Quaternary Research56, 401–410 (2001)doi:10.1006/qres.2001.2273, available online at http://www.idealibrary.com on

Late Quaternary Water-Level Variations and Vegetation Historyat Crooked Pond, Southeastern Massachusetts

Bryan Shuman, Jennifer Bravo, Jonathan Kaye,1 Jason A. Lynch,2 Paige Newby, and Thompson Webb III

Department of Geological Sciences, Box 1846, Brown University, Providence, Rhode Island 02912

Received July 14, 1999

Sediment cores collected along a transect in Crooked Pond,southeastern Massachusetts, provide evidence of water-levelchanges between 15,000 cal yr B.P. and present. The extent of fine-grained, detrital, organic accumulation in the basin, inferred fromsediment and pollen stratigraphies, varied over time and indicateslow water levels between 11,200 and 8000 cal yr B.P. and fromca. 5300 to 3200 cal yr B.P. This history is consistent with the pa-leohydrology records from nearby Makepeace Cedar Swamp andother sites from New England and eastern Canada and with tempo-ral patterns of regional changes in effective soil moisture inferredfrom pollen data. The similarities among these records indicatethat (1) regional conditions were drier than today when white pine(Pinus strobus) grew abundantly in southern New England (11,200to 9500 cal yr B.P.); (2) higher moisture levels existed between 8000and 5500 cal yr B.P., possibly caused by increased meridonal cir-culation as the influence of the Laurentide ice sheet waned; and(3) drier conditions possibly contributed to the regional decline inhemlock (Tsuga) abundances at 5300 cal yr B.P. Although sea-levelrise may have been an influence, moist climatic conditions duringthe late Holocene were the primary reason for a dramatic rise inwater-table elevations. C© 2001 University of Washington.

Key Words: lake levels; moisture balance; vegetation change; fos-sil pollen; kettle lakes.

oc

ctho

ni

yires,that

hich

d

t therentdry

ofrP.veg-de-ternhend

from

te-

nklernd isois-e of

er-heked

INTRODUCTION

During the past 21,000 yr, the retreating Laurentide ice shand changing seasonal insolation influenced surficial hydrolin New England by altering atmospheric circulation, surfatemperature, and moisture-balance. Therefore, stratigraphiidence of past lake-level fluctations provides insight intolate Quaternary climates of New England. Winkler and Sanf(1995) illustrated relatively low water levels between ca. 10,0and 5000 cal yr B.P. on Cape Cod, Massachusetts, and R. Wet al. (1993) summarized evidence from additional lake amire stratigraphies to demonstrate drier than present condit

1 Current Address: School of Oceanography, University of Washington, PBox 357940, Seattle, Washington 98195.

2 Current Address: Department of Botany, Duke University, Box 9033Durham, North Carolina 27708.

t-byTheatopand

401

eetgyeev-erd00ebbd

ons

.O.

9,

at∼10,000 cal yr. B.P. Newbyet al. (2000) used sedimentardata from transects of cores in two southern New England mMakepeace and Pequot Cedar Swamps (Fig. 1), to confirmdry conditions occurred during the early Holocene.

The lake-level data indicate moisture-balance changes wfit well with conditions inferred from pollen data (R. Webbet al.,1993; Newbyet al., 2000). Cool, dry conditions in New Englanwere initially inferred from abundant pine (Pinus) pollen in earlyHolocene sediments by Deevey (1939). He further noted thaHolocene sequence of temperate tree types indicated diffephases of warmer Holocene climates, including a warm,mid-Holocene reflected by abundant oak (Quercus) and hickory(Carya). Later, Daviset al. (1980) used changing elevationspine and hemlock (Tsuga) macrofossils to infer warmer and driethan modern conditions between 10,000 and 5000 cal yr B.

We studied the moisture-balance changes that influencedetation history in southern New England by obtaining atailed lake-level record from a closed kettle pond in southeasMassachusetts. We confirm that dry conditions prevailed wwhite pine (Pinus strobus) reached its broadest distribution anhighest abundances in New England (Jacobsonet al., 1987).We also provide data to complement sedimentary evidenceMakepeace Cedar Swamp (Newbyet al., 2000) and from sites inOntario (Yu and Andrews, 1994; Yuet al., 1997) that indicate thalow water levels coincided with the mid-Holocene hemlock dcline (Allisonet al., 1986; Bhiry and Filion, 1996). Like ponds oCape Cod that have been influenced by rising sea level (Winand Sanford, 1995), our site is located in a coastal aquifer apotentially sensitive to both sea level rise and changes in mture balance. We have examined our data for the influencboth factors.

SITE SELECTION AND RECONNAISSANCE

Kettle lakes without surface outlets are effectively wattable “outcrops,” where climate is the primary control on tlevel of the aquifer and the lake. Therefore, we chose CrooPond (41◦53′30′′N, 70◦39′W), a 9-ha oligotrophic closed ketle lake, which fulfilled the optimal basin criteria proposedDigerfeldt (1986) and Street-Perrott and Harrison (1985).pond lies 28 m above sea level, and 62 m above bedrock,the coarse sands of the Ellisville kame moraine (Hansen

0033-5894/01 $35.00Copyright C© 2001 by the University of Washington.

All rights of reproduction in any form reserved.

N

o

loer

aa

c

iiov

n

rsr

c-

tc

ils,

of

ters,ossral,iateesyingions.e.,silese

st, andons.ollenhichD),

the late-Holocene hemlock rise (HR), and the ragweed rise (RR). The core

402 SHUMA

FIG. 1. The upper map shows the present bathymetry of Crooked Pwith the locations of cores and probing transects. The profile of the basin shin Figure 2 is based on probing data from the northern (upper) transect. Themap shows the Plymouth-Carver Aquifer (dark grey) and the sites mentionthe text, which are used to provide a regional pollen stratigraphic framewo

Lapham, 1992). Like Makepeace Cedar Swamp, 15 km towest-northwest. Crooked Pond is an exposure of groundwin the coastal Plymouth-Carver aquifer (Hansen and Laph1992).

At present, groundwater enters the pond from the west andits to the east. Measurements of water-table fluctuations, tafrom wells within the surrounding aquifer, show annual flutuations related to variability in the local climate (Hansen aLapham, 1992). We assume that annual water-level variabcontributes to the processes that allow sediment distributwithin the basin to reflect the long-term average water leCoring and surface sample collection coincided with the timelowest water levels in the annual cycle, and water-level variawas minimal between coring dates.

