishighest in water less than 2 m deep, whererooted aquatic and emergent plants aregrowing (Fig. 10). The most spectacularexample is water-willow (Decodon), whichgrows prolifica lly in the swamp at theeastern end of the lake. The high frequencyof its pollen in the sediment there (Fig. 10)is quite characteristic, permitting identifica-tion of sediment from that end of the lake

from its pollen content alone. When theother aquatic and emergent plant genera,principa lly Potamogeton and Typhu. butalso including Ny~~phaea and Nuphar, areconsidered singly, they show simila rly local-ized distributions (Table 2).

RESUL,TS-ADDITIONAL L.qKESIN SOUTHERN MICHIGAN

The patterns of variation in pollen fre-quency found at Frains Lake are repeated,with some important differences, in severallakes in southern Michigan. Since the localvegetation was not mapped at these lakes,we have no information about the distribu-tion patterns of locally produced pollen.Among the regional types (as defined for


13/18

462 DAVIS, BRLJBAKER, AND BEISWENGERwil low /oak

totol wotar plods /oak

Decodon / ook

FIG. 10. Pollen rat ios in surface samp les from Frains Lake, shown at the sampl ing stat ions. The area ofthe circles is proportional to the value of the pollen ratios.Frains Lake), however, the deciduous treepollen percentages are nearly uniform at allthe lakes we have studied, just as they wereat Frains Lake. Figure 11 shows the ratioof oak to total tree pollen (excluding wil-low) plotted against water depth, in surfacesamples from five lakes. As at Frains, thereis little or no between-station variation, al-though the value of the ratio is higher atsome lakes than others. The ratio of oakpollen to total trees is highest at Blind,Picke rel and Sayles Lakes, located in thenorthwestern part of the county where oakforests are now more abundant.The regional pollen types that were varia-ble from one sampling station to another atFrains Lake are also irregularly distributedin all the other lakes. Most lakes showhigher frequencies for pine pollen in shallowwater. Several of the lakes studied showpatterns of distribution of ragweed pollenthat are similar to Frains. At Murray Lake

and Picke rel Lake, for example, the ratio ofragweed to oak pollen is generally higher inshallow water, and lower in deep-water sed-iment (Fig. 12). These two lakes are simi-lar in size and morphometry to FrainsLake. But at Sayles Lake, no pattern is ap-parent. Sayles is very shallow, with waterless than 1 m deep over most of its area.The entire lake is similar to the littoral areaof the other lakes we have studied. Pollenratios were more variable in littoral sedi-ment than in deeper water at Frains Lake(Fig. 9) ; it is not surprising that similarvariation characterizes the entire transectacross Sayles. Variab ility of pollen percent-ages in sediment collected in shallow wateris also reported by R. B. Davis et al. (1969))who used the value of chi-square to measurevariation in pollen assemblages from shallow-water and deep-water samples from Wis-consin lakes.The ragweed: oak ratio at Blind Lake


14/18

POLLEN PERCEN TAGES IN SURFACE SEDIMENTS 463

Frains Lake80. surface presettlement40. -w-b? b.

0 5 IO 0 5 1IOMurray Lake - surface

80- .e .4Q.e -* rr..0 5 IO 15

oak Sayles Lake - surface80-as% btree 40-

pollen 0. 5Pickerel Lake - surface80- .40-. I-.- - - 8

O- 5 IO 15Bli nd Lake - surface

80. . . l . 1 a 140 - . . - IO- 5 IO I5 20

/ Bli nd Lake - presettlement80 . l * 8 R 9. 840

0- IO ISwoter depth (m.) 20

FIG. 11. Pollen ratios (oak to total tree pollen, X 100) in f ive lakes, plotted against the water depth atwhich the samples were collected. Rat ios in presett lemen t sediment are shown for comparison from thesame stat ions in two of the lakes. Presett lement sediment was ident if ied from i ts low ragweed and weedpollen conte nt. S traight l ines were fitted to the data by eye.(Fig. 12) is unique among the lakes westudied. In samples from shallow water, theragweed : oak ratio is lower than in the othernearby lakes. It decreases to a minimum insediment from water 5-10 m deep. The rag-weed : oak ratio then rises again, to highestlevels in the deepest portion of the lake. Thepresence of a shoal-area near the center ofthe lake (Fig. 5) has permitted us to testthe idea that these variations in ratio are infact related to water-depth rather than todistance from shore. The shallow-water sam-ples are variable, both near shore and in thelake center. But the low ratio of ragweed:

oak that character izes samples from 5-10m depth near shore, is repeated at thatdepth near the center of the lake. The highragweed: oak ratio was found in both deepbasins of the lake. These occurrences implythat water-depth, not distance from shore, isthe important variable correlated with thisdistribution of pollen percentages.Another interesting datum is the differ-ence in ragweed : oak ratio in deep water innearby lakes. Many palynologists havethought that a sample or core taken fromthe deepest part of a lake basin is a stand-ard sample. Our study shows that the per-


15/18

464 DAVIS , BRUBAKER, AND BEISWENGER

Frains Lake

Murray Lake

Pickerel Lake

Bl ind Lake

I .a ------ic5 IO ~- 15 20 25water depth (ml

FIG. 12. Pollen ratios in surface samples from five lakes, plotted against th e water depths at which thesampleswere collected. The lines have been fitted to the points by eye,centage of ragweed in deep water is differ-ent, even in adjacent lakes, such as FrainsLake and Murray Lake. These lakes are thesame size, but Murray is deeper and shieldedfrom wind by adjacent hills. Oak woodlotsgrow close to the shores of Murray Lake,but the percentage of ragweed pollen indeep-water sediment there is higher than atFrains Lake. Data such as these have con-vinced us that the patterns of distribution ofpollen percentages that we are observing arecontrolled by limnological factors ratherthan by the distribution of nearby terrestrialvegetation.

IMPLICATIONS OF RESULTS TOPALEOECOLOGYA. Local Pollen

Pollen from local plants does not travelfar within the lake. This ca n be an advan-tage to the paleoecologist, since high per-centages in sediment indicate the nearby oc-

currence of the parent plants. Highfrequencies of willow pollen, for example,indicate that the shore is close to the sam-pling site, and that willow shrubs or treesare growing along the shore. This could beimportant information. Tauber (1967) hasshown that willow thickets along the shoreof a lake filter pollen as it is blown in fromthe surrounding forest, removing largergrains and thus changing the percentages ofpollen entering the lake. At Frains Lake thewinds are too strong for shrubs to be effec-tive as filters (Davis, 1968). The pollendiagrams from Frains Lake do, however,show a very recent, postsettlement increasein willow pollen. Th is effect is strongest innear-shore cores, reflecting the developmentof a band of willow shrubs after the forestwas cleared. Much of the pollen also comesfrom large weeping-willow trees (Sal& ba-bylonica) which were of course plantedafter the area was settled (F ig. 3).


16/18

POLLEN PERCENTAGES IN SURFACE S EDIMENTS 465Many pollen diagrams in the literature

show intriguing changes through time in thepercentages of aquatic plant pollen. Theseare often quite different from site to site, aswould be expected for locally distributedpollen types. Interpretations can probably bemade in great detail. If several cores arestudied from within a single basin, for ex-ample, changes in aquatic plant pollen mightbe used to detect changes in the distributionof aquatic and emergent plants, giving infor-mation on water transparency and/or waterlevel.

B. Regional Pollen, Variable TypesRegional pollen types whose percentages

are variable in lake sediments give only thecrudest quantitative information about thevegetation. The percentages of pine pollen,for example, will vary depending upon thewater depth where the sediment sample wascollected. Large variations also characterizeherb pollen percentages, especia lly ragweed.Of course, there is some relationship to veg-

etation: everywhere in the lakes we studied,the percentage of herbs is higher at the sur-face than in presettlement sediment, reflect-ing the increase in open land. Themagnitude of the percentage increase, how-ever, fails to give an accurate picture of themagnitude of increase in open land, sincethere is no single ratio of herb to tree pollenthat can be considered typical for any of thelakes, much less characteristic for the land-scape in general.

