the genesis of yedoma ice complex permafrost – grain-size

EampG Quaternary Sci J 69 33ndash53 2020httpsdoiorg105194egqsj-69-33-2020copy Author(s) 2020 This work is distributed underthe Creative Commons Attribution 40 License

Research

article

The genesis of Yedoma Ice Complex permafrost ndash grain-sizeendmember modeling analysis from Siberia and AlaskaLutz Schirrmeister1 Elisabeth Dietze12 Heidrun Matthes1 Guido Grosse13 Jens Strauss1 Sebastian Laboor1Mathias Ulrich4 Frank Kienast5 and Sebastian Wetterich1

1Alfred Wegener Institute Helmholtz Centre for Polar and Marine Research Potsdam Germany2Section 32 Organic Geochemistry German Research Centre for Geosciences GFZHelmholtz Centre Potsdam Potsdam Germany3Institute of Geosciences University of Potsdam Potsdam Germany4Institute for Geography Leipzig University Leipzig Germany5Research Station of Quaternary Palaeontology Senckenberg Research Institute Weimar Germany

Correspondence Lutz Schirrmeister (lutzschirrmeisterawide)

Relevant dates Received 3 December 2019 ndash Revised 6 March 2020 ndash Accepted 23 March 2020 ndashPublished 25 May 2020

How to cite Schirrmeister L Dietze E Matthes H Grosse G Strauss J Laboor S Ulrich M Kienast Fand Wetterich S The genesis of Yedoma Ice Complex permafrost ndash grain-size endmember modelinganalysis from Siberia and Alaska EampG Quaternary Sci J 69 33ndash53 httpsdoiorg105194egqsj-69-33-2020 2020

Abstract The late Pleistocene Yedoma Ice Complex is an ice-rich and organic-bearing type of permafrost de-posit widely distributed across Beringia and is assumed to be especially prone to deep degradationwith warming temperature which is a potential tipping point of the climate system To better under-stand Yedoma formation its local characteristics and its regional sedimentological composition wecompiled the grain-size distributions (GSDs) of 771 samples from 23 Yedoma locations across theArctic samples from sites located close together were pooled to form 17 study sites In addition westudied 160 samples from three non-Yedoma ice-wedge polygon and floodplain sites for the compar-ison of Yedoma samples with Holocene depositional environments The multimodal GSDs indicatethat a variety of sediment production transport and depositional processes were involved in Yedomaformation To disentangle these processes a robust endmember modeling analysis (rEMMA) was per-formed Nine robust grain-size endmembers (rEMs) characterize Yedoma deposits across BeringiaThe study sites of Yedoma deposits were finally classified using cluster analysis The resulting fourclusters consisted of two to five sites that are distributed randomly across northeastern Siberia andAlaska suggesting that the differences are associated with rather local conditions In contrast to priorstudies suggesting a largely aeolian contribution to Yedoma sedimentation the wide range of rEMsindicates that aeolian sedimentation processes cannot explain the entire variability found in GSDsof Yedoma deposits Instead Yedoma sedimentation is controlled by local conditions such as sourcerocks and weathering processes nearby paleotopography and diverse sediment transport processesOur findings support the hypothesis of a polygenetic Yedoma origin involving alluvial fluvial andniveo-aeolian transport accumulation in ponding waters and in situ frost weathering as well as post-depositional processes of solifluction cryoturbation and pedogenesis The characteristic rEM com-position of the Yedoma clusters will help to improve how grain-size-dependent parameters in per-

Published by Copernicus Publications on behalf of the Deutsche Quartaumlrvereinigung (DEUQUA) eV

34 L Schirrmeister et al The genesis of Yedoma Ice Complex permafrost

mafrost models and soil carbon budgets are considered Our results show the characteristic propertiesof ice-rich Yedoma deposits in the terrestrial Arctic Characterizing and quantifying site-specific pastdepositional processes is crucial for elucidating and understanding the trajectories of this unique kindof ice-rich permafrost in a warmer future

Kurzfassung Der spaumltpleistozaumlne Yedoma Eiskomplex ist ein eisreicher und organikhaltiger Permafrosttyp der inBeringia weit verbreitet ist Durch den hohen Eisanteil wird der Yedoma Eiskomplex im Zuge desKlimawandels als besonders anfaumlllig fuumlr tiefgreifende Stoumlrungen betrachtet und damit ein potentiellerKipppunkt des Klimasystems Um seine Entstehung die lokalen Eigenschaften und die regionale sed-imentologische Zusammensetzung besser zu verstehen haben wir die Korngroumlszligenverteilung von 771Proben an 23 Yedoma-Standorten in der Arktis zusammengestellt raumlumlich eng zusammenhaumlngendeProbenserien wurden zu 17 Untersuchungsstandorten zusammengefasst Daruumlber hinaus wurden 160Proben aus nicht Yedoma-Ablagerungen von drei Eiskeilpolygon- und Uumlberschwemmungsgebietenals holozaumlne Referenzen untersucht Die multimodalen Korngroumlszligenverteilungen zeigen dass eineVielzahl von Sedimentbildungs- Transport- und Ablagerungsprozessen an der Yedoma-Entstehungbeteiligt waren Um diese Prozesse zu erkennen wurde eine robuste Endmembermodellierungs-analyse (rEMMA) durchgefuumlhrt Neun robuste Endmember (rEM) charakterisieren die Yedoma-Ablagerungen uumlber ganz Beringia Die untersuchten Standorte der Yedoma-Ablagerungen wurdenanschlieszligend mittels Clusteranalyse klassifiziert Die daraus resultierenden vier Cluster umfassenzwei bis fuumlnf Untersuchungsstandorte die unregelmaumlszligig uumlber den Nordosten Sibiriens und Alaskaverteilt sind Die breite Palette von rEMs zeigt dass nicht allein aumlolische Sedimentationsprozessefuumlr die Variabilitaumlt in den Korngroumlszligenverteilungen von Yedoma-Ablagerungen verantwortlich sindVielmehr wird die Sedimentation der Yedoma-Ablagerungen eher durch lokale Bedingungen wieAusgangsgesteine ehemalige Topographie und multiple Transportprozesse gesteuert Das stuumltzt dieHypothese einer polygenetischen Yedoma-Entstehung die alluvialen fluvialen und nival-aumlolischenTransport und Akkumulation in polygonalen Tuumlmpeln und in-situ Frostverwitterung sowie postsedi-mentaumlre Frostverwitterung Solifluktion Kryoturbation und Pedogenese beinhaltet Die charakteristis-che rEM Zusammensetzung der Yedoma Cluster kann auch helfen korngroumlszligenspezifische Parameterbesser in der Kohlenstoffbudgetierung und Permafrostmodellierung zu beruumlcksichtigen Damit traumlgtdie Charakterisierung und Quantifizierung standortspezifischer Ablagerungsprozesse in der Vergan-genheit dazu bei die charakteristischen Eigenschaften eisreicher Yedoma-Ablagerungen in der ter-restrischen Arktis aufzuklaumlren Dies ist entscheidend fuumlr das Verstaumlndnis der Fortentwicklung diesesbesonderen Permafrosttyps in einer waumlrmeren Zukunft