Crooked Pond consists two basins (Fig. 1): a steep noeastern basin (ca. 5.5 m deep) and a shallower southwebasin (ca. 3 m deep). We studied the latter area because gually sloped basins experience wider lateral movement ofshoreline in association with changes in water surface eleva(Digerfeldt, 1986). Sediment depth was determined using 1-diameter metal probing rods, at 10-m intervals along a 205long transect down the long axis of the oval basin (N60W). Tmaximum sediment thickness (11.10 m) and water depthcur ca. 80 m south from the northwestern shore. To undersmodern sediment deposition patterns in the lake, we colle

surface samples, using an Eckman dredge, and comparedment characteristics with water depth.

ET AL.

nd,ownwerd ink.

theterm,

ex-ken-

ndlityns

el.ofce

th-ternad-thetionm-m-heoc-andted

Pine (Pinus strobusandP. rigida) and oak (Quercus rubraandQ. alba) dominate the forest of the sandy, well-drained sosurrounding the pond. Birch (Betulaspp.), heaths (Ericacedee.g.,Vacciniumspp.), and myrica (Myrica pennsylvanica, andM. gale) are also present. Aquatic vegetation near the edgethe pond is composed ofEriocaulon(water depths<1 m) andNuphar(water depths of 1 to 2 m).Myriophyllumforms a carpetthroughout the bottom of the pond (depths>1.5 m).

LAKE-LEVEL RECONSTRUCTION METHODS

Eight sediment cores were collected using a 5-cm-diamehand-driven Wright piston corer (Wright, 1967). Three coreB, C, and D, were taken along a transect, oriented N57W, acrthe basin in 1994. In 1996, core G was taken from a centdeep area, and cores H, I, J, and K were taken at intermedpoints to sample the early Holocene record in detail. All corcapture the contact between lake sediments and the underlcoarse glacial sands. To estimate the past water-level fluctuat(Digerfeldt, 1986), we used the sediment characteristics (iloss-on-ignition and gross grain-size), biostratigraphy (i.e. fospollen), and the estimated sediment accumulation rates of thcores (Fig. 2 and Table 1).

FIG. 2. A profile of the southwestern basin of Crooked Pond from northweto southeast showing A) the percent organic content of surface-samplesB) core locations, sediment stratigraphies, and pollen stratigraphic correlatiThe dashed lines connect the positions of the second peak in spruce pabundance (S2), the peak in heath pollen abundance (HE), the point at woak pollen first rises above 30% abundance (O), the hemlock decline (H

sedi-stratigraphies are generalized to show three types of sediment: detrital organicsediments, inorganic clay and sands.

O

th

1

5rlo

b

e

s

w,9

da

tho-aled

ast lo-),pastpre-ater

theange

nt in-

e-ore

uceti-New

erpo-ong

tescrosslogy

WATER-LEVEL AND VEGETATI

TABLE 1Core Lengths, Locations, and Water Depths

Water Sediment Portion of CorePositionb depth thickness sediment leng

Core Transecta (m) (m) (m) captured (m)

B N57W 10 1.13 1.19 All 1.19C N57W 30 3.03 2.44 All 2.44H N60W 65 1.70 4.25 2.91 to 4.25 1.34J N60W 75 1.80 4.22 3.31 to 4.22 0.9I N60W 85 2.27 4.16 2.34 to 1.82 1.82G – 115 3.28 4.05 All 4.05K N60W 125 3.39 6.70 4.72 to 6.70 1.98D N57W 140 3.70 7.52 All 7.52

a Cores B, C, and D were retrieved from sites along a transect oriented Nwhile cores H, J, I, and K were retrieved from locations along a line fixed acthe lake, oriented N60W, slightly askew to the first transect. Core G fell aneither transect.

b The distances given are measured from the southeastern shore of theThe distances for cores H, J, I, and K are known precisely due to the fixedused during the capture of these cores. The distances for the other cores arapproximate.

The sediments of each core and the additional surfaceples were described. Loss-on-ignition (LOI) at 550◦C estimatesthe percent organic content (Dean, 1974). Pollen samplesobtained at ca. 10-cm intervals from each core, processedcounted using standard techniques (Faegri and Iverson, 1Between 300 and 500 pollen grains were counted per samPollen stratigraphic correlations, guided by square-chordsimilarity measures (Overpecket al., 1985), helped ensure th

er

dates

e Dting

similar-age sediments were compared between cores when mak-ing water-level estimates.

TABLE 2Radiocarbon and Pollen Stratigraphic Dates for Crooked Pond

Calibrated ageRadiocarbon agea Calibrated age range (1 sigma)

Depth (m) (14C yr B.P.) (cal yr B.P.) (cal yr B.P.) Type of date and material Lab numb

Core D0.22 200 150–250 Pollen Stratigraphic1.00–1.05 2990± 90 3190 3000–3340 Conventional, Bulk Sediment Beta-942371.59 4700 ± 200 5300 5100–5500 Pollen Stratigraphic3.50–3.55 6860± 90 7650 7570–7720 Conventional, Bulk Sediment Beta-942386.24–6.32 9220± 70 10,230 10,050–10,300 Conventional, Bulk Sediment Beta-942396.42 10,700 10,200–11,200 Pollen Stratigraphic6.82 12,600 12,400–12,800 Pollen Stratigraphic6.92 13,200 12,900–13,500 Pollen Stratigraphic

Core K6.00 9990± 60 11,140 11,010–11,330 AMS, Bulk Sediment NSRL-108016.30 11,120± 65 13,040 12,970–13,100 AMS, Bulk Sediment NSRL-108026.53 11,890± 65 13,870 13,790–13,940 AMS, Bulk Sediment NSRL-10803

Core I3.53 8150± 70 8990 8770–9000 AMS, Bulk Sediment Beta-105371

a Regional pollen stratigraphic dates are shown in italics and radiocarbon dates on sediments are shown in bold. Calibrated ages for radiocarbonwere

by obtaining three bulk sediment radiocarbon dates from cor(Table 2). The radiocarbon ages were calibrated by interpola

obtained using the IntCal93 dataset (Stuiver and Riemer 1993).Note.AMS = accelerator mass spectrometry.

