University of Groningen

14C AMS wiggle matching of raised bog deposits and models of peat accumulationKilian, M.R.; Geel, B. van; Plicht, J. van der

Published in:Quaternary Science Reviews

DOI:10.1016/S0277-3791(99)00049-9

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:2000

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Kilian, M. R., Geel, B. V., & Plicht, J. V. D. (2000). 14C AMS wiggle matching of raised bog deposits andmodels of peat accumulation. Quaternary Science Reviews, 19(10), 1011-1033.https://doi.org/10.1016/S0277-3791(99)00049-9

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 05-07-2020

Quaternary Science Reviews 19 (2000) 1011}1033

14C AMS wiggle matching of raised bog depositsand models of peat accumulation

M.R. Kilian!,", B. van Geel!, J. van der Plicht",*!The Netherlands Centre for Geo-ecological Research, Universiteit van Amsterdam, Kruislaan 318, 1098 SM Amsterdam, The Netherlands

"Centre for Isotope Research and The Netherlands Centre for Geo-ecological Research, Rijks Universiteit Groningen, Nijenborgh 4,9747 AG Groningen, The Netherlands

Abstract

High-resolution Accelerator Mass Spectrometer (AMS) 14C dates of selected plant macrofossils from the raised bog Engberts-dijksvenen (Eastern Netherlands) show century-scale wiggles analogous to the radiocarbon calibration curve. We used three relativetime scales, viz., based on depth, mass, and pollen concentration, respectively, to match the peat AMS dates to the calibration curve.This procedure is repeated for one conventionally dated core. For each relative time scale, realistic con"dence intervals are calculated.Depth appears to be the best time scale for certain stratigraphical units of a core. This justi"es using depth for wiggle matchingconventionally dated mire cores published by others, even when 14C errors are larger. Our evidence shows four major sources of14C variation for mire deposits compared to treerings: (1) dating error, due to sample composition. This includes a reservoir e!ectdemonstrated for many bulk peat samples; (2) hiatuses, causing a sudden &leap' of peat 14C age; (3) changing accumulation rates,apparent from a break in the slope of the peat 14C ages; (4) sampling error. These results shed doubt on the assumed continuities intheoretical peat accumulation models. Both mire stratigraphy and changing accumulation rates can be explained in qualitativehydrological terms. More evidence for a previously discovered reservoir e!ect in bog deposits is presented, though the phys-ical/biological mechanisms remain uncertain. ( 2000 Elsevier Science Ltd. All rights reserved.

1. Introduction

Studies of Holocene sediments often rely on 14C datingto provide a time frame of their genesis. Both &absolute'ages and rates of accumulation are relevant for the inter-pretation of geological } a.o. botanical } sequences, andtheir comparison to independently dated records. Agesand accumulation rates are described by a time-depthrelation, which is usually constructed in two successivesteps. First, a series of 14C dates is calibrated (van derPlicht, 1993), i.e. converted to calendar ages individuallywith the 14C calibration curve. Then a function is cal-culated to describe ages throughout the sequence. Theresulting time-depth relations for sediments, includingbogs, are often nonlinear (Bennett, 1995; Clymo, 1984;Middeldorp, 1982). This may result either from real accu-mulation rate changes, or from inaccuracies in the datingand curve "tting procedures.

*Corresponding author.E-mail address: plicht@phys.rug.nl (J.van der Plicht).

Problems in this dating approach are twofold. First,calibrated 14C ages from peat bogs are usually inter-preted to be both accurate and precise (Olsson, 1991).Contamination of samples, either in the laboratory, orby intruding roots or uptake of older carbon (reservoire!ect: Stuiver and Pollach, 1977) are assumed to beabsent. However, there is ample evidence that bulk peatsamples may contain fractions di!ering by centuries ormillennia in age (Olsson, 1986; Kilian et al., 1995; Shoreet al., 1995).

Second, problems arise due to the nonlinearity of the14C treering curve (Fig. 1). Past atmospheric 14C concen-trations #uctuated mainly due to varying solar activity.The resulting century-scale atmospheric 14C &age' vari-ations, wiggles, transform single gauss-distributed BPages into complex calendar age probability distributions.These ages are generally simpli"ed to support a "ttedcurve (Bennett, 1995). But errors in individual datapoints and the time}depth relation as such are di$cult toevaluate.

The main aim of the present study is to developa method for reconstructing accurate time}depth

0277-3791/00/$ - see front matter ( 2000 Elsevier Science Ltd. All rights reserved.PII: S 0 2 7 7 - 3 7 9 1 ( 9 9 ) 0 0 0 4 9 - 9

Fig. 1. The radiocarbon calibration curve (Stuiver & Reimer, 1993). To avoid using error bars, in all "gures treering 14C age $2r is indicated(envelope). The long-term trend is sinusoidal, with century-scale #uctuations (wiggles) superimposed.

relations for Holocene sediments and peat deposits thatis less sensitive to these dating problems. This also in-volved selection of speci"c botanical fractions to assesstheir reliability for 14C dating. A comparison of relativetime scales serves to evaluate, and reject, some assump-tions of theoretical peat accumulation models.

2. Wiggle matching and relative time scales

In our dating approach, wiggle matching, the strati-graphic interrelation between 14C dates is used to pro-vide a more accurate age determination for a wholepro"le. By high-resolution 14C AMS dating of a series ofsediment samples we reconstruct the wiggles prevalent inthe 14C calibration curve, but with a separate relativetime scale (Neustupny, 1973). This time scale can bemanipulated to match the uncalibrated peat 14C agesto the treering curve, as was proposed by van Geel andMook (1989).

This approach inverts the usual order of reconstruct-ing a time}depth relation: before any individual calendarages are known, a relative time scale should already be

established. Therefore wiggle matching of #oating treer-ing series is the most straightforward application (Fer-guson et al., 1966; Stuiver et al., 1986; Pearson, 1986; vander Plicht et al., 1995). As for varved lake sediments(Goslar et al., 1995), absolute sample distances in calen-dar years are known. Nonvarved sediments can bewiggle-matched if plotted versus depth, or assuming con-stant input of a measured variable. For sediments thatare largely inorganic, depth may be the only relative timescale available. For raised bogs, two alternatives areused:

(1) Middeldorp (1982, 1984) assumed that tree pollenin#ux for a raised bog site is constant through time, sothat pollen concentrations would increase proportionallywith the amount of time represented by a sample. Mid-deldorp (1982, 1984) therefore used cumulative pollenin#uxes (&P) as a relative time scale, which is nonlinearcompared to depth.

(2) If we assume a constant net productivity (g/cm2 y)of the bog plant communities, there should be a linearrelation between cumulative bulk density (&m) and time(Clymo, 1978, 1983, 1984, Ilomets, 1980 in: Punning

1012 M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033

et al., 1993). Distortion of the time scale due to compac-tion increases speci"c weight, which therefore does nota!ect the mass-based time scale. However, according to(Clymo, 1978, 1983, 1984) the time}&m relation of bogs isconcave, because long-term decay leads to mass lossesfrom the bog. If this mass loss is assumed to be propor-tional to the amount remaining, the time-&m relationbecomes logarithmic [(Clymo, 1984) (see discussion be-low)].

Though a generally accepted convention for establish-ing time}depth relationship does not exist, it would belogical to assume a linear relationship as the norm. It isthe most simple and the easiest to apply. We compareother relative time scales to this standard, by employingthem to wiggle-match the same 14C data to the treeringcurve. If pollen concentration or mass input were moreconstant in time than depth increment, then the use oftheir cumulative values as a relative time scale shouldimprove the wiggle-match &"t'. This will be discussedbelow.

The wiggle-matching was carried out with a specialversion of the Groningen 14C calibration program Cal20(van der Plicht, 1993). The wiggle match option containsboth the sediment 14C data and the relevant calibrationcurve interval (as in Fig. 2A); the dataset can be matchedto the calibration curve interactively. The &#oating' sedi-mentary dates are plotted on a secondary x-axis, orrelative time scale (here depth, drawn explicitly). Thistime scale can be shifted, stretched and compressed to"nd the best match. Vertically shifting the whole14C series is an option to &correct' for systematic errorsuch as a general reservoir e!ect.

The visual approach is supplemented by the automaticcalculation of a goodness-of-"t (>), which is the distancein 14C years between average sediment and calibrationcurve (spline) data, expressed as a standard deviation. Itsminimum value, corresponding to the best "t, we calls8.,.*/

(minimum s found by wiggle matching; see Ap-pendix A). We use the goodness-of-"t (s

8.,.*/) for two

reasons: "rst, to judge whether the best "t foundis actually good enough; second, to construct realistic95%-con"dence intervals on all time}depth relations.

Even if a perfect dataset is matched to the treeringcalibration curve, there will be errors causing a scatteringof the ages of the geological samples around the curve.We consider the residual variance (the square ofs8.,.*/

as composed of two sources of variance,14C measurement error (s

BP) and sample error (s

4!.1-%).

The 14C measurement error is the usual one-sigma stan-dard deviation of the Gaussian probability distribution.Sample error, or sample time width, allows for the num-ber of years in which a sample formed (deposition time inBennett, 1995) and species error, arising from datingbotanically mixed samples (Mook, 1983; ToK rnqvist andBierkens, 1994).

F-test statistics can be used to evaluate the probabilitythat two sample variances are from the same population(Sokal and Rohlf, 1995). If the observed variance is small-er than predicted, a more complex time}depth relation isunnecessary. If greater variance was observed, we at-tributed this to species errors or the use of an erroneous(too simple) time}depth relation, which we subsequentlycorrected for.

The calibration curve is a nonlinear function. How-ever, if an accurate relative time scale, or x-axis, is se-lected, the sediment data should contain the samenonlinearities (wiggles). Therefore we consider wiggle-matching as a linear regression of relative time (depth,&p, &m) with calendar age. Error envelopes are createdby linear interpolation between the 14C dated levels.A more detailed description of our wiggle-match proced-ure is described in the appendix.

3. Methods of sampling and analysis

The core Eng-XIV was sampled from the Engberts-dijksvenen raised bog area (Eastern Netherlands;52328@N/6339@E) on November 13, 1991. We sampled theupper 189 cm in four stainless steel boxes, from which the33}120 cm interval was studied.

The boxes were taken to the laboratory and storedfrozen. The 0.5 cm-peat slices were cut in a climate room(at !23C) with a precise cutting device. The slices weredivided into subsamples with a stainless steel punch withknown dimensions. Subsamples were taken for microfos-sil analysis (0.72 cm3), macrofossil analysis and 14C AMSmeasurement (4.5 cm3), and chemical analyses and deter-mination of speci"c dry weight (25 cm3).

Macrofossil/14C AMS samples were disaggregated inhot KOH (10%) and washed over a 150 lm sieve. Theremaining material was transferred to demineralizedwater with 5% HCl, from which it was analysed witha Wild binocular (magni"cation 6}50]) and a Leitzmicroscope (100/400]). For 170 consecutive 0.5 cm-sli-ces, volume proportions were estimated of various peatmosses (Sphagnum spp.), and of the roots and vegetativeremains of phanerogams.