1 Introduction

The formation and distribution of late Pleistocene YedomaIce Complex deposits located in western (Siberia) andeastern (Alaska and northwest Canada) Beringia are stillwidely debated (Table S1 in the Supplement) These perma-nently frozen (permafrost) deposits are of silt- and sand-richorganic-bearing sediments up to tens of meters thick inter-spersed with large syngenetic ice wedges that contain highamounts of excess ground ice making them highly sensitiveto degradation in a warming climate Syngenetic ice wedgesand segregated intrasedimental ice (ice lenses and bands)constitute the largest portion (50 ndash95 ) of this type of de-posit by volume in most Yedoma regions (Kanevskiy et al2011 2016 Strauss et al 2013 Ulrich et al 2014) next toclastic and organic components (Schirrmeister et al 2013)Thus ground ice aggradation is clearly one of the most crit-ical factors in Yedoma Ice Complex genesis (hence also the

name ldquoIce Complexrdquo) and ice forms a main component ofthe entire deposit in clear contrast to the accumulation ofsedimentary deposits in temperate regions where ice does notplay a role either in the formation of deposits or as a struc-tural and stratigraphic component

In terms of depositional and stratigraphic characteristicsvarious Yedoma types seem to exist (Kaplina 1981 Sher etal 2005 Strauss et al 2012 Murton et al 2015) acrossthe area of 14times 106 km2 where Yedoma deposits currentlyoccur (Strauss et al 2017) Because of their carbon storageand high ice content Yedoma deposits have been suggestedas a potential ldquotipping elementrdquo for future climate warming(Lenton 2012) Permafrost models require the parametriza-tion of the types of Yedoma deposits for example in terms oftheir grain-size composition to better constrain factors suchas the hydraulic conductivity and pore space volume wherewater can freeze (DallrsquoAmico et al 2011) The estimation of

EampG Quaternary Sci J 69 33ndash53 2020 httpsdoiorg105194egqsj-69-33-2020

L Schirrmeister et al The genesis of Yedoma Ice Complex permafrost 35

carbon storage potential also seems to be linked with grain-size composition (Palmtag and Kuhry 2018) making a bettergranulometric characterization of Yedoma types useful forcarbon budget studies

Most of the studies on Yedoma formation agree that itwas dominated by the growth of syngenetic ice wedgesin polygonal tundra landscapes during the late Pleistocene(Schirrmeister et al 2013) The ice wedges formed in low-center polygon nets during the interstadial Marine IsotopeStage 3 (MIS 3) and the stadial MIS 2 promoted by long-lasting continental cold climate conditions with short thawphases during late Pleistocene summers (for references seeTable S1) The widespread formation of ice-wedge polygonsin much of Beringia was closely related to the persistence ofstable poorly drained accumulation areas with a low topo-graphic gradient (Schirrmeister et al 2013)

More debated is the origin of allochthonous clasticYedoma components Different hypotheses have been sug-gested pointing especially to the role of aeolian processesduring Yedoma formation Studies in Yukon and Alaskainterpret Yedoma as loess or retransported loess (ldquomuckrdquoPeacuteweacute 1955 1975 Muhs et al 2008) A range of other hy-potheses have emerged to explain the late Pleistocene depo-sition processes in the Siberian Yedoma region interpretingthe clastic Yedoma deposits as being derived from multi-ple rather local sediment sources and transport pathwaysas well as from secondary sediment deformation and cryo-genic reworking (Schirrmeister et al 2011 2013 Siegert etal 2002) and Cryosol formation (Orthels Turbels or His-tels Walter Anthony et al 2014 Table S1) with an aeolianfraction as one of many components of the sediment mate-rial

Grain-size distributions (GSDs) are known to provide es-sential information about source-to-sink relations transportmodes sorting and depositional processes (Folk and Ward1957 Visher 1969 Sun et al 2002 Bartholdy et al 2007Weltje and Prins 2007 Dietze et al 2014 Ulrich et al2019) Here we analyzed (i) to what extent regional- tocontinental-scale aeolian processes contributed to Yedomagenesis and (ii) what the role of local sedimentation pro-cesses was Assuming rather similar environmental (ie landcover) and climatic conditions across the Yedoma region atthe time of Yedoma formation during the late Pleistocene wesuggest that Yedoma types and varieties originated in differ-ent bedrock and paleotopographic configurations (Table S2)We test this hypothesis by analyzing the GSDs of more than700 Yedoma samples from across the Arctic in order to iden-tify sedimentological endmembers (EMs) that can be associ-ated with certain depositional regimes The development ofsite-specific and region-wide interpretations of Yedoma de-positional processes helps to elucidate the typical composi-tion formation and transformation conditions of these de-posits which are an important indicator for the late Pleis-tocene paleoenvironment in Beringia

2 Material and methods

21 Study region

The nonglaciated lowlands and formerly exposed shelf ar-eas between the Eurasian and Laurentide ice sheets formeda land bridge commonly named Beringia between Eurasiaand northern America during the late Pleistocene (Hulteacuten1937) Study sites in eastern Beringia are situated on theAlaska North Slope with exposures along the Itkillik andColville rivers on the northern part of the Seward Peninsulaand in the Vault Creek (VC) tunnel near Fairbanks in Inte-rior Alaska (Fig 1a Table S2) In western Beringia Yedomaexposures and drill cores from numerous coastal and deltasites in the Laptev and East Siberian seas region were stud-ied between 1998 and 2014 mainly along the Laptev Seaand New Siberian Islands coasts (Fig 1b c Table S2) Inaddition Yedoma sites were studied in the Yakutian inlandat the key site Duvanny Yar in the Kolyma lowlands at theKytalyk site in the Yana-Indigirka Lowland in the BatagayMega-slump in the Yana Highlands and in Central Yakutia(Table 2) All Yedoma deposits that contributed samples tothis study were formed during the late Pleistocene MIS 3and MIS 2 periods (Table S2) Sediments from Holocene ice-wedge polygons from two study areas in Yakutia (KytalykYana-Indigirka lowland and Pokhodsk Kolyma Lowland)were used for comparison as we consider those low-centerpolygons as final formation areas of the Yedoma Ice Com-plex (Tables 1 S2)