N HISTORY, MASSACHUSETTS 403

7W,ossng

asin.linemore

am-

ereand89).ple.is-t

A comparison of sediment types revealed changes in lifacies among cores and pollen stratigraphic correlations revehiatuses in shoreward records. These data constrained the pcation of the “sediment limit” (Digerfeldt, 1986; Dearing, 1997which parallels the water-surface level. We assumed thatchange in water level was represented accurately, if notcisely, by applying the modern relationship between the wsurface and the sediment limit. We focused on reconstructingpast water-surface elevations because water depths may chindependent of climate, due to processes such as sedimefilling.

CHRONOLOGY

For an initial chronology, we used linear interpolation btween five well-recognized pollen stratigraphic horizons in cD and their calibrated ages: the ragweed (Ambrosia) increase(200± 50 cal yr B.P.), the hemlock decline (5300± 200 cal yrB.P.), the increase in oak pollen above 10% (10,700± 500 calyr B.P.), the second (Younger Dryas-age) spruce (Picea) peak(12,600± 200 cal yr B.P.), and the low between peaks in sprpercentages (13,200± 300 cal yr B.P.). Each of these stragraphic events is observed at sites throughout southeasternEngland and the age of each event, as estimated by linear intlation between calibrated radiocarbon dates, is consistent amsites (Newbyet al., 2000). We estimated the error of these dabased on the standard deviation of the spread of dates athe region. We then filled in and tested the proposed chrono

N

nbaat

imttloas

e

ur

io

.

lna

tne

italsur-G,edcmtheof

hethe

s H,ativereynededi-nd

hick,andtopre-an. 2).rlym ofud,orepted

valsropentde-

nto

tesre G

hicternresialigra-, andces.lee

404 SHUMA

FIG. 3. Age model for core D, showing the dating error and depth rafor each date. Solid circles represent calibrated radiocarbon dates, openindicate the relative position of calibrated radiocarbon dates from core K,gray points represent the calibrated ages of several regional pollen stratigrevents, based on the ages calculated for other nearby sites. Table 2 lisspecific date information.

between dates given in the IntCal93 dataset (Stuiver and Re1993). The calibrated ages fall on or extremely close tolinearly interpolated lines between the pollen stratigraphic dathereby confirming the robustness of the estimated chrono(Fig. 3). Four additional AMS radiocarbon-dated sediment sples (Table 2) from two other cores also fit well with ages emated by regional pollen stratigraphic comparisons: two braeting the second (Younger Dryas-age) spruce peak in corone near the base of core K, and one from above a sand-laycore I. With the age-model for core D established (Fig. 3) atested, regional and local pollen stratigraphic events wereto correlate between cores to provide age assignments foobserved litho-stratigraphic horizons.

SEDIMENTARY AND STRATIGRAPHIC EVIDENCE

Surface Sediment Samples

Surface samples taken across the pond reveal that theper limit of organic sediment accumulation within the basthe “sediment limit” (Digerfeldt, 1986), lies at water depthsapproximately 1 m. Near-shore (water depths<1 m) modernsediments are mostly sands containing<15% organic matterAt greater depths, the sands grade into dark gelatinous mand the organic content of the substrate rises sharply to>40%(Fig. 2). Ten surface samples taken at depths of 60 cm orhad a mean organic content of 5% with a standard deviatio4%. Two samples from depths of 85 and 94 cm yielded 28%27% organic contents and eight samples taken from depth>120 cm contained 48± 2% organic matter. Because substrabelow the sediment limit are homogeneous with consistehigh organic content (Fig. 2), downcore variability likely r

flects environmental change rather than heterogeneity intrinto the detrital lacustrine sediments.

ove,

ET AL.

geoxesndphics the

er,hees,gy

m-ti-ck-

K,er inndsedthe

up-n,f

uds

essofnd

s ofestly-sic

Sediment Stratigraphy

The eight cores contain predominantly dark brown detrorganic mud overlying clean coarse sands, typical of theficial glacial deposits about the pond (Fig. 2). In cores D,and K, taken from the deep portion of the basin, fine-grainsediment is continuous to the top. In core D, the upper 732of fine-grained organic sediments overlie glacial sands, withlowermost 22 cm of the core containing alternating layersgray inorganic clay and fine sands.

Cores H, I, J, and C, taken from the shallow portion of tbasin, contain sand layers not found in the three cores fromdeeper parts of the lake (Fig. 2). The stratigraphies of coreJ, and I are similar to each other, and core J is representof the three. It contains a basal unit, 24 cm thick, of dark ginorganic silt and sand, which grades into 13.5 cm of fine-graidark organic-rich mud. A sharp contact separates these sments from an overlying sand layer, 20.5 cm thick. This salayer contains an upward coarsening sequence, 16.5 cm tthat grades from very fine to very coarse sand. Above the slayer, dark brown detrital organic mud is continuous to theof the core. Cores I and H contain similar sequences with pdominately dark-brown detrital organic mud interrupted byupward-coarsening sand layer, 6 cm thick in both cores (Fig

Further shoreward, in core C, a 69.5-cm-thick layer of poosorted sands, ranging from very fine to coarse, overlies 7.5 clacustrine mud at the base of the core. Dark brown organic m128 cm thick, overlies the sand. In contrast, core B, the cnearest the modern shoreline, contains 105 cm of uninterrubrown detrital organic mud and contains no sand intervals.

Loss on Ignition

Low percent organic content corresponds with sandy interin the cores. In the two long, deep cores, D and G, a sharp dto 10–20% approximately 100 cm below the modern sedimsurface (Fig. 4) coincides with a well-recognized and datedcrease in hemlock pollen abundances dated to 5300± 500 cal yrB.P. (Allisonet al., 1986). A radiocarbon date in core D, takefrom just above the interval of low organic content, calibrates3000–3300 cal yr B.P. Pollen stratigraphic correlation indicathat the rise in percent of organic matter is synchronous in co(Fig. 4).