Final 14C AMS samples were prepared by manualselection of plant remains (Table 1). The samples werestored in an acidi"ed millipore water solution (pH 2}3),and neutralized and dried. The samples were combustedto CO

2and reduced to graphite for 14C AMS measure-

ment with the Groningen AMS facility (Gottdang et al.,1995; van der Plicht et al., 1995). The radiocarbon datesare corrected for isotope fractionation tod13C"!25&, as measured by the AMS.

The d13C-values reported here were measured separ-ately by one of the Groningen Stable Isotope MassSpectrometers. For most radiocarbon dated samples,carbon concentrations (% of dry weight) were determined.

M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033 1013

Fig

.2.

Tim

e-dep

thre

lations

const

ruct

edby

wig

gle

mat

chin

g44

sele

cted

Eng-

XIV

AM

S14

Cdat

es.A

and

Dha

vetw

ox-

axes

:cal

endar

age

forth

eca

libra

tion

curv

ean

ddep

thfo

rth

era

ised

bog

sam

ple

s.(A

)B

estm

atch

forth

ew

hole

grou

pas

sum

ing

const

anta

ccum

ulat

ion

(cm

/yr;

Tab

le2,

N(1

)).(B

)T

ime}

depth

rela

tion

s(2

p)fo

rth

ebe

st"tca

lcula

ted

fordi!

eren

trel

ativ

etim

esc

ales

assu

min

gco

nst

anta

ccum

ula

tion

orin

put(

d,&

m,&

P).

(C)

Rel

ativ

etim

esc

ales

(mas

s,A

P,N

AP

and

Aln

us),

plo

tted

non-

cum

ulat

iveve

rsusdep

th.(

D)

Bes

tm

atch

afte

rsp

litting

these

ries

into

thre

egr

oup

sof

date

s,as

sum

ing

cons

tant

accu

mula

tion

forea

chin

terv

al(T

able

2,N(3

)).(

E)

Tim

e-de

pth

rela

tions

(2p)

forth

ebes

t"t

obta

ined

with

di!

eren

tre

lative

tim

esc

ales

(d,&

m,&

P)as

sum

ing

cons

tantac

cum

ulat

ion

orin

put

for

each

ofth

eth

ree

inte

rval

s.The

stra

tigr

aphi

calco

lum

nsco

rres

pond

with

(A),

(D),

and

depth

erro

ren

velo

pes

in(B

)an

d(E

).

1014 M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033

Table 1AMS 14C dates of selected plant macrofossils from the raised bog core Eng-XIV, Engbertsdijksvenen, The Netherlands

GrA-number SampleCode

d13C(&)

Carboncontent(%)

14C age(yr BP)

Laboratorystandard deviation(yr BP)

Sample depth(cm)

Sample composition

400 45A !22.33 39.8 1940 40 33.93 S.i.: Stems and leaves (100%)2144 47A !24.90 42.6 1935 40 35.88 S.i.: Stems and leaves ('99%%),

C. vulgaris few above ground rem.2143 49A !26.16 42.3 1930 40 37.83 S.i.: stems and leaves ('99%),

C. vulgaris, 2 #owers2151 50A !26.26 38.78 S.i.: Stems and leaves (100%)2153 50.5 !24.21 42.6 2000 40 39.57 S.i.: leaves, few stems (98.5%), Eric.

roots (1.5%)243 51 !23.68 36.5 2145 20 40.19 S.i.: leaves, few stems (96%), Eric.

roots (3%), above ground rem.(1%)

401 53A !27.13 43.1 2020 40 42.17 S.i.: stems and leaves, 2 sporangia(100%)

2402 55A !30.43 42.8 2040 45 44.14 S.p.: Stems and leaves (100%)2147 57A !30.29 42.8 2040 45 46.12 S.p.: Stems and leaves (100%)402 59A !26.44 42.3 2170 40 48.10 S.c./t/A.: Stems and leaves (100%)

2152 61A !24.68 42.7 2050 45 50.07 S.i.: Stems and leaves (99%),C. vulgaris: 1 #ower

5549 62.5A 2210 50 51.56 S.i.: Stems and leaves (80%),Ericaceae: #owers, leaves, R. alba:some seeds

815 64A !24.06 42.4 2140 45 53.04 S.i.: stems ('97.5%), Erc.: roots(2%), Indet.: epiderm (0.5%)

817/1345 66A !26.83 2115 45 55.02 S.i.: Stems and leaves (100%)403 67A !25.62 42.5 2240 40 56.01 S.i.: leaves, some stems (995),

E. tetralix: 5 branches2698 67.5 !27.01 2210 45 56.50 S.i.: Stems and leaves (95%), C. vul-

garis: 10 branches, E. tetralix:1 branch (5%)

814 68A !26.04 43.1 2140 45 57.00 S.i.: leaves, some stems ('99%)Eric.: roots ((1%)

812 69A !27.46 47.1 2150 45 57.99 S.: stems (34%), R. alba: 1 seed;Eric. #owers, branches and leaves(64%)

3095 69.5A !25.95 2410 140 58.48 Cyper.: epiderm indet (100%)2699 69.5B !30.70 2205 45 58.48 Eric. roots (d: 0.5}3 mm; 100%)652 70A !28.71 51.8 2260 35 58.98 Eric. roots (d: 0.5}3 mm; 100%)679 70B !28.63 58.98 Eric. roots (d: (1 mm; 100%)

2408 70C !26.54 44.7 2275 45 58.98 S.i.: leaves, few stems (87%), Eric,roots (3%); C. vulgaris: charredremains (10%)

2777 72 !25.77 2280 45 60.95 S.A.: stems ('99%), Eric.: roots((1%)

2767 72.5 !27.44 2285 45 61.45 S.A.: stems (100%)818 73A !28.13 46.7 2190 45 61.94 Eric.: roots (d: '1 mm. 100%)644 73B !22.89 42.8 2295 40 61.94 S.i.: stems, some leaves (100%)

2718 74.5 !28.33 2365 45 63.43 S.A.: stems, some leaves (100%)2689 76 !25.28 2280 45 64.92 S.i.: stems & leaves (100%)404 77A !25.28 41.5 2385 40 65.92 S.i.: stems & leaves (100%)

2688 78 !30.70 2415 45 66.91 S.i.: stems & leaves (100%)244 79 !23.72 41.4 2535 20 67.90 S.i.: stems & leaves (97%);

Eric.: roots (3%), few branches646 81A !21.49 45.0 2450 40 69.87 S.i.: stems & leaves (100%)245 83A !23.25 39.5 2595 20 71.85 S.i.: stems & leaves (96%),

Eric.: roots (4%), few branches654 85A !25.32 40.3 2475 40 73.83 S.A/i.: stems & leaves (100%)811 87A !25.64 41.5 2470 45 75.81 S.i.: stems and leaves (100%)

2398/3232 87B !23.88 32.4 2410 40 75.81 Erioph. vagniatum: veg. remains(95%); Eric.: roots (2%); S: (2%)

(Continued on next page)

M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033 1015

Table 1 Continued

GrA-number SampleCode

d13C(&)

Carboncontent(%)

14C age(yr BP)

Laboratorystandard deviation(yr BP)

Sample depth(cm)

Sample composition

246/256 87C !22.24 40.9 2600 20 75.81 S.: stems and leaves (92%); erioph.vaginatum: veg. remains (5%);Eric.: (roots (3%)

2145 89A !21.28 43.5 2490 40 77.79 S.A./c./t.: stems and leaves (90%);Eric.: some above ground rem.;R. alba: 3 seeds; Scirpus caes-pitosus: 2 seeds (10%)

2220 89B !25.33 2240 60 77.79 R. alba: 21 stem bases (100%)249 91A !23.68 40.2 2580 20 79.76 S.i.: stems and leaves (100%)250/259 91B !23.68 40.2 2560 20 79.76 S.i.: stems and leaves (98%);

Eric.: roots (1.5%), few branches251/642 95A !24.86 43.0 2550 20 83.73 S.i.: stems and leaves (99.5%); Eric.:

some branches and leaves,roots ((3%)

252 95B !23.51 41.9 2560 20 83.73 S.: stems and leaves (90%); Eric.:some branches and leaves (1%),roots (3%), Erioph. vaginatum (3%)

816 97A !31.30 53.5 2460 45 85.71 E. tetralix: branches and leaves(15%), C. bularis: branches andleaves (85%)

643 97B !23.53 43.3 2530 45 85.71 S.i.: stems and leaves (99%); Erioph.vaginatum (1%)

3231 97D !23.71 2375 70 85.71 Erioph. vaginatum: veg. remains(100%)

248/257 99A !24.03 42.0 2540 20 87.69 S.i.: (95%); C. vulgaris: aboveground remains (3d), roots (2%)

2397 101B !24.57 25.5 2445 45 90.99 S.A.: stems and leaves (95%); Eric.:roots (3%); Eroph. vaginaturm (1%)

645 103A !24.48 41.4 2480 40 92.93 S.A.: stems and leaves (100%)247/258 103B !24.73 44.4 2595 30 92.93 S.A.: stems and leaves (90%); Eric.:

roots (4%); Erioph. vaginaturm(2%)

655 105A !28.01 44.0 2525 35 94.87 S.A./i.: stems and leaves (99%);C. vulgaris: 2 branches; R. alba:1 seed

3050 1065A !28.32 2770 130 96.32 C. vulgaris: 54 branches, 17 #owers/#. bases; E. tetralix: 1 branch; R.alba: 1 stem base; Pohlia cf. nutans:6 branches with leaves; S.: 3 oper-culi, 1 sporangium, 50 stems

405 107A !23.83 61.0 2970 45 96.80 S.: (20%); Eric.: (40%); Indet.: 40%(100% charred remains)

2701 109.5/110 !27.86 2715 45 99.48 S.: 1 sporangium, 184 stems, 59small branches with leaves (47%);C. vulgaris: 40 branches, 21#owers/#. bases, 5 seeds; E. tetralix:24 seeds (47%); Pohlia cf. nutans:9 branches with leaves (6%)

653 111A !26.12 42.6 2765 40 100.69 S.A.: stems & leaves (99%); Eric.:roots (1%)

406 115A !24.25 35.9 2775 40 104.58 S.A.: stems (100%)641 117A !20.31 43.6 2775 45 106.52 S.A.: Stems, some leaves (100%)

2148 119A !26.22 41.9 2720 40 108.47 S.A.: Stems & leaves ('99%);C. vulgaris: some branches

407 121A !23.80 40.4 2765 45 110.41 S.A.: stems, some leaves (100%)2407 125A !26.01 42.7 2790 45 114.30 S.A.: stems, 5 sporangia (100%)2687 126A !27.54 2825 60 115.27 C. vulgaris: small rootlets (70%); S.:

stems & leaves (15%); Indet.: 15%2719 126/126.5 !26.67 2855 45 115.52 S.A./c./t.: stems (70%); C. vulgaris

branches (20%); R. alba: seeds/stem bases (10%)