22 Analytical methods

We compiled 671 samples from previous studies and 100unpublished samples from 23 individual Yedoma locationsin Alaska and Yakutia and added 103 samples of 13 drillcores and 57 modern surface sediments from non-Yedomaice-wedge polygons from five Yakutian sites as references(see Table S2) The modern polygon pond substrate was col-lected from the uppermost 5 cm at the substratendashwater inter-face For all Yedoma sites all available samples from eachsite were used for the following analysis

Sampling and grain-size analysis followed a similar proto-col for all samples (see references in Table S2) Frozen sed-iment samples were taken by hammer and hatchet from out-crops on seashores riverbanks and thaw slumps during ex-peditions Numerous vertically overlapping subprofiles 1 to5 m in height were sampled and merged into a combined pro-file of the site-specific stratigraphy The correlation of sam-pling positions was carried out by comparing height mea-surements using measuring tape or laser theodolite At VCAlaska (no 4 in Fig 1a) samples were taken in a tunnel withthe same approach In addition permafrost drill cores fromBolrsquoshoy Lyakhovsky Island (no 17) Buor Khaya Peninsula(no 11) and Yukechi (no 23 Table 1) were taken and latersubsampled at a 20 to 30 cm resolution in a cold laboratoryin Germany All samples were freeze-dried in the laboratory

httpsdoiorg105194egqsj-69-33-2020 EampG Quaternary Sci J 69 33ndash53 2020


Figure 1 (a) Study region showing the distribution of investigated Yedoma sites Study locations are grouped into three major Yedomaregions Alaska (diamonds) Laptev and East Siberian sea coasts including the Lena Delta (circles) and the Yakutian inland (triangles)Investigated non-Yedoma sites are also shown (squares) The background map indicates the outline of maximum Last Glacial Maximumareas according to Ehlers et al (2011) and the subaerially exposed Arctic shelf areas (Beringia) based on a minus125 m sea-level lowstand usingthe bathymetric data from ETOPO2 (2006) Location numbers and additional characteristics are explained in Table S2 Exemplary Yedomastudy sites shown are (b) Mamontov Klyk on the western Laptev Sea coast (no 5 in Fig 1a) and (c) Kurungnakh Sise Island in the LenaDelta (no 8 in Fig 1a)

manually homogenized without destroying the particles andsplit into subsamples for the various analyses

For grain-size analysis 5ndash10 g of a sample was treatedthree times a week over several weeks with 100 mL of 3 H2O2 in a horizontal shaker to remove organic matter Thesuspension was tested and the pH value adjusted to 6ndash8 At

the end of sample preparation the samples were centrifugedand dried Of the dry organic-free sediment samples 1 g wasthen dispersed in 1 L of 001 normal NH4OH and shaken forabout 24 h in an overhead shaker After that the sample wassplit into subsamples to obtain a solid content of 8 ndash12 (sufficient transparency for a laser beam) The subsamples



Table 1 Grain-size minimum and maximum arithmetic grain-size mean sorting and sample numbers of the collections from the combined17 Yedoma study sites and the non-Yedoma reference sites

Loc no Location Grain-size Grain-size Sorting No ofminndashmax meanplusmnSD minndashmax samples

(microm) (microm) (ϕ)

Alaska

1 Colville River 200ndash1618 367plusmn 241 18ndash25 342 Itkillik River 213ndash761 418plusmn 142 18ndash23 453 Seward Peninsula (Kitluk River) 263ndash637 356plusmn 96 19ndash24 244 Vault Creek tunnel 292ndash1082 480plusmn 200 14ndash19 24

Western Laptev Sea

5 Cape Mamontov Klyk 138ndash688 413plusmn 132 12ndash24 34

Lena Delta

6 7 Ebe Basyn Sise and Khardang Sise is-lands

26ndash3862 1547plusmn 723 10ndash27 47

8 Kurungnakh Sise Island 357ndash4673 1021plusmn 702 18ndash29 48

Central and eastern Laptev Sea

9 Bykovsky Peninsula 158ndash1614 654plusmn 349 18ndash31 3610 Muostakh Island 333ndash5909 1903plusmn 1299 18ndash35 2911 Buor Khaya Peninsula 208ndash909 471plusmn 193 18ndash24 90

New Siberian Islands and the Dmitry Laptev Strait

12 13 14 15 16 Stolbovoy Belrsquokovskiy northernKotelny southwestern Kotelny andMaly Lyakhovsky islands


17 Bolrsquoshoy Lyakhovsky Island 79ndash2913 446plusmn 338 16ndash27 10518 Oyogos Yar coast 157ndash527 285plusmn 84 16ndash21 44

Yakutian inland

19 Duvanny Yar 196ndash483 305plusmn 48 19ndash23 9420 Kytalyk 197ndash418 302plusmn 70 19ndash23 1821 Batagay Mega-slump 650ndash1262 875plusmn 127 14ndash24 3822 23 Tabaga and Yukechi 311ndash864 471plusmn 133 20ndash24 42

Non-Yedoma (as reference)

Pokhodsk polygon cores 211ndash1239 665plusmn 256 20ndash29 47Pokhodsk polygon bottom 256ndash3473 1054plusmn 791 20ndash31 31Kytalyk polygon cores 242ndash1378 469plusmn 235 18ndash26 28Kytalyk polygon bottom 155ndash2642 997plusmn 731 19ndash28 26Kolyma and Berelekh floodplains 201ndash1192 383plusmn 192 20ndash25 27

were sieved through a 1 mm sieve to avoid the destructionof the diffraction sample cell by larger particles There ismostly no or sometimes a little content (lt 1 ) of sieve re-mains larger than 1 mm It should be noted that due to themethodical assumption of spherical grains in laser grain-sizeanalysis the final grain size can sometimes be slightly abovethe upper limit of 1 mm Hence we here consider the GSDup to 2000 microm Finally the subsamples were measured in alaser diffraction particle analyzer (Beckman Coulter LS 200)with 92 channel sizes between 0375 and 2000 microm using

the Fraunhofer optical model preprogrammed in the LS 200analyzer Three or more subsamples of each main samplewere analyzed and their combined GSD was calculated withthe analytical software of the laser diffraction particle ana-lyzer Grain-size parameters such as sandndashsiltndashclay distribu-tion arithmetic mean in micrometers (microm) and sorting in phi(ϕ) were calculated using GRADISTAT 80 (Blott and Pye2001)