Pollen Stratigraphies

Core D is the only record that contains all pollen stratigrapzones, which represent the vegetation history of southeasNew England over the past ca. 15,000 cal yr B.P. (Fig. 5). CoK, I, J, and H extend to late-glacial times, although late-glacsedimentation rates were reduced and early Holocene stratphies are truncated, or at least compressed, in cores I, JH. Cores G, C, and B contain later Holocene pollen sequen

In cores D, K, I, J, and H, pine pollen (with no recognizabPinus subgenusstrobus-type pollen) is abundant, and sprucpollen percentages decline across pollen zone, 1a. Ab

in zone 1b (ca. 13,000–11,100 cal yr B.P.), spruce pollen

O

a

lo

l

e

J,

ared 3aK,s

nt inlock

hed atd Hdate

tain

n the

ores of

er-per-llene in

tcores

WATER-LEVEL AND VEGETATI

FIG. 4. Percent organic matter and percent hemlock pollen in cores DG. Both cores show a substantial reduction of organic content and hempollen percentages at ca. 1 to 1.5 m depth.

percentages rise to a second peak (>50%), typical of late-glaciarecords from southern New England (Deevey, 1939; Leop1956; Davis, 1969; Gaudreau and Webb, 1985). Peteetet al.(1990) have shown this spruce pattern to be related to coomates during Younger Dryas time. Peaks in alder (Alnus) and fir(Abies) pollen also occur in zone 1b.

Zone 2a is absent or reduced in thickness in cores I, J,H (Fig. 6). Peaks in the percentages of birch, myrica, and h

pollen distinguish the lowermost portion of zone 2a in cores Dat Crooked Pond (Fig. 3), and the pollen stratigraphic context

6;

and K. Zone 2a is also characterized by the highest percent-fits well with other regional records (Winkler and Sanford, 198

FIG. 5. Percentage pollen diagram from core D. A sum o

N HISTORY, MASSACHUSETTS 405

ndlock

ld,

cli-

andath

ages of pine pollen (48–72%), with white pine (Pinussubgenusstrobus-type) pollen being particularly frequent. In cores I,and H, sand layers clearly truncate this zone (Fig. 6).

The fine organic sediments found above the sand layersearly mid-Holocene age and correspond to zones 2b an(Fig. 6). Zone 2b, which is well recorded only in cores D,and I, has high percentages of pine (>30%, with low percentageof white pine-type pollen) and oak (>30%) pollen with lesseramounts of birch and hemlock pollen. Zone 3a, also presecores J and H, is marked by increased percentages of hemand beech (Fagus). The onset of organic deposition above tsand layer in core I occurred during zone 2b and is date8990 cal yr B.P. (Table 2). However, deposition in cores J andid not resume until zone 3a, 1000 years later, based on aof 7700 cal yr B.P. in core D (Table 2).

The mid- and late-Holocene pollen zones 3 and 4 conhigh percentages of pine and oak pollen (>30% and>20%, re-spectively) and relatively high values (>3%) of birch pollen.These zones are, further, distinguished by small changes ipercentages of hemlock, beech, hickory (Carya), and ragweedpollen (Figs. 5 and 6). In core C, the basal samples contain mthan 2% beech pollen and 6% hemlock pollen, characteristiczone 3a.

At the onset of zone 3b, a decline in hemlock pollen pcentages corresponds to a rise in hickory and birch pollencentages. We recognize this small (ca. 4%) change in popercentages as indicative of the pathogen-triggered declinhemlock populations (Alisonet al., 1986). We feel confidenin this conclusion because the pattern is consistent between

f all trees, shrubs, and herbs was used to calculate the percentages.

ies, with

406 SHUMAN ET AL.

FIG. 6. Selected pollen types from cores D, K, I, C, and B, with lines of correlation (as in Fig. 2). Bars to the left represent the sediment stratigraphdark gray for detrital organic mud and light gray for sand. Circles mark the position of calibrated radiocarbon dates (cal yr B.P.) The inset shows the position of

sand layers (shaded) in cores I, J, and H relative to the total pine (thin line) and white pine (Pinussubgenusstrobus-type; heavy line) stratigraphies. Light gray

i t

gd

l

w

aae

a-em

82;of

our3).

ow

re-nt).fa-evenon-

kel yrandn.

curves in the upper cores K and I stratigraphies also represent percent whpresentation is the same as for core D.

Davis, 1969). A relatively long interval of low hemlock pollepercentage follows (Figs. 5 and 6), with percentages rising aat approximately 3200 cal yr B.P., based on a calibrated racarbon date from core D.

The base of the sand layer in core C corresponds to the hlock decline of zone 3b. Samples from above the sand lacontain relatively high abundances of hemlock and beech poSquare-chord dissimilarity coefficients (<0.15) indicate thatthese core C pollen samples correlate to pollen spectra bet90 and 60 cm in core D. Therefore, organic deposition was likrenewed ca. 3000–2000 cal yr B.P. Similarly, the deepest sples from core B show an increase in hemlock pollen percentfrom<1% to>4%, suggesting that this core also began tocumulate organic sediments around 3000 cal yr B.P. A risthe abundance of ragweed pollen, dating to ca. 200± 50 cal yrB.P., marks the start of zone 4 at 10 cm in core B, 15 cm in cC, 14 cm in core G, and 22 cm in core D (Fig. 6).

INTERPRETING THE STRATIGRAPHIC DATA

Assumptions

The distribution of the modern sediment facies (Fig. 2) isaccordance with the model that wind-driven surface curre

te pine. Pollen taxa are abbreviated in diagrams for cores K, I, C, and B, buthe order of

nainio-

em-yerlen.

eenelyam-gesc-in

ore

and seasonal variance in water level winnow the fine-grain mterials from shallow near-shore sediments and redeposit thin the deeper portions of the basin (Davis and Ford, 19Dearing, 1997). Therefore, we used the maximum elevationfine-grained organic sediments in synchronous intervals ofcores as a proxy for the position of the sediment limit (TableThen, we assumed that (1) no more than 1 m of water can over-lie sandy deposits, (2) water levels were not significantly belsandy deposits, and (3) no less than 1 m of water can over-lie highly organic deposits (otherwise erosion and temperatudependent decomposition of the sediments would be evideIn general, we assume that the changing distribution of lithocies accurately represents the relative water-level changes,if our attempt to quantify the changes relies on modern relatiships that may have differed from those of the past.