1016 M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033

Table 1 Continued

GrA-number SampleCode

d13C(&)

Carboncontent(%)

14C age(yr BP)

Laboratorystandard deviation(yr BP)

Sample depth(cm)

Sample composition

2411 127A !25.8 48.7 2870 50 116.24 C. vulgaris: 55 branches, 7 #owers;E. tetralix: 8 branche (Eric. 40%);R. alba: 16 seeds (30%); S.: 65 stems(30%)

2779 126.5A/127.5B!29.92 2915 70 116.73 C. vulgaris: charred remains (100%)2721 127.5A !28.21 2885 85 116.73 C. vulgaris: #owers and branches

(40%); R. alba: seeds and stembases (40%); S.: stems (20%)

2686 128/128.5 2795 60 117.46 C. vulgaris: branches (20%); S.:(35%); R. alba: (45%)

408/3094 129A !25.77 45.4 2875 40 118.19 S.A.;t.: stems, sporangia, someleaves (97%); Eric.: roots (3%)

Note: Sample compositionA: Andromeda polifolia; C: Calluna vulgaris; E: Erica tetralix; Eric.: Ericaceae; R: Rhynchospora alba; S.A: Sphagnum sect. Acutifolia; S.c.: S. cuspidatum;

S.i.: S. imbricatum; S.p.: S. papillosum; S.t.: S. tenellum

Microfossil preparation included the adding of oneLycopodium tablet (13,500$500 spores/tablet) for thecalculation of pollen concentrations (Stockmarr, 1971).Speci"c weights were obtained in duplicate, dryingsamples at 1053C for 24 h and weighing immediatelyafterwards.

The detailed study of Eng-XIV (66 AMS dates) wassupplemented with 15 dates on contemporaneous plantmatter from the Meerstalblok bog reserve (NE Nether-lands), measured by the Groningen ConventionalLaboratory. Some radiocarbon results for the Eng-XIVraised bog core were published before (Kilian et al., 1995).One conventionally dated core from the same raised bog(Eng-Ib; van der Molen and Hoekstra, 1988) was wigglematched by the same method to extend our studyof relative time scales, as speci"c weight and pollenconcentration data were available.

4. Comparing relative time scales for raised bog deposits

4.1. Matching 44 AMS 14C dates from Eng-XIV

From an 85 cm-interval of the raised bog core Eng-XIV, 66 samples were AMS 14C dated, 45 consistingwholly of above-ground macrofossils (Fig. 2, Table 1).The latter were matched to the calibration curve usingdepth as relative time scale (Fig. 2A; Table 2, N(1)). Themain atmospheric 14C variations (wiggles) are clearlyrepresented in this series of dates, but in speci"c intervals,120}105 and 68}45 cm, correspondence with the treeringcurve is rather bad. Nevertheless, a preliminary time-depth error envelope was calculated assuming a linearrelation between depth and calendar time (Fig. 2B).

Substitution of the depth scale by cumulative mass(&m) or pollen (&p) changes the relative position of the

samples, ascribing most calendar time to samplesshowing high pollen concentrations and speci"c weightrespectively (Fig. 2C).

Speci"c weight was determined in duplicate, on small(0.72 cm3, n"168) and large (25 cm3: n"175) samples.It was calculated as the ratio of sample weight to volumeincluding voids, and therefore equals unit dry weight andnot speci"c gravity (Galvin, 1976). For the two sets ofsamples, speci"c dry weight was 0.084 ($0.015) and0.085 ($0.011) [g/cm3] respectively. Because of theirhigher number and lower variance, the large sampleswere used as the time scale based on cumulative mass(&m). As speci"c weight is rather constant throughout thecore, its error envelope almost coincides with that of thestraight time-depth relation (Fig. 2B). The "t is slightlybetter (s

8.,.*/lower), as the dates between 95-30 cm

move to slightly older ages.Four &p time scales were initially tried: &AP (Arboreal

Pollen), &AP excluding Alnus, &Alnus and &NAP (NonArboreal Pollen). Average AP (excluding Alnus) con-centrations are 21.4]103 ($15.4]103) pollen/cm3,which is roughly equal to Alnus values (20.9]103$17.0]103 pollen/cm3) and much higher than NAP(4.5]103$3.2]103 pollen/cm3). The two pollen timescales that produced the best match with the treeringcurve, viz. &AP (excluding Alnus) and &NAP, arediscussed.

The &AP (ex. Alnus) curve #uctuates around the lineartime-depth relation, but except for 120}100 cm where itprecedes the depth error envelope, and 46}33 cm where ittends to lag behind, di!erences are small. These are morepronounced in a time-depth relation based on &NAP (seeFigs. 2B and 3). While 14C dates in the plateau of95}65 cm are shifted to slightly higher calendar ages,bottom and top appear much younger compared toother x-axes. The younger age of the bottom of the core is

M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033 1017

Fig. 3. Best match for the 44 Eng-XIV samples assuming constant NAP in#ux (Table 2, N(1)).

considered realistic. It seems highly unlikely however,that the 14C dates between 70}55 cm are also younger as awhole, as the fall and rise in peat 14C ages reproduce the400}300 BC wiggle. That the "t is better (s

8.,.*/lower;

see Table 2, N(1)) than for the other axes therefore doesnot mean we accept this "t as the best solution (Fig. 3).

The cumulative pollen and mass time scales corres-pond with features of raised bog stratigraphy. Inferredchanges in accumulation rates take place on exactly thesame depth levels (Fig. 2B, shaded). Out of "ve zones(1}5) marked by high speci"c weights and pollen concen-trations, four are associated with local presence ofEricaceae, mostly Calluna. One (3 in Fig. 2B) coincideswith a S. cuspidatum/papillosum-layer (see below).

Short-term #uctuations are identical for all pollengroups, producing similarly shaped con"dence intervals.

Due to increasing human impact on the vegetation (vanGeel, 1978) however, &NAP increases relative to &AP(ex. Alnus) on the longer term, which leads to distinct ageintervals. The time scale is in#uenced by the selection ofpollen types (see also Middeldorp, 1982).

If all peat 14C dates are wiggle-matched using the samerelative time scale, sudden accumulation rate changesand hiatuses express themselves as general deviationsbetween the peat and treering data (Fig. 2A). The prelimi-nary linear time-depth relation can be improved prag-matically by subdividing the core into stratigraphicalunits, each of which can be matched separately to thecalibration curve. The best results for the Eng-XIV corewere obtained by dividing it into three units: the bottom,dominated by Sphagnum sect. Acutifolia (117.5}99.5 cm);a central part, of which the top is the 60}56 cm Ericaceae

1018 M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033

horizon (96.3}56 cm); and the top interval above(55}33.9 cm). The same four relative time scales as abovewere used again to match the three peat 14C series to thetreering curve. The assumption of constant accumulation[cm/y] per segment results in a close correlation of thepeat data with the calibration curve (Fig. 2D). Twodistinct features de"ne the stratigraphical intervals:a higher accumulation rate of Sphagnum sect. Acutifoliacompared to S. imbricatum, which is exceptional com-pared to other cores, and an apparent hiatus correspond-ing with the top of the 60}55 cm Ericaceae horizon. This130-year hiatus is bracketed by two 14C samples only1 cm apart.

Irrespective of the relative time scale applied, the good-ness-of-"t improves (s

8.,.*/decreases) though in di!er-

ing amounts. For the initially most successful x-axis(&NAP) s

8.,.*/drops only little, because the position

of the 60}55 cm date group hardly changes. It tendsto &"ll up' the hiatus with pollen accumulated in theEricaceae-dominated horizon (Figs. 2E and 3). Con-"dence intervals for the bottom of the core almostconverge completely.

The invariable e!ect of both cumulative pollen andmass axes is to attribute more time to layers domin-ated by Ericaceae, and perhaps aquatic Sphagna.However, we suggest with the "nal depth-based corre-lation of peat 14C dates with the calibration curve(see Figs. 2D and 5A) that the Ericaceae-horizons rough-ly accumulated at the same rate as the surroundingSphagnum peat.

4.2. Matching 11 conventional 14C dates from Eng-Ib

The approach described above was also applied toa 100 cm core from the same bog, Eng-Ib, for which 11conventional 14C dates were obtained in the past (vander Molen and Hoekstra, 1988). Fig. 4A shows a prelimi-nary match assuming a straight time}depth relation forall 14C dates (Table 2, N(0)). The resulting depth-basedcon"dence interval is compared with two relative timescales applied by van der Molen and Hoekstra (1988),&AP including Alnus and &m (Fig. 4B).

Total tree pollen concentrations average 1.2]105($9]104)/cm3. Because average tree pollen in#uxesdecrease towards the top of the core, the &AP-deptherror envelope is slightly concave. Speci"c weight valuesaverage 0.066 ($0.008) (g/cm3). The &m-depth envelopeis slightly convex, due to higher bulk densities in the top(Fig. 4C).

These shapes are basically unchanged when the bot-tom 14C date, taken below a Scheuchzeria palustris peatlayer, is removed from the x-axis, and the top 10 dates arematched again (Fig. 4D). This improves the "t for allrelative time scales. Most of the residual standard devi-ation (s

8.,.*/) is now caused by the anomalous 9.5 cm

date. Because the &AP-axis attributes a higher calendar

age to this level, the "t is numerically slightly betterthan of the straight time}depth relation (Table 2,N(1)). For a more detailed evaluation of the choiceof time axis we refer to the discussion and conclusionparagraphs.

van der Molen and Hoekstra (1988), applying Clymo'smodel to the data, concluded that the &m}age relationwas concave. As the raw &m}depth data are convex(Fig. 4C), the concavity noted by van der Molen andHoekstra (1988) must result from their iterative calcu-lations, assuming continuity of accumulation betweenthe two deepest 14C dates. Assuming constant accumula-tion, but positioning the "rst 14C date at the averageof its calibrated (BC) age range, produces the time}depthcorrelation of Fig. 4E. The 800}300 BC period represent-ed so well in the Eng-XIV AMS series, seems to bemissing from this core. If so, van der Molen andHoekstra (1988) quoted in Tolonen et al., 1992modelled a 500 yr hiatus as long-term methane loss to theatmosphere.

Site factors account for physical di!erences betweenthe Eng-Ib and Eng-XIV cores. Eng-Ib re#ects a hollow/lawn vegetation, in which Sphagnum papillosum andS. cuspidatum dominate, and Ericaceae are almostabsent. High &AP and &m values, unlike Eng-XIV, areassociated most with rather wet situations (Fig. 4C).