To have a sufficient number of individual samples for fur-ther analysis (here n gt 15) we combined the 23 studied



Figure 2 Example of grain-size analysis and the different steps of rEM modeling analysis (R package rEMMAgeo Dietze and Dietze2019) (a) All grain-distributions of a site (b) identification of rEM from all similarly likely endmembers with a gt 50 explained variance(c) mean and 1 standard deviation of rEM and (d) mean robust scores of respective rEM (more details are explained in the text)

Yedoma sites into 17 regional groups by merging sites lo-cated close together (Table 1) To distinguish characteristicgrain-size subpopulations from specific regions and to disen-tangle formation and transformation processes we unmixedthe polymodal GSDs of each site and the overall record us-ing a robust endmember modeling analysis (rEMMA) run inthe open-source R package EMMAgeo following Dietze etal (2012 2014) and Dietze and Dietze (2019) A type ofeigenspace analysis rEMMA is similar to principal compo-nent analysis but with the capacity to transform the endmem-ber (EM) components so that the loadings can be interpretedas GSDs (see details in Dietze et al 2012) The scores pro-vide a quantitative estimate of how much an EM contributesto a sample To obtain a robust estimate of EMs from a mea-sured GSD several EM models were analyzed with the fol-lowing steps (see background in Dietze et al 2012)

1 The ranges of a weight transformation parameter andlikely numbers of possible EMs were identified from themeasured data set (Fig 2a)

2 Robust EMs (rEMs) were defined as grain-size subpop-ulations that appear independent of model parametersFor a number of parameter sets rEMMA was then per-formed All model solutions with an overall explainedvariance of gt 50 were used to determine the rEMswhich consistently appeared among all chosen solu-tions (with similar main modes and shape identified inFig 2b) An average over all similar EMs was calcu-lated to describe the rEM

3 An uncertainty estimate for the loadings (contributionof grain-size classes to each rEM) of each mean rEMwas calculated from the spread of the modeled rEMloadings (Fig 2c)

4 Mean scores (Fig 2d) were calculated for the mean rEMloadings and a weight transformation limit that opti-mized the explained variance in the data set The un-certainty estimate for the scores was calculated via aMonte Carlo simulation (see Dietze and Dietze 2019)



From the rEM loadings and scores variances explainedby sample and class were calculated

To group the study sites further based on the rEM analy-sis the rEMs along with their explained variances wereused for a hierarchical cluster analysis (Anderberg 1973)The explained variances in the primary modes of all rEMswere summed within nine grain-size classes that containthe most common rEM modes (see rEMMA results below)into coarse sand (gt 750 microm) medium sand (281ndash750 microm)fine sand (101ndash280 microm) coarse silt (51ndash100 and 28ndash50 microm)medium silt (12ndash27 microm) fine silt (8ndash119 microm) very fine silt(4ndash79 microm) and clay (lt 4 microm) As the explained variancesrepresent the fraction of a certain rEM for a specific site achi-square distance measure was used distances dij betweentwo sites i and j were determined using the following for-mula

dij =

sum7k=1

[exvar (EMk (i))minus exvar (EMk (j ))

]2exvar (EMk (i))+ exvar (EMk (j ))

where exvar (EMk (i)) is the explained variance in the kthrEM of site i The clustering method used was ldquocompleterdquomeaning after the creation of a new cluster distances to theremaining clusters were calculated using the larger of bothoriginal distances The bootstrapping approach from the Rpackage pvclust (Suzuki and Shimodaira 2006) was usedto assess the significance of possible clusters Using thismethod supplied probabilities for each edge in the clusterdendrogram allowing the choice of statistically certain clus-ters The package supplies a basic bootstrapping probabilitysignificance value and a corrected approximately unbiasedsignificance value which we used to assess the statistical sig-nificance of the cluster edges and to determine the significantclusters

In addition to cluster analyses we combined the stud-ied Yedoma sites into three spatially explicit regions forArctic-wide comparisons (1) Alaska (2) the Laptev and EastSiberian sea coasts including the Lena Delta and (3) theYakutian inland

3 Results

31 Results of classical grain-size analyses

GSD curves of Yedoma sites reflect strong regional hetero-geneity The sandndashsiltndashclay diagram of all studied samples(Fig 3) shows the various compositions of the Yedoma se-quences The studied Yedoma deposits consisted mostly ofpoorly to very poorly sorted material with maxima in the siltand fine sand fractions (Table 1) with a certain proportion ofthe clay fraction In addition coarse sand and gravels werealso observed in the field Most sites are silt-dominated butsites in the Lena Delta and on the Laptev Sea coast are sand-dominated

The Colville site (Fig S31a in the Supplement) on theAlaska North Slope is characterized by uni- bi- tri- and

polymodal distributions of poorly to very poorly sorted mudto medium sandy silt (Fig S41) The Itkillik site exhibits bi-tri- and polymodal distributions of coarse silt to fine sandycoarse silt The VC tunnel Yedoma site near Fairbanks showsmostly unimodal and sometimes bi- and trimodal distribu-tions of poorly sorted coarse silt to very fine sandy coarsesilt The Kitluk site on the Seward Peninsula contains sortedto very poorly sorted very fine sandy coarse silt that is uni-bi- tri- and polymodally distributed

The Yedoma sites on the coasts of the western and cen-tral Laptev Sea as well as in the Lena Delta (Fig S32) allshow a wide range of GSDs including uni- bi- tri- andpolymodal curves (see examples in Fig 4a) The sorting andthe prevalent particle sizes however differ from site to siteThe Mamontov Klyk site on the western Laptev Sea coastis characterized by poorly to very poorly sorted medium siltto very fine sandy coarse silt The Ebe Sise (Nagym) andKhardang Sise Lena Delta sites are composed of moderatelyto very poorly sorted clay to fine sand The KurungnakhLena Delta site is characterized by poorly to very poorlysorted fine sandy coarse silt to coarse silty medium sandThe Bykovsky site includes sorted coarse silt to coarse siltyfine sand The adjacent Muostakh site has the widest range inGSD of poorly to very poorly sorted fine sandy coarse silt tofine silty coarse sand The Buor Khaya site contains poorlyto very poorly sorted coarse silt to fine sandy coarse silt

The Yedoma sites of the New Siberian Islands are uni-to bimodally distributed poorly sorted fine silt to fine sandycoarse silt (Fig S43) The largest data set from BolrsquoshoyLyakhovsky Island is characterized by uni- bi- tri- andpolymodal distributions and poorly to very poorly sorted finesilt to coarse silty coarse sand The Oyogos Yar site locatedon the opposite side of the Dmitry Laptev Strait is composedof uni- bi- tri- and polymodal poorly to very poorly sortedcoarse silt to fine sandy coarse silt