Inferred Water Levels

The lowermost sediments in the basin indicate that the laformed during a period of abundant spruce at ca. 14,900 caB.P. The absence of late-Pleistocene sediments in cores CB indicates that water levels were initially lower than moder

inntsCores I, J, and H were near to shore, and initially experiencedsignificantly slower sedimentation rates than cores D and K. The

ver

sed

WATER-LEVEL AND VEGETATION HISTORY, MASSACHUSETTS 407

TABLE 3Inferred Water Levels and Water Depths

Water surface Past sediment surface Past water depth oAge range elevationsa,b elevation in core Dc core D sedimentsd

Cal yr B.P. (m) Rationaleb (m) (m) (m)

14,900–13,800 4.09 m max Maximum of 1 m of water over sand deposition at 5.09 m in C 10.7–10.9 m 6.81–4.186.52 m min >1 m of water over organic deposition at 7.52 m in H

12,800–12,400 4.09 m max Maximum of 1 m of water over sand deposition at 5.09 m in C 10.4–10.6 m 6.51–5.864.54 m min >1 m of water over organic deposition at 5.54 m in H

11,700–10,700 4.85 m max Maximum of 1 m of water over sand deposition at 5.85 m in I 10.12–10.32 m 5.47–4.885.24 m min Water level not below sand deposition at 5.24 m in H

9000–8800 4.82 m max Maximum of 1 m of water over sand deposition at 5.82 m in I 8.19–8.89 m 4.07–3.015.18 m min Water level not below sand deposition at 5.18 m in H

8300–7800 4.09 m max Maximum of 1 m of water over sand deposition at 5.09 m in C 7.39–7.89 m 3.8–3.214.18 m min >1 m of water over organic deposition at 5.18 m in H

6600–5500 1.18 m max Maximum of 1 m of water over sand deposition at 2.18 m in B 5.39–6.29 m 5.11–1.334.06 m min >1 m of water over organic deposition at 5.06 m in C

5500–5000 4.02 m max Maximum of 1 m of water over sand deposition at 5.02 m in C 5.19–5.39 m 1.38–0.185.02 m min Water level not below sand deposition of 5.02 m in C

3300–3150 3.31 m max Maximum of 1 m of water over sand deposition at 4.31 m in C 4.70–4.75 m 1.44–0.394.31 m min Water level not below sand deposition at 4.31 m in C

3150–3000 1.18 m max Maximum of 1 m of water over sand deposition at 2.18 m in B 4.70–4.75 m 3.57–1.393.31 m min >1 m of water over organic deposition at 4.31 m in C

2700–1800 1.18 m min >1 m of water over organic deposition at 2.18 m in B 4.34–4.59 m 4.59–3.16

a Min refers to the lowest possible water level elevation; max refers to highest possible water level elevation.b Meters below modern water surface.c Past sediment surface is estimated from the sediment interval in core D determined by pollen stratigraphic correlation to be the equivalent of theiments

used to determine the water level.

d The range of water depths has been calculated by subtracting the max water elevation from the min sediment surface elevation and the min water elevation

e.

ant

pa

fa-gerun-

thethe

tentastt thetherlowthe

andstese-yrore.

re H, inwa-

from the max sediment surface elevation.

water surface was likely more than 4 m below the modern laksurface or lower until ca. 11,200 cal yr B.P. (Table 3; Fig. 7)cores I, J, and H, sand layers overlie the late Pleistoceneiments (Fig. 6). These stratigraphies suggest that while wlevels were initially high enough to accumulate detrital orgamud up to the elevation in core H, the water surface fell in

FIG. 7. Water table elevations at Crooked Pond through time. Theillustrates the estimated elevation of the water surface, based on stratigr

and sedimentary evidence summarized in Table 3. Boxes indicate the conson water-level elevations, as well as dating error.

Insed-teric

he

lotphic

early Holocene. The lower water level caused a shift in lithocies, dropping the sediment limit to below core I. This chanreduced accumulation in the shallow cores, as indicated by tcated pollen stratigraphies (Fig. 6).

Although factors such as mean wind velocity can affectdepth of the sediment limit, upward coarsening sequences insand layers are typical of a prograding shoreline and consiswith a drop in water level. The pollen data and the contrbetween radiocarbon ages in cores D and I also indicate thasand layers accumulated over 1200 to 2000 yr (Fig. 6), rathan during an instantaneous event. Therefore, we infer thatwater levels lasted from ca. 11,200 to 8000 cal yr B.P., withmaximum lowering from ca. 10,200 to 9000 cal yr B.P.

The time-transgressive nature of the contact between the sand the overlying organic muds in cores I, J, and H indicathat water levels rose slowly at first. The sediment limit rmained>5.8 m below the modern water level ca. 9000 calB.P., when fine-grained sediment began to accumulate in cI. The sediment limit did not rise significantly until after ca8000 cal yr B.P., when fine-grained deposition resumed at co(at 5.2 m below the present water level). Higher in the basincore C, the oldest organic sediments (5.1 m below modern

traintter level) are much younger, dating only to about 5800 cal yrB.P. After 8000 cal yr B.P., the sediment limit rose to a position

N

0ti

t

balnoli

rsaeht

dn2

.

fe

fo

l

a

io

a

P

-velse,

ord

ta-er-is-rnat

di-0 to

n of0

ndaen

hiteold,

lsohiftasganntil

bil-hicioner-

atthetil

in-me

risegheace

the

408 SHUMA

between cores H and C and continued to rise until after 580yr B.P., when deposition began in core C. Rising water, shifthe littoral vegetation away from the center of the pond, maccount for the low abundance of aquatic pollen in core D attime (Fig. 5).

The basal lacustrine sediments in core C are overlainsand layer, which corresponds to the decline in the orgcontent of sediment in cores D and G. The hemlock dec(5300 cal yr B.P.) also correlates with these sedimentary cha(Figs. 3 and 6). Using the elevation of the sand layer in cC as a constraint on the past elevation of the sediment(5.0–4.3 m below the modern water surface), we infer that5300 cal yr B.P., the water surface lay>4 m lower than to-day. Thus, the water was shallow (∼0.7 m) even in the centeof the basin (Table 3), causing the organic content of theiments to drop in cores D and G. As in modern surface sples, shallow (warmer and more turbulent) conditions causreduction in organic preservation and accumulation. This slowing resulted from either sediment infilling or a drop in walevel.

Deposition of organic-rich mud resumed in cores D, G, anby 3300–2800 cal yr B.P. After ca. 2800 cal yr B.P., fine-grailacustrine sediments were first deposited in core B (at 2.below the modern water level). For this reason, we infer a larise in the water surface elevation during the late Holocene

ROLE OF SEA LEVEL

Variations in sea level may be expected to play a role in forcchanges in the water level of a coastal aquifer. However, betw14,000 and 3,000 cal yr B.P., sea level rose at least 45 m ofMassachusetts coast (Oldale and O’Hara, 1980), but water lat Crooked Pond fluctuated only a few meters or less abolong-term mean level (ca. 4.5 m below the modern water levPeriods of low water levels at Crooked Pond near 9000 and 5cal yr B.P. demonstrate the influence of climate rather than olevel, because the water level changed in the direction oppto the sea-level rise.