4.3. Factors contributing to errors in peat dates

4.3.1. The Engbertsdijksvenen datasetFig. 5A shows the "nal correlation of all Eng-XIV

14C AMS dates with the calibration curve. We will dis-cuss here this large dataset in terms of six botanicalcategories: (1) pure Sphagnum, (2) Sphagnum with mac-rofossils, (3) charred material, (4) cyperaceous samples,(5) Ericaceae roots and (6) Spagnum with ericaceousrootlets.

(1) Pure ('98%) Sphagnum samples (n"37) havebeen the main basis for the correlation, as they wereexpected to reproduce atmospheric 14C values. Exceptfor one sample (64.9 cm) the correlation is almostperfect. Even the single deviant sample may re#ectnatural 14C variation, as the peat samples encompassonly c. 6 yr. In Sphagnum imbricatum-dominated layersgenerally enough Sphagnum could be collected for AMSdating.

(2) Sphagnum with higher quantities of other above-ground macrofossils (n"9) were used for the correlationas well, especially in Sphagnum sect. Acutifolia-strata,where the peat contains many Ericaceae roots, and lessSphagnum. Deviations from atmospheric 14C are notobservable in the results.

(3) Charred material was collected from two levels.The sample that consists of identi"ed Calluna vulgarisremnants (116.7 cm) returns a 14C age consistent withpure Sphagnum, which points to a local origin of the

M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033 1019

Fig

.4.

Tim

e-dep

thre

lations

cons

truc

ted

byw

iggl

em

atch

ing

11an

d10

conv

entiona

l14C

dat

esfrom

raised

bog

core

Eng

bert

sdijks

venen

Ib.(

A)

Bes

t"tas

sum

ing

cons

tantac

cum

ulat

ion

(Tab

le2,

N(0

)).(B

)T

ime-

dep

thre

lation

s(2

p)fo

rth

ebe

st"t

obt

aine

dw

ith

di!er

ent

rela

tive

tim

esc

ales

(d,&

m,&

P)as

sum

ing

cons

tant

accu

mula

tion

orin

put

.(C

)R

elat

ive

tim

esc

ales

(mas

s,A

P)pl

otted

noncu

mul

ativ

eve

rsusde

pth

.(D

)B

est"tas

sum

ing

const

antac

cum

ula

tion

(Tab

le2,

N(1

)).T

hebot

tom

date

ispo

sitioned

atth

eav

erag

eofi

tsca

libr

ated

age

inte

rval

,95%

-con"de

nce

lim

itsin

dic

ated

byth

eda

shed

box.

(E)

Tim

e}dep

thre

lations

(2r

)fo

rth

ebes

t"t

obta

ined

with

di!er

ent

rela

tive

tim

esc

ales

assu

min

gco

nsta

ntac

cum

ulat

ion

or

input

(d,&

m,&

P).

The

stra

tigr

aphi

calco

lum

nsco

rres

pond

with

(A),

(D),

and

dep

ther

ror

enve

lopes

in(B

)an

d(E

).A

lso

the

cons

tant

deca

ytim

e/dep

thcu

rve

(Cly

mo,

1984

)is

show

nher

e.

1020 M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033

material. The sample at 96.8 cm appears 350 14C yrtoo old: the 40% unidenti"ed charred wood fragments inthe sample probably originate from man-induced forest"re around the bog (Kilian et al., 1995; Kilian et al.,in prep.).

(4) Two cyperaceous samples are Eriophorum va-ginatum, a sedge forming tussocks on many raised bogsites. It is an opportunistic species which prefers dryhummocks, but is also able to resist prolonged wettingbecause of its aerenchymatic roots (Barber, 1981). In-growing of Eriophorum, transporting carbon to its stemsand roots, explains the too young 14C age of the samples(75.8 and 85.7 cm).

One cyperaceous sample consisted of 21 Rhynchosporaalba stem bases (77.8 cm). It yields a date much tooyoung, but it is unlikely to have rooted 20 cm into thepeat. Because of the small sample size (0.4 mg), onlynatural or human contamination seems a possibleexplanation.

(5) Samples of thick Ericaceae (Calluna) roots (n"4)show 14C variation di!erent to the treering curve.Ericaceae on bogs develop a lax growth form, of whichthe stems when overgrown by Sphagnum change intothick roots. Though the number of samples is small, thepattern can be explained by assuming the roots alive (andin atmospheric equilibrium) until the plants "nally died,some 100 yr (or 5}10 cm of peat) later. The 3 datesbetween 62}58.5 cm probably re#ect the 14C history ofthe hiatus, 300}150 BC, but 100 yr &earlier'. The degree towhich individual dates are composed of roots from di!er-ent levels however, is unknown.

(6) Sphagnum samples containing 2}4% of verysmall ericaceous rootlets (n"11) in the majority (6 or7) of cases show ages signi"cantly higher than pureSphagnum samples. This will be discussed in more detailbelow.

4.3.2. The reservoir ewectOlder 14C ages can be caused by the uptake of carbon

from an older source. This is termed &reservoir e!ect'(Stuiver and Pollach, 1977). Olsson (1983) discussesexamples of reservoir ages from early 14C studies. Kilianet al. (1995) demonstrated wider presence of this e!ectin raised bogs by wiggle matching four other raisedbog cores, of which bulk samples were conventionally14C dated. Because a reservoir e!ect of 100}250 14C y isrelated to such small contamination (2}4% rootlets), thecarbon source directly implied is methane (CH

4) produc-

ed in the anaerobic, deeper part of the bog. Wespeculated about two possible paths through which theCH

4could reach the rootlets, after oxidation to CO

2by

bacteria:

(1) uptake through interaction with mycorrhizal fungi,which are known to a!ect nutrient budgets of their host,or

(2) uptake by fungi feeding on decaying root tissue,and for carbon on CH

4/CO

2escaping from deeper levels

of the bog.

For further study of the reservoir e!ect we tried toconcentrate fungal material in a 10 lm-sieve, which failedas this mesh size was still too large. Therefore we tried tocon"rm the reservoir e!ect by 14C dating living plants,collected on various sites in the Meerstalblok raised bogreserve (Eastern Netherlands). They were cleaned andcut into macrobotanically homogeneous subsamples(Table 3). Except for one sample (an old peat surface), allmaterials have modern 14C activities (14a in pMC, per-cent Modern Carbon). Modern 14C activities forEriophorum tussocks and rootlets, the latter penetratingsome decimetres into the peat, was expected and ob-tained. The very small, modern pMC value range forSphagnum spp. con"rms our view that these species areextremely suited for 14C AMS dating. Even S. cus-pidatum, growing on the old peat surface, does not havelower 14C activity despite the proximity of a possibleold-carbon source.

The pattern within Calluna and Erica however isopposed to our geological study: the roots, especiallythe smallest rootlets collected, have the highest activities.This apparent disagreement between past and presentmay have various origins. Firstly, all factors involvedhave been subject to drastic human impact. Theatmospheric 14C activities, which rose to a maxi-mum of ca.200 pMC in 1963 due to nuclear testing, isfalling since (Olsson, 1983, 1991). The (seasonally aver-aged) atmospheric 14C activity value for 1995 is110 pMC, as monitored by the Groningen laboratoryin nearby Smilde (Meijer et al., 1995). Possibly the14C activity of the ericaceous rootlets simply lagsbehind that of the photosynthesizing parts. Due to peatcutting, the hydrology, ecology, and carbon balanceof the bog reserve may also di!er from virgin raisedbogs.

More fundamentally, the reservoir e!ect is associatedwith geological samples containing Ericaceae roots, butclearly not always present. This suggests that a complexof factors is included, such as the availability of CH

4, and

probably the presence of crucial but unknown fungi.That the only &aged' sample of ENG-Ib is associated withthe very maximum of fungal hyphae in the core (van derMolen and Hoekstra, 1988, Fig. 14) is more indirectevidence for our hypothesis. This however still says noth-ing on the mechanisms or pathways involved. Kerley andRead (1995) showed that the ericoid fungus Hymenos-cyphus ericae can degrade lignin and hexosamines, andtherefore it could also thrive on (sub)fossil carbon.Because Calluna rootlets consist partly of mycorrhizalfungal material, it is also possible that the old carbon&only' remains in these fungi and species feeding onthem.

M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033 1021

Tab

le2

Num

eric

alre

sults

ofth

era

ised

bog

core

sEng.

XIV

and

Eng-

Ibw

iggl

em

atch

edw

ith

variou

sre

lative

tim

esc

ales

.The

bes

tso

lution

sar

eplo

tted

inbol

dan

dsh

ow

nin

the"gu

res

1022 M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033

Fig. 5. Final plot showing all Eng-XIV carbon isotope data. (A) Best "t obtained by wiggle matching the dates as three groups to the calibration curve,now showing all samples dated. (B) d13C for three sample groups. (C) Carbon percentages (% dry weight) for the same groups.

There is also a small possibility of a reservoir e!ect inSphagnum species themselves, occurring through directuptake of aged CO

2from the bog water. This hypothesis

is based on data of Proctor et al., (1992), who determinedd13C-values of di!erent bryophytes. Sphagnum cus-pidatum showed an anomalously low d13C (!34.7&),

M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033 1023

Table 3AMS and conventional 14C dates of recent plant fractions from the bog reserve Meerstalblok (Netherlands)

Laboratory code Laboratory number 14a (%) Laboratory standarddeviation

d13C (&) Sample composition

AMS dataGrA 1361 122.40 0.70 !26.04 Calluna vulgaris: roots (diameter (2 mm)GrA 1362 122.60 0.60 !24.87 Erica tetralix: roots (diameter (0.5 mm)Conventional dataGrN 21218 112.33 0.52 !26.18 S. cuspidatum on old peat surfaceGrN 21219 111.62 0.38 !28.48 Calluna vulgaris: branches and leavesGrN 21220 111.76 0.71 !29.59 Calluna vulgaris: #owersGrN 21221 113.39 0.32 !28.04 Calluna vulgaris: rootsGrN 21222 112.56 0.52 !25.67 S. paillosumGrN 21223 112.73 0.53 !27.26 Erica tetralix: #owersGrN 21224 112.97 0.49 !29.23 S. recurvumGrN 21225 112.56 0.49 !26.66 S. tenellumGrN 21226 91.79 0.35 !24.69 Old peat surfaceGrN 21227 113.25 0.27 !26.01 Eriophorum sp.:leaf basesGrN 21228 112.39 0.74 !26.77 Eriophorum sp: rootletsGrN 21229 113.28 0.42 !26.85 Erica tetralix: branches and leavesGrN 21230 120.16 0.55 !26.11 Erica tewtralix: roots (d'2 mm)

which can be explained with its preference for aquatichabitats with possibly diverging 14CO

2-concentrations.

Geochemical con"rmation of this may be found in thehigh reservoir e!ect found for Draved Mose (240 14C yr;Kilian et al., 1995), which is the only high-resolutiondated bog core dominated by S. cuspidatumfor a prolonged period.