32 Results of site-specific endmember modelinganalyses

The rEMMA method was applied to data sets of each siteseparately Main modes of rEMs their explained variancesand the total grain-size variability explained by the averagerobust model for each study site are presented in Table 2 andFigs 4 and S41ndashS45 Each data set can be described bydifferent numbers of rEMs with modes in various grain-sizefractions between clay fine silt and coarse sand fractions(rEM 9= 10 microm from Oyogos Yar to rEM 1= 8639 micromfrom Muostakh Island Fig 4a) Across all sites the meanof the robust models explains between 54 and 855 ofthe total grain-size variability in the 17 studied data sets(Table 2) The non-Yedoma ice-wedge polygons have sim-ilar distributions from the clay fraction (rEM 9= 32 microm)to the coarse sand fraction (rEM 1= 7169 microm Table 2bFigs S45 S55)


40 L Schirrmeister et al The genesis of Yedoma Ice Complex permafrostTable

2(a)Main

modes

ofrobustgrain-sizeE

Ms

(micrombold)theirexplained

variances(

belowm

ode)andthe

totalgrain-sizevariability

()explained

bythe

averagerobustm

odelforeach

studysite

(seealso

Figs14aandS4)(b)M

ainm

odesofrobustgrain-size

EM

s(microm

bold)theirexplainedvariances

(below

mode)and

thetotalgrain-size

variability(

)explained

bythe

averagerobustm

odelfornon-Y

edoma

sitesof

modern

ice-wedge

polygons(see

alsoFigs1S45)(c)

Robustgrain-size

EM

s(in

microm

etersof

main

modebold)

andrespective

explainedvariances

(below

mode)in

regionalandA

rctic-wide

combinations

(seealso

FigS61)

(a)

Locno

Location

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1Total

clayvery

finesilt

finesilt

medium

siltcoarse

siltvery

finesand

finesand

medium

sandcoarse

sandexplained

lt4

microm(4ndash8

microm)

(79ndash119microm

)(12ndash27

microm)

(28ndash50microm

)(51ndash100

microm)

(101ndash280microm

)(281ndash750

microm)

(750ndash865microm

)variability

Alaska

1C

olvilleR

iver39

172330

234182

249278

340133

2Itkillik

River

29172

362634

213273

271260

173147

1483

Seward

Peninsula(K

itlukR

iver)32

157398

5782341

75227

250258

32233

4V

aultCreek

tunnel172

479764

339874

476267

17286

Western

Laptev

Sea

5C

apeM

amontov

Klyk

74300

4791768

786982

409340

71150

31

Lena

Delta

67E

beB

asynSise

andK

hardangSise

islands578

19424096

86213

423153

634211

8K

urungnakhSise

Island39

4792821

786973

347239

276139

Centraland

easternL

aptevSea

9B

ykovskyPeninsula

47250

16123731

69128

249313

31010

Muostakh

Island89

3981612

786974

214115

178384

8639109

11B

uorKhaya

Peninsula32

250839

121876

95320

316269

New

SiberianIslands

andthe

Dm

itryL

aptevStrait

1213141516StolbovoyB

elrsquokovskiynorthernK

otelnysouthwestern

Kotelnyand

Maly

29108

43768

Lyakhovskyislands

116332

55217

Bolrsquoshoy

LyakhovskyIsland

35108

330696

75339

135254

27218

Oyogos

Yarcoast

35108

2741768

72232

128250

23710

153

Yakutian

inland

19D

uvannyY

ar32

330634

66306

183511

20K

ytalyk62

3982132

54378

269353

21B

atagayM

ega-slump

301634

213276

115200

152839

2568433

1002223

Tabagaand

Yukechi

39362

6961612

69124

448167

260



Tabl

e2

Con

tinue

d

(b)

Loc

no

Loc

atio

nrE

M9

rEM

8rE

M7

rEM

6rE

M5

rEM

4rE

M3

rEM

2rE

M1

Tota

lcl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndex

plai

ned

lt4

microm(4

ndash8microm

)(7

9ndash1

19

microm)

(12ndash

27microm

)(2

8ndash50

microm)

(51ndash

100

microm)

(101

ndash280

microm)

(281

ndash750

microm)

(750

ndash865

microm)

vari

abili

ty

Non

-Yed

oma

site

s(m

oder

nic

e-w

edge

poly

gons

)

Pokh

odsk

poly

gon

core

s3

917

263

413

37

7934

68

823

033

5Po

khod

skpo

lygo

nbo

ttom

32

250

121

871

69

7438

216

614

530

7K

ytal

ykpo

lygo

nco

res

32

156

398

194

270

168

309

351

171

Kyt

alyk

poly

gon

botto

m3

218

969

630

96

7326

413

412

647

5K

olym

aan

dB

erel

ekh

flood

plai

ns4

247

917

68

7944

239

516

3

(c)

Reg

ion

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1cl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndlt

4microm

(4ndash8

microm)

(79

ndash11

9microm

)(1

2ndash27

)microm

(28ndash

50microm

)(5

1ndash10

0microm

)(1

01ndash2

80microm

)(2

81ndash7

50microm

)(7

50ndash8

65microm

)

Ala

ska

(site

nos

1ndash4)

35

172

437

282

178

69

289

170

355

130

56

Lap

tev

Sea

and

Eas

tSi

beri

anSe

aco

asts

in-

clud

ing

the

Len

aD

elta

(site

nos

5ndash18

)

35

300

111

086

39

194

346

153

107

213

220

0

Yak

utia

nin

land

(site

nos

19ndash2

3)25

047

992

125

68

386

212

396

07

Arc

tic-w

ide

35

330

921

234

178

69

220

346

209

158

68



Figure 3 Sandndashsiltndashclay diagram of the Yedoma sites in Alaska (diamonds) on the Laptev Sea and East Siberian Sea coasts including theLena Delta (circles) and in the Yakutian inland (triangles) and non-Yedoma sites (squares)

In 4 out of 17 sites (Seward Peninsula New Siberian Is-lands Duvanny Yar Kytalyk) only three rEMs contribute tothe grain-size variability whereas the sites from the LaptevSea coast and the Lena Delta are composed of four to fiverEMs Figure 4a shows examples of the rEM loadings andthe original GSDs from the Laptev Sea coast and the LenaDelta The rEM modeling results for all study sites shown inFigs S41ndashS45 4b and S51ndashS55 present the mean scoresie the relative contribution of a rEM to each sample Themean scores of the studied sample sets show the internalvariability in the composite Yedoma profiles or Yedoma se-quences taken from different sites along riverbank and coastsegments