During the past 3000 cal yrs B.P., the increase in water leveCrooked Pond is consistent with regional evidence for increamoisture levels (Newbyet al., 2000; R. Webbet al., 1993).However, this trend is also similar in magnitude and timinga regional sea-level rise. Crooked Pond water levels increapproximately 0.5 mm/yr after 2300 cal yr B.P., which is slowthan estimated rates of regional sea-level rise based on radbon dates from basal salt marsh peats (Donnelly, 1998), incing data from nearby Neponset River and Barnstable Salt M(Redfield, 1967). Although sea-level changes may have inenced water-table elevations in the Plymouth-Carver aquover the past 3100 cal yr B.P., the rate differences indicatesea-level influences must have been attenuated at Crooked

For these reasons, we consider climate to be the primary forcmechanism behind the inferred water-level changes.

ET AL.

calngayhis

y anicinegesre

mitca.

ed-m-d aal-

er

Cedm

rge

ingeenthevelsut ael).300seasite

s atsed

tosedercar-

lud-rsh

flu-iferthatond.

REGIONAL CLIMATE HISTORY ANDVEGETATION RESPONSES

Climatic controls on the input of moisture into the PlymouthCarver aquifer over the past 15,000 cal yr caused the lake-lefluctuations at Crooked Pond. As would be expected in this cathe water-level history from Crooked Pond matches the recfrom nearby Makepeace Cedar Swamp (Newbyet al., 2000).These hydrological changes also coincide with shifts in vegetion such as the early Holocene increase in white pine. Furthmore, the paleohydrology record from Crooked Pond is constent with the variations in soil moisture described for southeNew England (R. Webbet al., 1993). These results suggests thvegetation was responding to climatic changes.

Late-Glacial and Early-Holocene Climate Changes(15,000 to 8,000 cal yr B.P.)

Changing abundances in spruce, pine and oak pollen incate an increase in surface temperatures from ca. 11,20ca. 9500 cal yr B.P. (e.g. Daviset al., 1980; T. Webbet al., 1993).The decrease in spruce pollen percentages and the initiatiohigh abundances of white pine pollen at Crooked Pond (11,20±500 cal yr B.P.) coincide with a warming at ca. 11,650± 200 calyr B.P., as inferred from geochemical records from Greenla(Stuiveret al., 1995), fossil chironomids from Atlantic Canad(Walkeret al., 1991; Cwynar and Levesque, 1995), and polldata from across New England (e.g. Peteetet al., 1990). How-ever, the water-level records suggest that the increased wpine populations across New England (Deevey, 1939; Leop1956; Davis, 1969; Peteetet al., 1990) resulted not only fromwarming at the end of the Younger Dryas chronozone, but afrom a synchronous decrease in available moisture. This sto drier conditions is recorded in Crooked Pond (Fig. 7),well as at Makepeace Cedar Swamp, where shallowing beca. 11,000 cal yr B.P. and culminated in a hiatus that lasted uca. 8000 cal yr B.P. (Newbyet al., 2000). A slight increase inheath populations at this time may reflect the greater availaity of suitable habitats caused by the drainage of topograpdepressions. In contrast, the arrival of hemlock into the regca. 10,200 cal yr B.P. attest to the importance of rising tempatures for vegetation change.

Hemlock pollen percentages were initially extremely lowCrooked Pond, suggesting that the arrival of hemlock intolocal landscape (Davis and Webb, 1975) did not occur unca. 8000 cal yr B.P., when the lake-level record indicatescreased moisture availability. Beech populations also becamore frequent at this time. Water levels in the pond began toslowly ca. 9000 cal yr B.P. but remained relatively low, allowinsand deposition to continue at site H until ca. 8,000 cal yr B.P. Tlatter age corresponds to the time when the hiatus at MakepeCedar Swamp ended (Newbyet al., 2000). At Crooked Pond,the water-table elevations rose until 5800 cal yr B.P. Again,

ingpattern of Makepeace Cedar Swamp (Newbyet al., 2000) issimilar, with fine-grain organic sediments, which overlie peats,

O

Ne

a

n

nN0wssio

c

a

a

ed

0t

v

in

s,

n

o

ast-quotsell

bec

-nson

nd.P.,ster

re-a-

hats of500te

sub-Newoc-on-calsus-

lsr andrds

sus-oth

byandac-d byeteetne,on,ce.

ed

WATER-LEVEL AND VEGETATI

being deposited between 8000 and 5300 cal yr B.P. Basethese records, the replacement of white pine populations inEngland by birch, beech, and hemlock likely reflects an incrin effective moisture after 8000 cal yr B.P.

Even as North Atlantic sea-surface temperatures increat the end of the Younger Dryas chronozone, a well-develoglacial anticyclone and a strong belt of westerlies focusedthe southern margin of the ice sheet continued to prevent sigcant meridonal flow over eastern North America. Consequemoist air masses were prevented from reaching southernEngland. Only as the Laurentide ice sheet waned, ca. 800yr B.P., did the glacial anticyclonic circulation diminish, alloing meridonal flow to increase and moist Atlantic air massebecome more frequent in southern New England. Recentulations with the NCAR CCM1 corroborate this interpretat(Bartleinet al., 1998).

Mid-Holocene Shallowing of Crooked Pond(5500 to 3100 cal yr B.P.)

Between 5300 and 3100 cal yr B.P., sandy deposition inC and the low organic content of cores D and G indicate a peof shallow water. The stratigraphy at Makepeace Cedar Sw(Newbyet al., 2000) also supports a change in moisture regimca. 5300 cal yr B.P., when the basin began its most recent phpeat accumulation over consolidated organic mud. Duringdrier phase, Makepeace Cedar Swamp shallowed sufficito allow peat formation whereas Crooked Pond remainedenough for open water to remain in the central part of the baThe pollen record from Owl Pond on Cape Cod (Tzedakis, 19contains a peak in pine pollen percentages between 5503000 cal yr B.P. This peak may also indicate conditionswere drier than modern.