Raised bog plants, as aquatic plants (Marcenco et al.,1989), show a strong relation between species and d13C-values. In our data, Sphagnum imbricatum and S. sect.Acutifolia samples have an average d13C of !25.3&.S. papillosum is more depleted in the heavier isotopes(d13C below !30.3&), as are Ericaceae (average!28.7&) * see Fig. 5B.

4.4. 14C variation in some conventionally dated mire cores

In the previous paragraphs we identi"ed three majorsources of 14C variation within raised bog deposits:

(1) error due to botanically mixed samples, leading toindividual or systematic deviations,

(2) hiatuses, causing sudden leaps in 14C age of a se-quence, and

(3) stratigraphic change, indicating changing localconditions, and therefore possibly of changing net accu-mulation rate.

Medium-term accumulation rate variations suggestedby relative time scales composed of cumulative pollen ormass could not be con"rmed, leaving depth (or height) asthe preferred time scale. This justi"es using depth as

a relative time scale for wiggle matching other conven-tionally dated mire cores. This was also done to con"rmthe sources of 14C variation discussed, to stress the ad-vantages and some limits to the approach, and underlinethe di!erences with theoretical peat accumulation mod-els (Middeldorp, 1982; Clymo, 1984).

Four conventionally dated pro"les from the Engberts-dijksvenen (Eng-I: van Geel, 1978; Eng-II: van Geeland Dallmeijer, 1986; Eng-V: Middeldorp, 1982; Eng-VII: Dupont and Brenninkmeijer, 1984) complete ourpicture of peat accumulation in this bog, and servefurther discussion of Middeldorp's (1982) model(Fig. 6). Three other raised bog sequences (AgeroK dsMosse: Nilsson, 1964; Draved Mose: Aaby and Tauber,1975; Varassuo: Donner et al., 1978) were matchedto highlight quantitative (age) and qualitative (func-tional) disagreements with Clymo's (1984) model(Figs. 7 and 8).

Rather than gradual changes in the slope of the peat14C data, wiggle (or calibration curve) matching showedsharp changes in slope, and leaps in 14C age, both coin-ciding with major stratigraphic changes in the mires. Byprogressively subdividing the peat series into more ho-mogeneous stratigraphic units, we improved the "t withthe calibration curve. The numerical results of this pro-cedure are presented in Table 4.

Besides stratigraphical transitions, two other limits toour approach are visible (Figs. 6 and 7). First, the numberof dates for wiggle matching must always exceed two,because any two dates give at least one perfect "t with thetreering curve. Second, due to species error the ages of

1024 M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033

Tab

le4

Num

eric

alre

sults

ofth

eE

ngb

erts

dijk

sven

enra

ised

bog

core

s,w

iggl

em

atch

edw

ith

adep

thsc

ale.

The

bes

tso

lutions

are

plo

tted

inbold

and

show

nin

the"gu

res

Abr

evia

tion

stra

tigr

aphy

'is

succ

eeded

by/

RE

mat

ched

corr

ecting

for

rese

rvoir

e!ec

t;A

cut.

Acu

tifo

lia;c

usp

cusp

idat

um;E

riop

h.E

riop

horu

mva

gina

tum

;im

br.i

mbr

icat

um;M

agel

l.m

agel

lani

cum

;Mol

.M

olin

iaca

erul

ea;P

Pea

t;pa

p.pa

pill

osum

;S.

Spha

gnum

;Sch

euch

zSc

heuc

hzer

iapa

lust

ris

Fig

.6.

Tim

e}de

pth

rela

tion

sfo

rth

eEngb

erts

dijk

sven

enco

res.

Dep

ther

roren

velo

pesin

dic

ate95

%co

n"de

nce

inte

rval

susing

stra

ight

tim

e}de

pth

rela

tions

.Thehei

ghtsc

aleis

forEng-

I,II

,Van

dV

II;

dept

hfo

rE

ng-X

IVan

dE

ng-

IB.T

hecu

rve

isM

iddel

dor

p's

(198

2)poly

nom

for

Eng-

V.

M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033 1025

Fig. 7. Time}depth relations for AgeroK ds Mosse, Draved Mose and Varassuo. Depth error envelopes indicate 95% con"dence intervals using straighttime}depth relations, assuming no species error. The depth scale (left) is for AgeroK ds Mosse and Varassuo; height (cm, right) for Draved Mose. Curvesare Clymo's (1984) model results, using constant speci"c weight.

some linear segments show an arti"cial overlap, as in thetop of Draved Mose, or gap, as in Eng-I. This resultsfrom our age correction for the intervals in which wigglesmade the recognition possible of a reservoir e!ect, whileleaving this aside for segments without such evidence.

Our overview of the Engbertsdijksvenen sections indi-cates two important times of change in the bog. Anapparent interruption of the formation of Eriophorumvaginatum/Scheuchzeria palustris-dominated transitionalpeat is shown by a sudden 14C leap in three peat cores.Con"rmation of this feature as a hiatus, and "re as itsprobable cause, is found by high charcoal percentages inEng-I and an omnipresent Molinia caerulea-layer abovethis approximately synchronal level. Later, between 850nd 750 cal BC, net peat accumulation increases about50% when Sphagna of the section Cymbifolia start todominate the moss layer.

Similar results are obtained for AgeroK ds Mosse andDraved Mose (Fig. 7). Two further key levels of change inAgeroK ds Mosse constitute the transition from algaeousgyttja, representing open water, to Phragmites reed-swamp, and the subsequent transition to Sphagnumspecies, or raised bog conditions.

The wiggle matching results can be compared to thecalculations of Middeldorp (1982) for Eng-V, andClymo's (1984) model for AgeroK ds Mosse, Draved Mose,Varassuo (Clymo, 1984) and Eng-Ib (van der Molen andHoekstra, 1988).

Middeldorp's polynomial (Fig. 6) shows a systematicerror compared to wiggle matching, due to the use ofa di!erent calibration curve. More important, as in Eng-XIV, the time represented by the hiatus is ascribed to thepeat below because the polynomial assumes continuousaccumulation and includes one sample (20 cm) contain-ing older peat and Molinia roots. The erroneous assump-tions of continuous accumulation and dating accuracyserve to attribute vastly younger ages to the 10}30 cmsamples, while their resolution appears unrealisticallylow (many yr/cm).

For four mire sequences we recalculated an alternativetime}depth relation using Clymo's (1984) equation, as-suming constant speci"c weight for each sequence(AgeroK ds Mosse 0.069, Draved Mose 0.1 as Clymo (1984)and Eng-Ib 0.066 g/cm3 as above) and the parameters ofthe original authors (Clymo, 1984; van der Molen andHoekstra, 1988). The results are shown in Fig. 7, while

1026 M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033

Fig. 8. Age di!erences of Clymo's (1984) model with wiggle matching assuming a straight time}depth relation for four raised bog sequences (Eng-Ib,AgeroK ds Mosse, Draved Mose and Varassuo).

Fig. 8 shows the age di!erences of the model with wigglematching using a depth scale. Generally speaking, thetwo agree best in the top and bottom of the mire sections,but the curve rarely coincides with our error envelopes.Though this could be explained in part by our under-estimation of the e!ect of botanically mixed samples,average deviations between the models are too large:circa 67 yr for Eng-Ib, 178 yr for Draved Mose, 434 yr forAgeroK ds Mosse and 1422 yr for Varassuo.

The 14C pattern of Varassuo could at "rst not beexplained, because stratigraphic information is crude and14C variability very large. When the "t of the centralinterval was optimized however, two 14C age overlapsappeared in top and bottom. These appeared to be twooverlaps in real ages, both more than 500 calendar years,of three segments (Fig. 7), as the levels of 14C age leapscoincide with boundaries between separate cores, takenwith a piston corer (Donner et al., 1978).

Clymo's (1984) model "ts the data only in the crudecontrast between generally low accumulation in the earlystages of peat formation and higher rates in the lateHolocene. By assuming constant input and continuousaccumulation, major 14C changes in the deposits remainunnoticed. Besides the sources of variation mentionedalready (species error, hiatuses and accumulation rate),this includes sample error.

5. Discussion

Raised bogs are convex landforms which maintain andextend themselves by impeded drainage (Ingram, 1982,1983). Their top layer, or acrotelm, which supports theliving vegetation, ensures a gradual discharge of the year-ly precipitation surplus over the bottom layer, thecatotelm (Ingram, 1978). This structure maintains a dy-namical steady state by a number of negative feedbackmechanisms (Joosten, 1993). Variations in precipitationand water level can be met by species changes at thesurface. Falling bog water results in a decrease of Sphag-num, and an increase of Ericaceae, and therefore anincrease of unit dry weight (Figs. 2 and 4) and decline inconductivity of the acrotelm; rising water tables cause theopposite. The self-stabilising tendencies of a raised bog,discussed elaborately by Joosten (1993) in fact underlayour use of one or more straight time}depth relations forwiggle matching.

This would be supported by Ingram's (1982) applica-tion of Marino's equation for ground water mounds tobogs. While this method does have quantitative prob-lems, some of its qualitative conclusions are very useful.First, Ingram (1982) concludes that &growth at all pointson the surface of a raised bog is likely to proceed atsynchronal rates, even under #uctuating climates'. The

M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033 1027

assumption implicit in this is an unchanged quantitativerelation between net recharge and hydraulic conductiv-ity. This only holds if major species changes are absent,which allows us to use di!erent time}depth relationsfor distinct homogeneous stratigraphic units. Second,given a certain species composition, accumulationrates for Holocene bogs should converge as they increasein size. The accumulation rate changes assumed byMiddeldorp (1982) and Clymo (1984) disagree with thisapproach.

5.1. Pollen density dating

Middeldorp's (1982) &pollen density dating' typicallyoperates on a depth scale of decimeters, and a time scaleof 100}1000 yr. The pollen concentrations correlate cha-nges in accumulation rate with microrelief. Raised bogsare dominated over large areas by a hummock-hollowtopography (Casparie, 1972; Walker and Walker, 1961;van der Molen and Hoekstra, 1988). The driest parts(hummocks) are covered by Ericaceae, most Calluna,sometimes accompanied by Scirpus caespitosus orEriophorum vaginatum. The intermediate slopes andlawns are dominated by various Sphagna. The wet hol-lows and pools are inhabited by Sphagnum cuspidatum,Eriophorum angustifolium and in weakly minerotrophicconditions Scheuchzeria palustris.

Remarkably, high pollen concentrations are observedin the environments forming the extremes on the wet}drygradient. Our high concentrations of regional pollen forpeat in which Calluna was locally present are con"rmedby studies of other bogs (Dickinson, 1975; Dupont, 1985;Middeldorp, 1984, van der Molen, 1992). Similarly raisedpollen concentrations occur in Sphagnum cuspidatum-pools (Eng-Ib, see above). Middeldorp (1984) also re-ported high pollen concentrations in Eriophorum- andScheuchzeria-peat. Combined, this would mean thatgrowth rates are highest in the intermediate habitats,which however leads to contradictions. The intermediateslopes would tend to catch up with the Calluna hum-mocks, assimilate this vegetation and stagnate in turn. Infact, the stratigraphic variation of Eng-XIV could thenlargely be described as autogenic regeneration (Barber,1981). Simultaneously, the low accumulation rates infer-red for pools should lead to their permanent stagnation.Both inferences con#ict with stratigraphic evidence,which has shown that hummocks and hollows in temper-ate raised bogs are more or less spatially stable through-out millennia (Casparie, 1972; Walker and Walker, 1961;Barber, 1981; "kland and Ohlson, 1998).