In order to demonstrate this observation,we compared pollen in single samples fromseven lakes in southeastern Michigan, in-cluding the five %Tashtenaw County lakes,Sayles Lake in Livingston County, andSodon Lake in Oakland County (Fig. 2)(Table 1). Each surface sample was fromnear the center of the deepest basin of thelake. The deepest area was chosen for sam-pling because pollen assemblages have lowervariance in sediments deposited in deepwater (Davis et al., 1969). The results (aspercent total pollen) are arranged in geo-graphic order, with the easternmost lake at

o*A. Surface $J

as% A P +N A P t4

LS a c8. Surface 2

as% A P L

6S o LC. Presettle- $a

ment as F% A P w

B E0 20 40 60FIG. 13. Pol len percentages in surface and presettlement sediment at seven lakes in southern Michi gan.

The upper group of lakes in each part of the figure is located to the east of the lower group. Codeletters for lakes: So is Sodon Lake, M is Murray Lake, F is Frains Lake, W is Whitmore Lake,P is Pickerel Lake, B is Bli nd Lake, Sa is Sayles Lake. See Fig. 2 and Table 1 for locations of lakes.


17/18

466 DAVIS , BRUBAKER, AND BEISWENGERthe top of the figure and the westernmost atthe bottom (Fig. 13A). The percentages arehighly variable from lake to lake. The varia-tion in the assemblages is caused largely byvariat ions in the quantity of nonarborealpollen. The resulting noise in the datamakes it difficu lt to detect regional trends inthe assemblages. The percentages of treepollen are much less variable, and a clearerpicture emerges when they are consideredalone. The reduction in variab ility can beexpressed quantitatively. We calculated chi-square, as for a homogeneity test, using allof the assemblages, and including the countsfor nonarboreal pollen. When all pollentypes are included the value of x2 is 407.7with 35 df; p < 0.00001. When nonarborealpollen is not included (Fig. 13 B ) , x2 falls to45.7 with 20 df; p = 0.0009. The highervalue of p is a measure of the lesser hetero-geneity of the assemblages when tree pollenalone is considered. The vegetational signi-ficance of the assemblages wi ll be discussedin the next section.C. Regional Pollen, Uniform TypesPollen types that show little within-basinvariation in their frequencies, such as thedeciduous tree pollen in southern Michiganlakes, should be most valuable for character-ization of regional vegetation. Our datashow, for example, that in southern Michi-gan a single sample is enough to character-ize pollen frequencies of regional deciduoustrees in each lake. Although there is somevariation in sediment in shallow water, thetree pollen frequencies are neverthelessroughly similar everywhere in the lake(Fig. 8). Th is result means that even smalldifferences in the pollen percentages pro-duced by the vegetation wi ll be detectable,as there is little background noise to maskthe true differences. This point is illustratedby the comparison of tree pollen percentagesin the seven lakes (Fig. 13 B) ; the per-centages of tree pollen are remarkably uni-form in nearby lakes. The four eastern lakes

(M, F, W and So), located in an area offarmland, originally mixed-oak forest, aresimilar to one another. They contain appre-ciable percentages of pine, ash, beech, mapleand hornbeam pollen. The three westernlakes (P, B, and Sa), located where oakforest has regenerated, have higher oak pol-len percentages, reflecting the greater ab-solute abundance of oak trees in the vic inity.

A chi-square test shows that the deci-duous tree pollen assemblages from theseven lakes are nonhomogeneous when con-sidered as a single group (x2 = 45.7, df= 20, p = 0.0009). In presettlement time,however, the same test indicates that all sixlakes in Washtenaw County were similar intree pollen content. The percentages of treepollen in presettlement samples are shownin Fig. 13C. They produce a value of x2 =24.02 (df = 20, p = 0.24). This resultmeans the samples are homogeneous by thecriterion of a chi-square test (with 20 df).This similarity in pollen assemblages pre-vailed, even though we know from land sur-vey records made at that time (Figs. 1 and2) that the forests at the two locations werequite different. Maple, ash, elm, beech andironwood were much more common in theeastern part of the county than farther west(Merk, 1951). This result implies that pol-len dispersal by wind was so effective (eventhough the landscape was forested) thatminor differences in pollen production bythe forests were not registered in the lakes.Pollen from lakes such as these studied canbe used in this region to record the meanabundance of trees over a very large area.But the pollen seems ineffective in register-ing minor differences in forest compositionover distances measured in tens of kilome-ters.