Coarse sand rEMs (rEM 1 between 750 and 865 microm ex-plained variance 31 to 384 ) were calculated for threesites (Table 2a Fig 5) in the Laptev Sea region Mediumsand rEMs (rEM 2) between 280 and 750 microm (explained vari-ances 86 to 310 ) occurred at one site in Alaska ontwo islands in the Lena Delta and at one site on the easternLaptev Sea coast (Table 2a Fig 5) Fine sand rEMs (rEM 3101ndash280 microm) are present in all regions (Table 2a Fig 5) egforming the coarsest grain-size components at two sites inAlaska at two sites in Central Yakutia and on the OyogosYar coast

Two or three rEMs out of all rEMs are in the silt rangeSeveral rEMs with main modes in the coarse silt to veryfine sand range that we have grouped into two main rEMsndash rEM 4 (51ndash100 microm) and rEM 5 (28ndash50 microm) ndash occur inmost of the sites These size classes constitute the finest andsecond-finest rEMs of the combined Ebe Basyn Sise andKhardang Sise islands regional group (ie main mode at

578 microm) These rEMs have the highest explained variancesin the Yakutian inland (Table 2a Fig 5)

Medium silt rEMs (rEM 6 12ndash27 microm) explain variance of115 to 476 of eight sites in the three regions (Fig 5)

Fine silt rEMs (rEM 7) between 8 and 119 microm (explainedvariance 135 to 332 ) are calculated for four sites butnot in the Yakutian inland Very fine silt rEMs (rEM 8) be-tween 4 and 79 microm (explained variance 124 to 378 )occur at six sites (Table 2a) throughout the study regions Theclay rEMs (rEM 9) between 1 and 4 microm (explained varianceof 92 to 339 ) are present at seven sites (Table 2a) thefinest rEM with a main mode at 1 microm occurs on the OyogosYar coast

The reference sites of non-Yedoma ice-wedge polygonshave clay (rEM 9) and very fine silt (rEM 8) with explainedvariances between 168 and 442 fine to coarse silt(rEMs 7 to 5) with explained variances of 88 to 351 and very coarse silt to coarse sand (rEMs 4 to 2) with ex-plained variances of 126 to 475 (Table 2b Fig 5)

33 Results of regional and Arctic-wide endmemberanalyses

The rEMs of the regional aggregation of sites as well as therEMs from all Arctic-wide sites are presented in Table 2c andFigs 5 S61 and S63 (see scores in Figs S62 and S63)They show that very fine silt rEMs exist in Alaska and theLaptev and East Siberian seas region but not in the Yaku-tian inland Fine and medium silt rEMs are missing in theregional and Arctic-wide combinations Coarse silt rEMs oc-cur in all three regions with the coarsest silt rEMs occur-



Figure 4



Figure 4 (a) Examples of GSD curves and rEM modeling from six sites on the Laptev Sea coast and from islands in the Lena Delta TheGSD diagrams reflect different curves for each site GSD curves from all studied Yedoma sites and non-Yedoma reference sites are given inFigs S41ndash45 (sediment nomenclature according to Blott and Pye 2001) The rEM modeling shows a great variety with four to five differentrEMs The dotted lines show the standard deviation for each rEM Grey lines show original GSDs The EM modeling results for all studysites are also shown in Figs S41ndash45 (b) Examples of the mean scores (ie the relative contribution of a rEM to each sample) for the LaptevSea coast and the Lena Delta sites Please note for better comparability scores are plotted in the same stratigraphic order ie from top (left)to bottom (right) The mean score results for all study sites are shown in Figs S51ndash55

ring in Alaska and the Yakutian inland Very fine sand rEMsare calculated for sites on the Laptev Sea and East SiberianSea coasts and for the Yakutian inland but not for AlaskaA medium sand rEM occurs only in the Laptev and EastSiberian seas region while coarse sand rEMs are found inAlaska and the Yakutian inland

34 Results of cluster analysis

Clustering the study sites based on the explained variancesin the rEMs resulted in four significant clusters and threeadditional sites that could not be combined into a cluster(Fig S71) according to a corrected approximately unbiasedsignificance value above 09 There are two to five sites in one

cluster The outliers (black in Fig S71) comprise the NewSiberian Islands Muostakh Island and Kurungnakh Sise Is-land Cluster 1 consists of Bolrsquoshoy Lyakhovsky Island andthe Duvanny Yar site cluster 2 includes one Yakutian inlandsite (Batagay) three sites at the Laptev Sea (Ebe Basyn Siseand Khardang Sise Bykovsky Buor Khaya) and the AlaskanVC tunnel All three other Alaskan sites fall into cluster 3together with two Siberian sites from the Laptev Sea coast(Oyogos Yar) and the Yakutian inland (Tabaga and Yukechi)Finally Cluster 4 consists of Kytalyk and Cape MamontovKlyk both in the Laptev Sea region The sites that cluster to-gether are often hundreds or thousands of kilometers away



Figure 5 Position of the rEMs and their explained variances and the overall explained variance for each calculated rEM in the entire samplecollection the regional and Arctic-wide combinations and the non-Yedoma references On the left side the different clusters from the clusteranalyses are shown (more details are explained in the text)



from each other and have no common geological historysource rocks or typical distances to late Pleistocene glaciers

The clusters that include Siberian sites are more or lessrandomly distributed (Fig 6) indicating that grouping ofsites is determined by factors other than geographical loca-tion Instead clusters show a typical rEM composition Clus-ter 1 lacks the coarse rEMs 1ndash3 the medium silt rEM 6and very fine silt rEM 8 while cluster 4 is dominated bythe coarser rEMs 1 3 and 5 and a distinctive contributionof rEM 8 Cluster 3 like cluster 1 lacks the coarse rEMs 1and 2 and the very fine silt rEM 8 Cluster 2 only lacks thecoarse sand rEM 1 and the fine silt rEM 7 Statistically sig-nificant differences between the clusters are evident for therEMs with main modes in the very fine sand and very finesilt classes (rEMs 4 and 8 p values le 005 Fig S72)

4 Discussion

41 Interpretation of endmember modeling analyses

The multimodal GSD curves of Yedoma deposits from thestudied sites of the Laptev Sea region from Alaska andfrom the Yakutian inland indicate a wide range of contribut-ing grain-size subpopulations that could be unmixed by arEMMA Grain-size characteristics and contributing rEMsdiffered within the horizons of a site as well as between studysites confirming the results of heterogeneous GSDs acrossthe Yedoma region (Schirrmeister et al 2011)