The obvious solution is "nding an explanation forthe raised pollen concentration alternative to the concen-tration of time. Pollen deposition occurs by two distinc-tive mechanisms, viz. dry and wet deposition, of varyingrelative importance (Grosse-Brauckmann and Stix,1979). Because pollen in#uxes are highest in plant

communities with higher phanerogams, we suggest thatvegetation structure plays a key role in determiningin#ux. By increasing local surface roughness, andreducing air movement, both Ericaceae and Cyperaceaemay promote the "ltering of pollen from the air. Theirenhanced exposure to rain, compared to the moss layer,may have similar e!ect on wet pollen deposition. HighAP concentrations of S. cuspidatum-pools may be ex-plained by transport from neighbouring hummocks, andperhaps pollen carried by #owing surface water.

Two independent lines of evidence strongly supportthese claims. Old"eld et al. (1979) calculated in#uxesof magnetite in hummocks and pools in some bogs inCumbria (Britain), "nding that the hummocks trap anorder of magnitude more of magnetic spherules. Old"eldet al., (1981)calculated deposition rates for magnetite andiron in four Finnish bogs, based on moss-incrementcounting. Three of these bogs show large inter-site in#uxdi!erences, while only maxima and minima are corre-lated. Belyea and Warner (1994) who tried to date peatwith 210Pb, concluded that &the hummock cores2accu-mulated a greater burden of unsupported 210Pb than didthe depressions, despite small di!erences in the atmo-spheric #ux of 210Pb to these two microhabitat types'.

The second argument are the set of observations col-lected by Joosten and van den Brink (1992) on pollenentrapment by Secale and Hordeum. The ears of Secalecontained considerably higher regional pollen concentra-tions than those of Hordeum. They attributed this to earmorphology, and di!erences between other tissues by&the surface structure of the stems and leaves. The stemsurface is rather smooth, whereas the proximal side ofleaves is very rough and hairy. The lower part of theSecale stem bears more leaves than the upper part, whichwould explain the higher pollen concentration on thelower part of the stem'. Surface morphology thereforeseems an important component determining in#uxes ofparticulate matter.

5.2. Constant decay model

The conclusions of Ingram (1982,1983) were con-tested by Clymo (1984), who claimed that anaerobicdecay in the catotelm causes progressive mass loss,resulting in a concave age}depth relation for bogs. Heproposed an equation based on constant input fromthe acrotelm and decay proportional to the amount ofmatter left:

xc"p

c(1!e~ac tc )

where xcis height, expressed as cumulative mass, p

cis the

input from the acrotelm, ac

is the constant decayfactor, and t

cis time. The equation (above) would mean

that ageing raised bogs eventually cease to accumulate,and merely pass sequestered CO

2back into the atmo-

sphere as CH4

(Clymo, 1984). This proposition was

1028 M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033

supported by calculations on conventionally dated mirepro"les.

The application of one single curve to complex miresequences however, involves the highly unlikely assump-tion that both (past) productivity and decay are constantfrom top to bottom, i.e. &vegetation is unchanged'(Clymo, 1984, p. 615). This incorrect assumption ignoresthe systematic changes in stratigraphy as much as theimpact of species on productivity and decay. Variousauthors have noted di!erent productivities for distinctspecies under identical conditions (e.g. Clymo and Redd-away (1971) later Rochefort et al. (1990)). Species alsoseem to in#uence decay rate directly (Clymo, 1984; Joh-nson et al., 1990; Johnson and Damman, 1991) andindirectly as substrate (Coulson and Butter"eld, 1978).Moreover decay is not constant but decreasing in time(see data of Clymo, 1984; Johnson and Damman, 1991)because the amount of resistant structures increases intime. In Sphagnum species in general, relatively low decayrates should be expected (van der Heijden, 1994). Never-theless, Nilsson and Bohlin (1993) found higher methaneand carbon dioxide concentrations in Sphagnum peatthan in Carex peat. This may be due to higher transportrates in the latter, however.

The stratigraphic changes are not explained by selec-tive decay changing the macrofossil composition (Clymo,1984, p. 641), but rather by environmental changesexpressed as successive species producing di!erent netaccumulation rates. Our stratigraphic results "t two dis-tinct hydroseres described by Walker (1970), which canqualitatively explain the observed changes in accumula-tion.

The basal gyttja in AgeroK ds Mosse, containing shellsand "ne sand in the bottom and algae in the top, repres-ents an open water-environment with low net organicproductivity. Though inorganic input may raise accumu-lation rates in this mire stage, they are generally low dueto the small amount of resistant tissues formed by openwater-species. The invasion of Phragmites, which mayrapidly colonize shallow lakes, re#ects the decline ofwater depth to about 1m (Walker, 1970). The higheraccumulation rate observed is attributed to a higher netproductivity of Phragmites, mainly of rhizomes and roots,and its creation of anoxic conditions (inhibiting decay) atshallower depths.

Raised bog plant communities in our data follow oneither Phragmites peat, as in AgeroK ds Mosse, or on fenpeat with Carex species, as in Varassuo. This transition isthe result of preceding peat accumulation, which reducedwater depth and isolated the surface from the lake or soilwater. &Once a fen surface is isolated from periodic su!u-sion by lake water it must become a bog or maintaina tenuous stability by patchy, cyclical, breakdown andregeneration of the fen itself ' (Walker, 1970, p. 135). Thetransition to raised bog does not automatically involveincreasing accumulation rates; in fact a decline is ob-

served in AgeroK ds Mosse and Varassuo. The Sphagnum-dominated raised bog community results from irrevers-ible changes creating a nutrient-poor rain water lens inthe central mire, and ampli"es those changes by acidify-ing its environment (Walker, 1970; Ingram, 1983). Thesespecies e!ect two qualitative hydrological changes, whichare the development of a genuine acrotelm and the asso-ciated transition from a centripetal to a centrifugal drain-age system (Ingram, 1983).

The increased capacity of Sphagnum to reproduce itsown conditions is also apparent within the raised bogdeposits. In three mires } AgeroK ds Mosse, Draved Moseand the Engbertsdijksvenen } the replacement of thesmall-leaved Sphagnum sect. Acutifolia by broad-leavedS. sect. Cymbifolia is accompanied by an increase inaccumulation rate. This represents a threshold, whichbogs may have crossed due to a climatic wetting pulse.This does however not explain the persistence of thebroad-leaved species and high accumulation rate. Wesuggest that species-related hydrological performance,resulting in more e!ective storage of precipitation, maybe the cause.

The exclusive occurrence of accumulation rate changescombined with major stratigraphical transitions, and thegenerally good matches of mire data to the calibrationcurve, leads us to conclude that stratigraphy } fossilizedhydrology } explains the roughly concave age}depthrelation of mire deposits. The productivity and decaycomponents of Clymo (1983, 1984) model have beenexpressed as a contrast between the apparent and &actual'rate of carbon accumulation. The actual rate of carbonaccumulation, which is catotelm input minus decay, isestimated to be 2/3, 70%, 2/3 and 75% respectively of theapparent "gure (i.e. input only) in subsequent studies ofFinnish mires (Tolonen et al., 1992; Korhola et al., 1995;Tolonen and Turunen, 1996; MaK kilaK , 1997). A similar"gure can be obtained directly by dividing the accumula-tion rates of broad-leaved Sphagna (S. imbricatum, S.papillosum, S. magellanicum), as the &apparent' rate, bythose of the preceding Sphagna, which would re#ect the&actual' long-term accumulation. For the Eng-I core,the quotient of accumulation rates (S. sect. Acutifolia/S.imbricatum) would be 0.48/0.77"0.62. For AgeroK dsMosse and Draved Mose the values are 0.51 and 0.46respectively.

6. Conclusions

For the construction of a time}depth relation fora sediment a number of problems should be solved,depending on the aims. To be empirically reliable withinone sequence, we should allow for dating error to trans-late the existing age variation into realistic con"denceintervals. If error sources can be identi"ed (and dealtwith), the relation becomes more accurate. If a number of

M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033 1029

time}depth relations are transformed into a model claim-ing physical explanation, the impact of the original as-sumptions should be reviewed.

Wiggle matching a series of uncalibrated 14C dates ofabove ground plant macrofossils provides the possibilityto compare the peat age #uctuations directly to thetreering curve. Relative time scales alternative to depth(cumulative pollen, mass) were not found to improve thewiggle match "t. For conventionally dated raised bogdeposits, several sources of 14C variation in sequences ofRadiocarbon dates were dealt with: errors due to botan-ically mixed species, sample error due to time width,hiatuses, and stratigraphic change.

Middeldorp's (1982) &pollen density dating' assumesa stable pollen in#ux and continuous accumulation (nomissing pollen) which is not always justi"ed. In additionit suggests stagnation of both hummocks and pools,which con#icts with stratigraphic evidence.

Clymo's (1984) method is hampered by dating errorsand the assumption of continuity as well. For the sake ofclarity, long-term decay of bog deposits is not a prioriimpossible. It seems unlikely that no methane from bogsescapes to the atmosphere at all. Methane oxidation infreshwater samples is coupled to methanogenesis, andalways results in net methane production (Harder, 1997).However, the reservoir e!ect described by us may reducethe input to the atmosphere by storing fossil carbon onericaceous roots (Kilian et al., 1995).

Unrealistic decay rates are obtained by contrastingearly and late Holocene peat accumulation. Clymo'smethod excludes the possibility of species-related in-creases and decreases of accumulation rate; mire inter-vals with a roughly convex age}depth relation can not bedescribed. Korhola et al. (1995) also observed that &inseveral peatlands, the vertical growth contradicted theconcave plot predicted of age vs. depth by Clymo's (1984)model of peat growth2'. In other cases, the large in-crease in peat accumulation resulting from the rise ofSphagnum sect. Cymbifolia is presented as its opposite,viz. a decreased accumulation. This erroneous notion hasbecome scienti"c common sense (Gorham, 1991; Joosten,1993; Ovenden, 1990; Tolonen and Turunen, 1996;"kland and Ohlson, 1998).

Whether time}depth relations of bog deposits are lin-ear, or perhaps slightly concave due to long-term decay(Clymo, 1984), is a question for further research. Ourresults suggest the former, but subtle changes in14C gradient within homogeneous stratigraphic units, ifexistent, may be demonstrated by future high-resolution14C AMS dating of selected macrofossils.