As suggested by Bhiry and Filion (1996) and Yuet al.(1997),drier-than-present conditions at 5300 cal yr B.P. perhaps strethe hemlock populations, thereby increasing their susceptibto pathogens. Low percentages of hemlock pollen indicatepopulations were infrequent or absent on the sandy soilsrounding Crooked Pond. However, populations in more faable habitats across the pollen-source region declined in adance as dry conditions caused water levels to fall. The decllikely to have been more severe than otherwise anticipated uthese conditions due to the role of insect activity or disease.conditions about the time of the hemlock decline are consiswith evidence for lower than modern water levels at ManPond, Maine (Almquist-Jacobsonet al., in press), Albion LakeQuebec (Lavoie and Richard, 2000), Lac du Triangle, Que(Asnong and Richard, 1998), Rice Lake, Ontario (Yuet al.,1994), and Crawford Lake, Ontario (Yuet al., 1997).

Late-Holocene Water-Level Rise (3100 cal yr B.P. to Prese

After 3000 cal yr B.P., water levels rose significantly ab

earlier levels, as shown by the onset of lake sediment deption in core B at Crooked Pond and peat deposition in a sh

N HISTORY, MASSACHUSETTS 409

d onew

ase

sedpedearnifi-tly,ewcal

-toim-n

oreriodmp

e atse ofthisntlyeepsin.92)and

hat

ssedilitythatsur-or-bun-e is

nderDrytentell

bec

t)

veosi-

near-shore core at Makepeace Cedar Swamp (Newbyet al.,2000). During this time, other sites from New England and eern Canada also record higher water levels, including PeCedar Swamp, Connecticut (Thorson and Webb, 1991), ManPond, Maine (Almquist-Jacobsonet al., in review), Albion Lake,Quebec (Lavoie and Richard, 2000), Lac du Triangle, Que(Asnong and Richard, 1998), Rice Lake, Ontario (Yuet al.,1994), and Crawford Lake, Ontario (Yuet al., 1997). The expansions of chestnut eastward (Gaudreau and Webb, 1985; Johand Webb, 1989) and spruce southward (Webbet al., 1983; Daviset al., 1980; Spearet al., 1994; Gaudreau and Webb, 1985) athe recovery of the hemlock populations after 3000 cal yr Bis consistent with a regional climate that had become moiand cooler than earlier.

CONCLUSIONS

The sediment and pollen stratigraphies from eight coresveal a history of fluctuating water levels at Crooked Pond. Wter levels were low during the early Holocene, indicating tdry conditions prevailed during the time of high percentagewhite pine pollen in southern New England (11,200 and 9cal yr B.P.). After 8,000 cal yr B.P., rising water levels indicathat the influence of the Laurentide ice sheet waned andtropical air masses became more frequent over southernEngland. In the mid-Holocene, the shallowest conditionscurred at Crooked Pond, indicating that drier than modern cditions coincided with the regional hemlock decline (5300yr B.P.). These dry conditions may have contributed to theceptibility of hemlock to attacks by pathogens (Yuet al., 1997;Bhiry and Filion, 1996). After 3000 cal yr B.P., water leverose when a change the regional vegetation indicated coolemoister conditions than earlier. Thus, the water-level recofrom southern New England show that, as Deevey (1939)pected, the vegetation history reflects intermixed effects of btemperature and moisture balance.

ACKNOWLEDGMENTS

This project was funded by NSF Grant ATM-9523585 to TEMPO andundergraduate research fellowships from Brown University to J. LynchJ. Kaye. We thank A. Mason, T. French, and T. Graham for helping uscess Crooked Pond. Valuable insight into the regional geology was provideB. Hansen and R. Oldale. We also thank H. Almquist-Jacobson and D. Pfor helpful reviews; J. Donnelly, R. Matthews, W. Prell, P. Richard, J. Stoand J. Williams for useful discussion; and A. Allen, J. Dowling, G. JohnsK. Killian, D. Murray, G. Newby, and R. Summers for additional assistanP. Leduc and G. Johnson provided technical support.

REFERENCES

Allison, T. D., Moeller, R. E., and Davis, M. B. (1986). Pollen in laminat

ort,sediments provides evidence for a mid-Holocene forest pathogen outbreak.Ecology67,1101–1105.

N

,

Cce

r

e

la

le

r

u

e

j

ti

e

i

m

c

of a

ut.

).eace

nerl rise

re-of

r, L.for

rth

y ofre.

re-,

noes.

p in

u-

ke

aceorth-

3).he

t-ess,

of

vel-

rio,

te

410 SHUMA

Almquist-Jacobson, H., Dieffenbacher-Krall, A. C., Brown, R., and Sanger(2001). An 8000-yr Holocene record of lake-levels at Mansell Pond, cenMaine, U.S.A.The Holocene11(2), 189–201.

Ansong, H., and Richard, P. J. H. (1998). Postglacial sedimentation and wlevel changes in Lac du Triangle, central Gaspe peninsula, Quebec.In “Ge-ological Society of America 33rd Annual Meeting, Abstracts with PrograNortheastern Section,” Vol. 30, p. 3.

Bartlein, P. J., Anderson, K., Anderson, P., Edwards, M., Mock,Thompson, R., Webb, R., Webb III, T., and Whitlock, C. (1998) Paleomate simulations for North America over the past 21,000 years: Featurthe simulated climate and comparisons with paleoenvironmental data.Qua-ternary Science Reviews17,549–586.

Bhiry, N., and Filion, L. (1996). Mid-Holocene hemlock decline in eastern NoAmerica linked with phytophagous insect activity.Quaternary Research45,312–320.

Cwynar, L. C., and Levesque, A. J. (1995). Chironomid evidence for late-glaclimatic reversals in Maine.Quaternary Research43,405–413.

Davis, M. B. (1969). Climatic changes in southern Connecticut recordedpollen deposition at Rogers Lake.Ecology50,409–422.

Davis, M. B., and Ford, M. S. (1982). Sediment focusing in Mirror Lake, NHampshire.Limnology and Oceanography27,137–150.

Davis, M. B., Spear, R., and Shane, L. (1980). Holocene climate of New EngQuaternary Research14,240–250.

Davis, R. and Webb III, T. (1975). The contemporary distribution of polin eastern North America: A comparison with the vegetation.QuaternaryResearch5, 395–434.

Dean, Jr., W. E. (1974). Determination of carbonate and organic matter incareous sediments and sedimentary rocks by loss on ignition: Compawith other methods.Journal of Sedimentary Petrology44,242–248.

Dearing, J. A. (1997). Sedimentary indicators of lake-level changes in the htemperate zone: A critical review.Journal of Paleolimnology18,1–14.