Important perspective is added to thisconclusion by preliminary results from asimilar study of between-lake variation inthe Upper Peninsula of Michigan. Here wehave found large variations between lakes,related to lake size. The variations involve


18/18

POLLEN PERCENTAGES IN S IJ RFACE SEDIMENTS 46 7tree pollen, including pine, birch, and oakpollen, although the latter two genera showlittle or no variation within or betweenlakes in southern Michigan. The extent towhich limnological factors rather than dif-ferential input are involved is unclear. Butone can say that results from one regionshould not be applied uncritically to an-other, where vegetation, pollen flora, and to-pography are different.The limnological factors responsible formore or less uniform deposition of tree pol-len in southern Michigan, and variab le dep-osition of herb pollen, are the subject oflater papers in this series.

ACKNOWLEDGMENTSThis work has been supported by the Nat ional

Science Foundat ion, Grants GB 2377, GB 5320, andGB 7727.

REFERENCESAXOXYM O~-S. (1881). History of Wash tenaw

County , Michigan. Chas . C. Chapman & Co. ,Chicago.

DA VIS, M . B. (1968). Pollen grains in lake sedi-men ts : redeposit ion caused by seasona l watercirculation. Science 162, 796-799.

DAV IS. M. B. Redeposi t ion of pol len grains in lakesediments. Limnolog~~ and Oceanografihy, in press.

DAVIS, R. B. (1967). Pollen studies of near-sur-face sediments in Maine lakes. In QuaternaryPaleoecology. (E. J. Cushing and H. E. Wright,Jr . , Eds.) , pp. 143-173. Yale Press, New Haven .

DAVIS, R. B. , BREW STER, L. A. , and SUTH ERLAN D,J. (1969). Variation in pollen spectra withinlakes. Pollew et Spores 11 , 557-572.

FAE GR I, K., and IVERS EN, J. (1964). Textbook ofPol len A nalysis, 2nd ed. Munksgaa rd, Copen-hagen.FR IES , M . (1962). Pollen profi les of late Pleisto-cene and Recen t sediments f rom Weber Lake,Minnesota. Ecology 43, 295-308.J ANSSEN, C . R . (1966). Recent pol len spectra f romthe deciduous and coni ferous-deciduous forestsof northeastern Minnesota: A study in pol lendispersal. Ecology 47, 804-825.KABAL EINE, M. (1970). On format ion of pol lenspectra and restorat ion of vegetat ion ( In Rus-s ian, Engl ish Sum mary). Ministry of Geologyof the USS R, Inst . of Geol. (Vi lnius) Trans.,Vol . I I , 147 p.MC AN DR EW S, J. (1966). Postglacial history ofprairie, sa vanna and forest in northw estern Min-nesota. Torrey Botanical Club Memoir 22, 72 pp.MER K, J. W . (1951). Tree Species Distr ibut ionon the Basis of the Original Land Survey ofWash tenaw Cou nty, Michigan. M.S . thesis, Uni-vers i ty of Michigan.

PHLEG ER, F. B. (1951). Ecology of Foramini fera,northwest Gulf of Mexico: Part 1. Foramini feradistribution. Geological Society of America Mewair 46, l-88.

POH L, F. (1937). D ie Pol lenerzeugung der Wind-bliitler. Beih . B at. Cerctralbl. 56 R, 365-470.ROW LEY, J. R., and DAHL , A . 0. (1956). Modif ica-tions in design and use of the Livingstone pistonsampler. Ecology 37, 849-851.TAU BER , H. (1967). Di f ferent ial pol len dispers ionand fi l tration. Zn Quaternary Paleoeco logy.(E. J. Cushing and H. E. Wright, Jr . , Eds.) ,pp. 131-141. Yale Press, New H aven.

Documents

Pollen Surface Michigan