These subpopulations likely reflect different sediment pro-duction transport depositional and postdepositional pro-cesses Here we interpret the main modes of the nine rEMsthat cluster in similar grain-size classes across sites (Table 2Fig 5) suggesting that common processes were involvedin Yedoma formation The main assumption is that higherenergy is required to mobilize and transport coarser com-pared to finer sediment We consider fluvial deposits fromlarge streams and from temporary meltwater creeks as simi-larly important parts of the periglacial sediments that com-prise Yedoma deposits as are deposits from aeolian pro-cesses (Murton et al 2015 2017 Peacuteweacute and Journaux 1983Tomirdiaro 1996)

However the interpretation of the contributions of rEMsubpopulations to the individual sample compositions (ierEM scores) can only occur in general terms becausecomparing samples and depositional environments acrossYedoma sites is hampered by postdepositional cryogenicprocesses such as cryoturbation and ground ice formationthat complicate assigning an age to individual samples andhence their temporal comparison In addition due to neotec-tonics (seismotectonics and isostatic adjustments followingthe deglaciation) we cannot reconstruct site-specific catch-ments and ancient fluvial sediment pathways to determinelocal sediment transport processes

42 Yedoma grain-size endmembers and associatedprocesses

High transport energies are required to move the coarse sandgrains (rEM 1) that are found at three sites in the LaptevSea region with the highest contributions to the sedimentof Muostakh Island deposited ca 20ndash39 kyr ago as well asmedium sands (rEM 2) that dominate on three other LaptevSea islands The two rEMs found at these sites (explainedvariance 109 to 552 ) point to high-energy processesie saltation and traction processes in confined running wa-ter such as during strong meltwater runoff with the mediumsands forming the main saltation component (Visher 1969Sun et al 2002 Cockburn and Lamoureux 2008) Fluvialsands of rEMs 1 and 2 are coarser compared to aeoliansands from modes in the medium sand (eg 200ndash400 micromSun et al 2002) to coarser sand fractions and they are morepoorly sorted as occurs for example in alluvial fan envi-ronments (Tsoar and Pye 1987 Pendea et al 2009 Northand Davidson 2012) On the Tibetan Plateau fluvial sandswith modes of around 450 microm appeared as rEMs in lakesediments (Dietze et al 2014) We cannot assess the an-cient topographic position of these coarse-grained depositionsites anymore Yet all these sites are located in the vicin-ity of Permo-Carboniferous sandstone outcrops (Table S2)which could have served as sand sources Modern pond sub-strates from modern ice-wedge polygons also exhibit rEM 2with explained variances of 307 to 475 (Table 2bFigs 5 S25)

The fine sand rEM 3 between 101 and 280 microm (explainedvariance 14 to 511 ) was found for 11 sites (Table 2a)in all study regions Studies of modern snow patches showgrain-size means in the same fraction linking this rEM withniveo-aeolian deposition (Galabala 1997 Kunitsky et al2002) which could have been reworked postdepositionallyby runoff below or on top of the snow patch These rEMsalso include the size classes of local aeolian sands that rolland saltate due to strong surface winds (Tsoar and Pye 1987Sun et al 2002 Vandenberghe 2013 Dietze et al 2014)When showing a well-sorted rEM distribution as at the Itkil-lik Colville and Mamontov Klyk sites these sands couldindicate local dune deposits (Tsoar and Pye 1987 Sun et al2002) andor additional sorting by unconfined alluvial flow(North and Davidson 2012 Pendea et al 2009) The rEM 3explained 145 to 335 of the variance on average in themodern polygon tundra samples (Table 2b Fig S25) sug-gesting that polygonal structures are ideal sites to retain snowpatches and trap local sediments

Two rEMs fall within the very fine sand to coarse siltfractions (rEM 4 51ndash100 microm explained variance of 32 to 511 and rEM 5 28ndash50 microm explained variance of115 to 448 ) Only two sites (Bykovsky Peninsula NewSiberian Islands) did not exhibit these rEMs These sub-populations could be explained as fine-grained overbank de-posits or settled suspended loads in temporarily flooded sec-



Figure 6 Distribution of the clustered sites in Beringia according to the hierarchical cluster analyses (Fig S71) Location numbers areexplained in Table 1 The color codes of the legend denote the cluster corresponding to Fig S72

tions with small slopes or shallow flow depth (Visher 1969Cockburn and Lamoureux 2008) Accordingly rEM 4 ex-plains most of the variance in modern floodplain sedimentof the Berelekh and the Kolyma rivers (Table 2b) and bothrEM 4 and rEM 5 were present in drill cores from ice-wedge polygons (Table 2b Fig S25) In addition the sub-population rEM 5 could also be explained as primary orsecondary (reworked) aeolian material (Vandenberghe 2013Vandenberghe et al 2018) The disintegration of coarsergrains by repeated frost weathering processes (Viran and Bi-nal 2018) could also contribute to these rEM 4 and 5 frac-tions Schwamborn et al (2012) showed that experimen-tal frost weathering of fine sand samples (63ndash125 microm) byup to 230 freezendashthaw cycles leads to an increase of up to25 in the lt 63 microm fraction of a sand sample this pro-cess seems likely to occur in ice-rich Yedoma sedimentsSnow patches could also have acted as sediment traps form-ing niveo-aeolian deposits (Galabala 1997 Kunitsky et al2002) The studied snow patch samples from Kunitsky etal (2002) exhibit an arithmetic mean from 22 to 491 microm(median 10ndash381 microm) Very fine sand to silt fractions are alsocharacteristic of grain-size subpopulations of dune sand andcoarse local dust in present-day arid and periglacial environ-ments (Tsoar and Pye 1987 Dietze et al 2014 Vanden-berghe 2013) These rEMs explain most of the grain-sizevariances at Yakutian inland sites (Table 2a Fig 5) whereduring glacial time a grassy steppe environment (Fradkinaet al 2005a b Ashastina et al 2018) could have providedthe surface roughness required for the deposition of sedimentfrom local low-energy floods andor aeolian transport duringstorms and short-term near-surface suspension clouds as de-

scribed from modern-day periglacial settings (Stauch et al2012 Dietze et al 2014)