Acknowledgements

We express our gratitude to many who contributed tothis study. Ton van Druten, Hans Joosten, Pim de Klerk

and Jan-Bart Laan helped sampling the Eng-XIV peatcore with the Damocles cutting device (Laboratory forPalynology and Palaeobotany, Utrecht University). An-nemarie Philip and Elly Beglinger prepared the pollenslides. In Groningen, Anita Aerts-Bijma prepared theAMS samples, Stef Wijma operated the AMS andHarm-Jan Streurman measured the conventional14C samples. Alessandra Speranza is acknowledged forhelp with the illustrations. Mr. J. de Vries of Staatsbos-beheer Zwartemeer assisted in the Meerstalblok for col-lecting recent plants. Eva Ran is thanked for discussionsconcerning this study. The "rst author was supported bythe Netherlands Geosciences Foundation (GOA) with"nancial aid from the Netherlands Organisation forScienti"c Research (NWO).

Appendix A. Wiggle and curve match error estimate

A.1. Introduction

Various authors have derived error estimates forwiggle matched sequences. Pearson (1986) used as2-minimization technique to "t "ve 20 yr treeringsamples to a calibration curve with the same resolution.However, his method to estimate a 95%-con"denceinterval was complex. Pilcher et al. (1995) use a similartechnique, minimizing the sum of squares (SS), measuredas the squared di!erences between splined calibrationcurve ages and 14C ages of peat samples. By amethod analogous to our version in Cal20, they varyboth sedimentation rate and average age of the core byshifting, stretching and compressing the 14C dateson a linear depth scale. Their error evaluation of"ve raised bog 14C dates on humin leads to a 2310$20BC estimate for the Hekla 4 eruption, which is amuch narrower con"dence interval than previously pub-lished (Pilcher et al., 1995). But it is unclear whethersample time width is taken into account in the resultingerror.

Goslar et al. (1995) "t the Gosciaz #oating varvechronology to the early Holocene part of the calibra-tion curve, with samples of on average 37 varve years.They use a minimum sum of squares method weighingindividual 14C dates by their errors. In addition,an error term accounting for deposition time (0.5dt) isincluded in the calculation of minimum sum of squares(SS).

With both the s2- and the SS-minimization techniques,a minimum will be found for any number of (good orbad) dates, giving a best solution for a straighttime}depth curve. But there has not been an e!ort yet tocalculate realistic con"dence intervals through time (cru-cial for non-varved sediments), and no acceptance cri-teria of the goodness-of-"t were formulated independentfrom the "tting procedure.

1030 M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033

A.2. General formula for linear interpolation

By means of wiggle (or curve) matching we try to relatea relative time scale &x (depth, pollen or mass) of a sedi-ment directly to calendar time t. The age of sample iin a continuously sampled section is given by a linearregression equation of the form:

ti"¹

0#Cdx

¹.!9

!¹0

X.!9

!X0

[yr BC/AD] (A.1)

whereti

average age of sample i (yr)¹0

age of the basal 14C sample (yr)¹.!9

age of top 14C sample (yr)dx average sample size (cm, pollen or mass)dt average deposition time (y)X

0position of bottom 14C sample (cm, pollen ormass)

X.!9

position of top 14C sample (cm, pollen or mass)C constant.

If ¹0"X

0"0,

ti"Cdt"Cdx

¹.!9

X.!9

(yr BC/AD) (A.2)

This equation we used to calculate dt from the best "tfound by wiggle matching.

A.3. Linearity test

We compared the minimum variance obtained bywiggle matching all 14C samples from a series with themaximum variance expected. If the assumption of cons-tant accumulation (or input) is valid, and samples exhibitno species error, we expect all variation in sediment14C ages can be explained by natural radiocarbon vari-ations and statistical error. We expect a maximum vari-ance, for the best "t, calculated by a linear combinationof laboratory standard deviation s

BPand sample error

s4!.1-%

:

s2%91

"s2BP

#s4!.1-%

(A.3)

A direct estimate for s%91

is provided by wiggle matching.The goodness-of-"t parameter s

8.,.*/calculated with

Cal20 represents the root of the mean square s2Y >X

, esti-mating the standard error for the best "t:

s28.,.*/

"s2Y >X

"

+(BP!BP@)2N

(A.4)

in which BP is a sediment 14C date, BP@ the correspond-ing calibration curve (spline) 14C date at the best "tposition, and N the number of 14C samples.

We use s2%91

to predict s28.,.*/

, using F-test statistics.Depending upon the number of dates N, we can constuctan acceptance threshold s

#3*5for the linear time-depth

relation found by wiggle matching:

s#3*5

"J(Fs2%91

) (A.5)

F(N~1),0.05

can be directly read from an F-table. Sampletime width consists of species error and deposition time(dt). Assuming species errors absent (which is only correctfor AMS-dates), dt was estimated with Eq. (A.1). For longcores this underestimates dt with circa 10%, as the calib-ration curve has a 90% slope. To simplify calculation, forall intervals s

#3*5was calculated with:

s#3*5

"J(F(s2BP

#(0.5dt/0.9)2)) (A.6)

If s8.,.*/

exceeded s#3*5

, we concluded that either accu-mulation was not constant as considered, or that other(e.g. species) error raised s

8.,.*/. In that case we adapted

our original hypothesis by: (1) splitting up the section instratigraphically homogeneous units of progressivelysmaller size and N (This is indicated in the tables byN(0)..N(..)), or: (2) compensating for a species e!ect, untilfor each core segment s

8.,.*/(s

#3*5, or N(4.

A.4. Error estimate

Though the calibration curve is an non linear function,the assumption of constant sediment accumulationdirectly produces a linear time}depth regression. If thenonlinearity of the calibration curve BP age is matchedby an analogous nonlinearity of the sediment samples BPage, we may use wiggle matching between X

0and

X.!9

as a linear age interpolation method. This ageinterpolation is used to calculate con"dence intervals forages from depths.

A careful age prediction interval (95%-limits) for tiis

ti"t

i,8.,.*/$2J(s2

BP#(0.5dt)2) (A.7)

in which ti

and sample resolution dt [yr/sample] areobtained from the best "t. If, however, the lineartime}depth relation is used to predict ages within thedated interval, the combined use of many 14C datesbecomes a decisive advantage:

ti"t

i,8.,.*/$2SAs2Y >XC

1

N#

(xi!X

!7')2

&x2 DB. (A.8)

In which xiis the depth of the 14C sample for which the

interval is calculated, X!7'

average 14C sample depth,and Rx2 the sum of squared sample depths. Throughouta core/segment, this error estimate procedure createsnarrow con"dence intervals slightly widening towardsthe ends (Sokal and Rohlf, 1981; see also Boreux et al.,1997. Fig. 4). Provided the dates are more or less evenlydistributed throughout a core, this yields a minimumerror for the center of:

ti"t

i,8.,.*/$2SA

(s2Y >X

)

N B (A.9)

M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033 1031

However, this approach only produces accurate results ifsampling error is smaller than the calibration curve res-olution. Especially if 0.5dt'20 y, a modi"ed form of Eq.(A.8) is proposed:

ti"t

i,8.,.*/$2SAs2Y >XC

1

N#

(xi!X

!7')2

+x2 D#(0.5dt)2B(A.10)

Throughout a core/segment, this formula could be usedfor estimating the con"dence interval of the regression.Assuming s

8.,.*/equals s

Y >X, and adding (0.5dt)2 to

account for average uncertainty in real sample mean age,con"dence intervals can now be calculated from our esti-mate of s

8.,.*/, dt and the depth level of every 14C date.

Comparison with cumulative mass or pollen (&m, &p)wiggle/curve match procedure was done by substitutingdepths by sample &m/&p-values. In analogy to (0.5dt) asdepth error estimate, we used 0.5m

ior 0.5p

iwhere m

iand

pi

are the mass and pollen concentration (g/cm3, pol-len/cm3) of the ith sample.

References

Aaby, B., Tauber, H., 1975. Rates of peat formation in relation to degreeof humi"cation and local environment, as shown by studies ofa raised bog in Denmark. Boreas 4, 1}17.

Barber, K.E., 1981. Peat Stratigraphy and Climatic Change. Balkema,Rotterdam, 219 pp.

Belyea, L.R., Warner, B.G., 1994. Dating of the near-surface layer of apeatland in northwestern Ontario, Canada. Boreas 23, 259}269.

Bennett, K.D., 1995. Con"dence intervals for age estimates and depos-ition times in late-Quaternary sediment sequences. The Holocene 4,337}348.

Boreux, J.-J., Pesti, G., Duckstein, L., Nicolas, J., 1997. Age modelestimation in paleoclimatic research: fuzzy regression and radiocar-bon uncertainties. Palaeogeography, Palaeoclimatology, Palaeoecol-ogy 128, 29}37.

Casparie, W.A., 1972. Bog development in Southeastern Drenthe, The-sis, University of Groningen, The Netherlands, Junk, 271 pp.

Coulson, J.C., Butter"eld, J., 1978. An investigation of the biotic factorsdetermining the rates of plant decomposition on blanket bog. Jour-nal of Ecology 66, 631}650.

Clymo, R.S., 1978. A model of peat bog growth. In: Heal, O.W., Perkins,D.F. (Eds.), Production Ecology of British Moors and MontaneGrasslands; Ecological Studies, vol. 27 (pp. 187}223).

Clymo, R.S., 1983. Peat. In: Gore, A.J.P. (Ed.), Ecosystems of the world4A. Mires: Swamp, Bog, Fen and Moor. Elsevier, Amsterdam,pp. 159}224.

Clymo, R.S., 1984. The limits to peat bog growth. Proceedings of theRoyal Society of London B 303, 605}654.

Clymo, R.S., Reddaway, E.J.F., 1971. Productivity of Sphagnum (bog-moss) and peat accumulation. Hidrobiologia (Bucharest) 12,181}192.

Dickinson, W., 1975. Recurrence surfaces in Rusland Moss, Cumbria(formerly North Lancashire). Journal of Ecology 63, 913}935.

Donner, J.J., Alhonen, P., Eronen, M., Jungner, H., Vuorela, I., 1978.Biostratigraphy and radiocarbon dating of the Holocene lakesediments of TyoK tjaK rvi and the peats in the adjoining bog Varassuowest of Lahti in southern Finland. Annales Botanici Fennici 15,258}280.

Dupont, L.M., 1985. Temperature and rainfall variation in a raised bogecosystem: a palaeoecological and isotope-geological study. Thesis,Universiteit van Amsterdam, 62 pp.

Dupont, L.M., Brenninkmeijer, C.A.M., 1984. Palaeobotanic and iso-topic analyses of late Subboreal and early Subatlantic peat fromEngbertsdijksveen VII. The Netherlands. Review of Palaeobotanyand Palynology 41, 241}271.

Ferguson, C.W., Huber, B., Suess, H.E., 1966. Determination of the ageof Swiss lake dwellings as an example of dendrochronologicallycalibrated radiocarbon dating. Zeitschrift fuK r Naturforschung (SerieA) 21, 1173}1177.