Deevey, Jr., E. S. (1939). A postglacial climatic chronology for southNew England.American Journal of Science237,691–721.

Digerfeldt, G. (1986). Studies on past lake-level fluctuations.In “Handbookof Holocene Palaeoecology and Palaeohydrology” (B. E. Berglund, Epp. 127–142. Wiley, Chichester.

Donnelly, J. P. (1998). Evidence of late Holocene post-glacial isostatic adment in coastal wetland deposits of eastern North America.In “Dynamics ofthe Ice Age Earth” (P. Wu, Ed.) GeoResearch Forum. Trans Tech PublicaUetikon–Zuerich.

Faegri, K., and Iverson, J. (1989). “Textbook of Pollen Analysis.” WilNew York.

Gaudreau, D., and Webb III, T. (1985). Late-Quaternary pollen stratigraphyisochrone maps for the northeastern United States.In “Pollen Records ofLate-Quaternary North American Sediments” (J. Bryant and R. G. HallowEds.), pp. 247–280. American Association of Stratigraphic PalynologDallas.

Hansen, B. P., and Lapham, W. W. (1992). “Geohydrology and Silated Ground-Water Flow, Plymouth–Carver Aquifer, Southeastern Msachusetts.” United States Geological Survey Water Resources InvestigReport 90-4204, Marlborough, Massachusetts.

Jacobson, G. L., Jr., Webb III, T., and Grimm, E. C. (1987). Patteand rates of vegetation change during the deglaciation of eastern NAmerica.In “North America and Adjacent Oceans during the Last Deglaction” (W. F. Ruddiman and H. E. Wright, Jr., Eds.), Geology of North AmeriVol. K-3, pp. 277–288. Geol. Soc. Amer., Boulder, Colorado.

Johnson, W. C., and Webb III, T. (1989). The role of blue jays (Cyanocitta

cristata L.) in the postglacial dispersal of fagaceous trees in eastern NoAmerica.Journal of Biogeography16,561–571.

ET AL.

D.tral

ater-

m,

.,li-s of

th

cial

by

w

nd.

n

cal-ison

mid

rn

d.),

ust-

ons,

y,

and

ay,sts,

u-as-ation

rnsorthia-a,

Lavoie, M., and Richard, P. J. H. (2000). Postglacial water-level changessmall lake in southern Quebec, Canada.The Holocene10,621–634.

Leopold, E. B. (1956). Two late-glacial deposits from southern ConnecticProceedings of the National Academy of Sciences52,863–867.

Newby, P. E., Killoran, P., Waldorf, M. R., Webb III, T., and Webb, R. S. (200011,500 years of sediment, vegetation, and water level changes at MakepCedar Swamp, southeastern MA.Quaternary Research53(3), 352–368.

Oldale, R. N., and O’Hara, C. J. (1980). New radiocarbon dates from the incontinental shelf off southeastern Massachusetts and a local sea-levecurve for the past 12,000 yr.Geology8, 102–106.

Overpeck, J. T., Webb III, T., and Prentice, I. C. (1985). Quantitative interptation of fossil pollen spectra: Dissimilarity coefficients and the methodsmodern analogs.Quaternary Research23,87–108.

Peteet, D. M., Vogel, J., Nelson, D., Southon, J., Nickmann, R., and Heusse(1990). Younger Dryas climatic reversal in northeastern USA? AMS agesan old problem.Quaternary Research33,219–230.

Redfield, A. C. (1967). Postglacial change in sea level in the western NoAtlantic Ocean.Science147,50–55.

Spear, R. W., Davis, M. B., and Shane, L. C. K. (1994). Late Quaternary historlow- and mid-elevation vegetation in the White Mountains of New HampshiEcological Monographs64,85–109.

Street-Perrott, F. A., and Harrison, S. P. (1985). Lake levels and climateconstruction.In “Paleoclimate Analysis and Modeling” (A. D. Hecht, Ed.)pp. 163–195. Wiley, New York.

Stuiver, M., Grootes, P. M., and Braziunas, T. F. (1995) The GISP2δ18O climaterecord of the past 16,500 years and the role of the sun, oceans, and volcaQuaternary Research44,341–354.

Stuiver, M., and Reimer, P. J. (1993). Extended14C data base and revised CALIB3.014C calibration program.Radiocarbon35,215–230.

Thorson, R., and Webb, R. (1991). Postglacial history of a cedar swamsoutheastern Connecticut.Journal of Paleolimnology6, 17–35.

Tzedakis, P. C. (1992). Effects of soils on the Holocene history of forest commnities, Cape Cod, Massachusetts, USA.Geographie Physique et Quaternaire46,113–124.

Walker, I. R., Mott, R. J., and Smol, J. P. (1991). Allerød-Younger Dryas latemperatures from midge fossils in Atlantic Canada.Science253,1010–1012.

Webb, R. S., Anderson, K. H., and Webb III, T. (1993). Pollen response-surfestimates of late-Quaternary changes in the moisture balance of the neastern United States.Quaternary Research40,213–227.

Webb III, T., Bartlein, P. J., Harrison, S. P., and Anderson, K. H. (199Vegetation, lake levels, and climate in eastern North America for tpast 18,000 years.In “Global Climates since the Last Glacial Maximum”(H. E. Wright, Jr., J. E. Kutzbach, T. Webb III, W. F. Ruddiman, F. A. StreePerrott, and P. J. Bartlein, Eds.), pp. 415–467. University of Minnesota PrMinneapolis.

Webb III, T., Richard, P. J. H., and Mott, R. J. (1983). A mapped historyHolocene vegetation in southern Quebec.Syllogeus49,273–336.

Winkler, M. G., and Sanford, P. R. (1995). Coastal Massachusetts pond deopment: Edaphic, climatic, and sea level impacts since deglaciation.Journalof Paleolimnology14,311–336.

Wright, Jr., H. E. (1967). A square-rod piston sampler for lake sediments.Journalof Sedimentary Petrology27,957–976.

Yu, Z., and Andrews, J. H. (1994). Holocene water levels at Rice Lake, OntaCanada: Sediment, pollen and plant macrofossil evidence.The Holocene4,141–152.

Yu, Z., Andrews, J. H., and Eicher, Ueli. (1997). Middle Holocene dry clima

rthcaused by change in atmospheric circulation patterns: Evidence from lake

levels and stable isotopes.Geology25,251–254.