Galvin, L.F., 1976. Physical properties of Irish peats. Irish Journal ofagricultural Research 15, 207}221.

Gorham, E., 1991. Northern peatlands: role in the carbon cycle andprobable responses to climatic warming. Ecological Applications 1,182}195.

Goslar, T., Arnold, M., Bard, E., Kuc, T., Pazdur, M., Ralska-Jawic-zowna, M., RoH zaH nski, K., Tisnerat, N., Walanus, A., Wicik, B.,Wieckowski, K., 1995. High concentration of atmospheric 14C dur-ing the Younger Dryas cold episode. Nature 377, 414}417.

Gottdang, A., Mous, D.J.W., van der Plicht, J., 1995. The HVEE14C system at Groningen. In: Cook, G.T., Harkness, D.D., Miller,B.F., Scott, E.M. (Eds.) Proceedings of the 15th InternationalRadiocarbon Conference; Radiocarbon 37, 649-656.

Grosse-Brauckmann, G., Stix, E., 1979. Beziehungen zwischen Pollen-konzentratio-nen in der Luft und Pollenniederschlagswerten. Flora168, 53}84.

Harder, J., 1997. Anaerobic methane oxidation by bacteria employing14C-methane uncontaminated with 14C-carbon monoxide. MarineGeology 137, 13}23.

Ingram, H.A.P., 1978. Soil layers in mires: Function and terminology.Journal of Soil Science 29, 224}227.

Ingram, H.A.P., 1982. Size and shape in raised mire ecosystems: A geo-physical model. Nature 297, 300}303.

Ingram, H.A.P., 1983. Hydrology. In: Gore, A.J.P. (Ed.), Ecosystems ofthe world 4A. Mires: Swamp, Bog, Fen and Moor. Elsevier, Amster-dam, pp. 67}158.

Johnson, L.C., Damman, A.W.H., 1991. Species-controlled Sphagnumdecay on a South Swedish raised bog. Oikos 61, 234}242.

Johnson, L.C., Damman, A.W.H., Malmer, N., 1990. Sphagnum macro-structure as an indicator of decay and compaction in peat coresfrom an ombrotrophic South Swedish peat-bog. Journal of Ecology78, 633}647.

Joosten, J.H.J., 1993. Denken wie ein Hochmoor: Hydrologische Sel-bstregulation von Hochmooren und deren bedeutung fuK r Wieder-vernaK ssung und Restauration. Telma 23, 95}115.

Joosten, J.H.J., van den Brink, L.M., 1992. Some notes on pollenentrapment by rye (Secale cereale L.). Review of Palaeobotany andPalynology 73, 145}151.

Kerley, S.J., Read, D.J., 1995. The biology of mycorrhiza in theEricaceae. XVIII. Chitin degradation by Hymenoscyphus ericae andtransfer of chitin-nitrogen to the host plant. New Phytologist 131,369}375.

Kilian, M.R., van der Plicht, J., van Geel, B., 1995. Dating raised bogs:New aspects of 14C AMS wiggle matching, a reservoir e!ect andclimatic change. Quaternary Science Reviews 14, 959}966.

Korhola, A., Tolonen, K, Turunen, J., Jungner, H., 1995. Estimatinglong-term accumulation rates in Boreal peatlands by radiocarbondating. In: Cook, G. T., Harkness, D. D., Miller, B. F., Scott, E. M.(Eds.), Proceedings of the 15th International 14C Conference;Radiocarbon, 37, 575}584.

MaK kilaK , M., 1997. Holocene lateral expansion, peat growth and carbonaccumulation on Haukkansuo, a raised bog in southeastern Fin-land. Boreas 26, 1}14.

Marcenco, E., Srdoc, D., Golubi'c, S., Pezdic, J., Head, M.J., 1989.Carbon uptake in aquatic plants deduced from their natural13C and 13C content. Radiocarbon 31, 785}794.

1032 M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033

Meijer, H.A.J., van der Plicht, J., Gislefoss, J.S., Nydal, R., 1995. Com-paring long-term atmospheric 14C and 3H records near Groningen,The Netherlands, with Fruholmen. Norway, and Izana, CanaryIslands. Radiocarbon 37, 39}50.

Middeldorp, A.A., 1982. Pollen concentration as a basis for indirectdating and quantifying net organic and fungal production in a peatbog ecosystem. Review of Palaeobotany and Palynology 37, 225}282.

Middeldorp, A.A., 1984. Functional palaeoecology of raised bogs: Ananalysis by means of pollen density dating, in connection with theregional forest history. Thesis, Universiteit van Amsterdam, 124 pp.

Mook, W.G., 1983. 14C calibration curves depending on sample time-width. PACT 8, 517}525.

Neustupny, E., 1973. Absolute chronology of the Aeneolithic period.Actes du VIIIe Congres International de Sciences Prehistoriqueset Protohistoriques, Beograd, 9}15 september 1971, pp. 243}248.

Nilsson, M., Bohlin, E., 1993. Methane and carbon dioxide concen-trations in bogs and fens - with special reference to the e!ects of thebotanical composition of the peat. Journal of Ecology 81, 615}625.

Nilsson, T., 1964. Standardpollendiagramme und C14-Datierungen ausdem AgeroK ds Mosse im Mittleren Schonen. Lunds UniversitetsAs rsskrift N.F. Avd. 2, 59-7, 1-52.

"kland, R.H., Ohlson, M., 1998. Age-depth relationships in Scandina-vian surface peat: A quantitative analysis. Oikos 82, 29}36.

Old"eld, F., Brown, A., Thompson, R., 1979. The e!ect of microtopo-graphy and vegetation on the catchment of airborne particles mea-sured by remanent magnetism. Quaternary Research 12, 326}332.

Old"eld, F., Tolonen, K., Thompson, R., 1981. History of particulateatmospheric pollution from magnetic measurements in dated Finn-ish peat pro"les. Ambio 10, 185}188.

Olsson, I.U., 1983. Dating non-terrestrial materials. PACT 8, 277}294.Olsson, I.U., 1986. A study of errors in 14C dates of peat and sediment.

Radiocarbon 28, 429}435.Olsson, I.U., 1991. Accuracy and precision in sediment chronology.

Hydrobiologia 214, 25}34.Ovenden, L., 1990. Peat accumulation in northern wetlands. Quater-

nary Research 33, 377}386.Pearson, G.W., 1986. Precise calendrical dating of known growth-

period samples using a &curve "tting' technique. Radiocarbon 28,292}299.

Pilcher, J.R., Hall, V.A., McCormac, F.G., 1995. Dates of HoloceneIcelandic volcanic eruptions from tephra layers in Irish peats. TheHolocene 5, 103}110.

Proctor, M.C.F., Raven, J.A., Rice, S.K., 1992. Stable carbon isotopediscrimination measurements in Sphagnum and other bryophytes:Physiological and ecological implications. Journal for Bryology 17,193}202.

Punning, J.-M., Ilomets, M., Ko!, T., 1993. Possibilities for detaileddating of peat bog deposits. Radiocarbon 35, 379}385.

Rochefort, L., Vitt, D.H., Bayley, S.E., 1990. Growth, production, anddecomposition dynamics of Sphagnum under natural and experi-mentally acidi"ed conditions. Ecology 71, 1986}2000.

Shore, J.S., Bartley, D.D., Harkness, D.D., 1995. Problems encounteredwith the 14C dating of peat. Quaternary Science Reviews (Quater-nary Geochronology) 14, 373}383.

Sokal, R. R., Rohlf, F. J., 1995. Biometry, 3rd Edition. Freeman andCompany, New York, 859pp.

Stockmarr, J., 1971. Tablets with spores used in absolute pollen analy-sis. Pollen et Spores 13, 615}621.

Stuiver, M., Kromer, B., Becker, B., Ferguson, C.W., 1986. Radiocarbonage calibration back to 13,300 years BP and the 14C age matching ofthe German Oak and US Bristlecone Pine chronologies. Radiocar-bon 28, 969}971.

Stuiver, M., Pollach, H.A., 1977. Discussions of reporting 14C.Radiocarbon 19, 355}363.

Stuiver, M., Reimer, P.J., 1993. Extended 14C data base and revisedCALIB 3.0 14C age calibration program. Radiocarbon 35,215}230.

Tolonen, K., Vasander, H., Damman, A.W.H., Clymo, R.S., 1992. Rateof apparent and true carbon accumulation in Boreal peatlands.Proceedings of the 9th international peat congress, 22}26 June,Upssala, pp. 319}333.

Tolonen, K., Turunen, J., 1996. Accumulation rates of carbon in miresin Finland and implications for climate change. The Holocene 6,171}178.

ToK rnqvist, T.T., Bierkens, M.F.P., 1994. How smooth should curves befor calibration of radiocarbon ages?. Radiocarbon 36, 11}26.

van der Heijden, E., 1994. A combined anatomical and pyrolysis massspectrometric study of peati"ed plant tissues. Thesis, Universiteitvan Amsterdam, 127 pp.

van der Molen, P.C.M., 1992. Hummock-hollow complexes on Irishraised bogs: A palaeo/actuoecological approach of environ-mental and climatic change. Thesis, Universiteit van Amsterdam,213 pp.

van der Molen, P.C.M., Hoekstra, S.P., 1988. A palaeoecologicalstudy of a hummock-hollow complex from Engbertsdijksveen, inthe Netherlands. Review of Palaeobotany and Palynology 56,213}274.

van der Plicht, J., 1993. The Groningen radiocarbon calibration pro-gram. Radiocarbon 35, 231}237.

van der Plicht, J., Jansma, E., Kars, H., 1995. The Amsterdam castle.A case study of wiggle matching and the proper calibration curve.Radiocarbon 37, 965}968.

van Geel, B., 1978. A palaeoecological study of Holocene peat bogsections in Germany and the Netherlands. Review of Palaeobotanyand Palynology 25, 1}120.

van Geel, B., Dallmeijer, A.A., 1986. Eine Molinia-Tor#age als E!ekteines Moorbrandes aus dem fruK hen Subboreal im Hochmoor En-gbertsdijksveen (Niederlande). Abhandlungen Landesmuseum fuK rNaturkunde 48, 471}479.

van Geel, B., Mook, W.G., 1989. High-resolution 14C dating of organicdeposits using natural atmospheric 14C variations. Radiocarbon,31, 151}155.

Walker, D., 1970. Direction and rate in some British post-glacial hydroseres. In: Walker, D., West, G. (Eds.), Studies in theVegetational History of the British Isles. Cambridge UniversityPress, Cambridge, pp. 117}139.

Walker, D., Walker, P.M., 1961. Stratigraphic evidence of regenerationin some Irish bogs. Journal of Ecology 49, 169}185.

M.R. Kilian et al. / Quaternary Science Reviews 19 (2000) 1011}1033 1033