Medium silt rEM 6 between 12 and 27 microm with an ex-plained variance between 249 and 476 dominates atsix sites and is present in Alaska and the Laptev Sea (Ta-ble 2a Fig 5) This could be the result of unconfined flu-vial and alluvial sediments that settled out of suspension inponding water when the Shields stress fell below a criti-cal threshold needed for motion (Dietrich 1982) Mediumsilts can either reflect low-energy sediment delivery from themain fluvial suspension component during snowmelt (Sun etal 2002 Macumber et al 2018 Cockburn and Lamoureux2008 Visher 1969) or represent coarse regional dust that canremain in suspension for several days during storm events(Tsoar and Pye 1987 Dietze et al 2014) The rEM 6 ex-plained 134 to 309 of the explained variance in themodern polygon tundra samples (Table 2b Fig S45) sug-gesting that fluvial suspension components contributed tomodern and Yedoma deposits

The fine silt rEM 7 between 8 and 12 microm was present atfour sites in the Laptev Sea on the New Siberian Islandsand in the Dmitry Laptev Strait (Table 2a Fig 5) but notin Alaska the Lena Delta or the Yakutian inland Our ice-wedge polygon references do not show this rEM 7 In addi-tion dust that traveled over distances of several hundred kilo-meters (Vandenberghe 2013 Tsoar and Pye 1969 Dietze etal 2014) could have accumulated along rough surfaces sim-ilar to other aeolian components (see also experiments on drydust deposition by Goossens 2005)

Very fine silt rEM 8 between 4 and 79 microm (explained vari-ance between 128 and 409 ) occurred at three sites



which are 500 to 1000 km apart from each other (Table 2a)This could have originated from fine silts that were trans-ported in low-energy fluvial suspension for a long time ascould occur in larger streams (eg Kytalyk in the Yana-Indigirka Lowland) or in recurring meltwater runoffs whichwould require still water conditions for the sediments to set-tle (Tsoar and Pye 1987 Visher 1969) On the Kolyma andBerelekh river floodplains rEM 8 deposits explain 442 ofthe variance (Table 2b Fig S45) This rEM 8 could alsoderive from primary or fluvially reworked background dust(Vandenberghe 2013 Dietze et al 2014) or even from pri-mary or fluvially reworked pedogenic clay (Vandenberghe etal 2018) potentially deposited in polygon ponds such as thepolygon structures where the reference samples are from

Similarly clays between 1 and 4 microm (rEM 9 explainedvariance of 92 to 339 ) were present at 10 sites (Ta-ble 2a Fig 5) Similar to the very fine silts calm water con-ditions such as under frozen surfaces of ponds or small lakesie palustrine conditions are required for these clays to set-tle (Cockburn and Lamoureux 2008 Francus et al 2008Dietze et al 2014) The finest rEM at the Oyogos Yar coastsite can only be explained by postdepositional conditionsClay was formed in situ during pedogenesis (Schirrmeisteret al 2013 Strauss et al 2017) as it happened during theshort but warm summers of the interstadials (Kienast et al2005 Andreev et al 2011) Clays could also be concentratedinto polygon ponds by cryogenic reworking rEM 9 consti-tuted 168 to 382 of the sediment in modern ice-wedgepolygon samples (Table 2b Fig 5) Hence frost weatheringmight have been more important for postdepositional graindisintegration in polygon ponds compared to at dry sites

43 Synthesis

Overall various Yedoma types exist across the large regionwhere Yedoma occurs (Fig 6) ranging from spatially con-fined Yedoma valley fills along for example the Lena YanaIndigirka and Kolyma rivers to vast accumulation plains onArctic lowlands and shelves Cluster analyses revealed fourdistinct site clusters each comprised of two to five sites thatare distributed across Siberia Yet the three major regionsAlaska the Laptev and East Siberian seas and Yakutian in-land could not be differentiated by unique rEM configura-tions Accordingly rEM modes did not show a distinct re-gional or Arctic-wide pattern

The large variety and spread of rEMs from very coarse tovery fine grain sizes suggests different source areas and dif-ferent transport and depositional processes that act along thesediment cascades from source rock to final deposition andsubsequent reworking (Fig 7) In cold environments physi-cal weathering of clayey silty and sandy source rocks deter-mines the grain sizes available for transport from local andregional sources From an energetic point of view grain-sizerEMs gt 250 microm and lt 2 microm can only be explained by flu-vial transport (Dietze et al 2014 and references therein) or

in the case of the lt 2 microm fraction also by chemical weather-ing Silts and very fine sands could have derived from sev-eral transport and depositional mechanisms that depend onshear stress and shear velocities modified by surface rough-ness grain densities cohesiveness and other properties ofthe transport medium such as water flow depth or wind fields(Tsoar and Pye 1987 Visher 1969 Dietrich 1982 Van-denberghe 2013 Dietze et al 2014) In the Yedoma sam-ples we find a link in that sites of coarser rEMs tend tobe located close to sandy source rocks Yet de facto fluvialndashalluvial catchment areas and fluvial pathways are difficult toconstrain for our sites and samples due to isostatic neotecton-ics during the Holocene and seismotectonics in modern times(eg Franke et al 2000 Grigoriev et al 1996) Thereforethe identification of specific catchments as sources of fluvialand aeolian deposits is not possible without further for ex-ample mineralogical information Yet both local to regionalfluvial and aeolian transport and depositional processes arerepresented by several rEMs especially in deposits of theperiglacial environments of MIS 2 and 3 during the late Pleis-tocene (Fig 7) For example the trapping of all types of ae-olian sediment in snow patches forming niveo-aeolian de-posits (Bateman 2013) is supported by snow patch samplesfrom Kunitsky et al (2002) that showed arithmetic meansfrom 22 to 491 microm

Yet the finding of primary grain-size modes across allYedoma deposits outside the range of classical aeolian de-posits (ie coarser and finer) supports the hypothesis of apolygenetic origin of Yedoma deposits including alluvialfluvial and palustrine processes (Sher 1997 Schirrmeisteret al 2013 Fig 7) In addition further postdepositional pro-cesses occur in periglacial environments such as cryotur-bation mass wasting solifluction frost weathering and re-working (Francus et al 2008 French 2018 Bateman 2013van Huissteden et al 2013 Strauss et al 2012 Dietze et al2014 Fig 7) These processes might have affected the grain-size composition directly via postdepositional disaggrega-tion (eg Schwarmborn 2012) or soil (clay) formation dur-ing warmer interstadials (eg Munroe and Bockheim 2001Ping et al 2015)

However the dominance of ground ice and cryostrati-graphic and cryolithologic properties in Yedoma depositsis key to understanding the genetic distinction between theYedoma Ice Complex and other permafrost deposits of al-luvial fluvial and aeolian origin that did not have a ma-jor ground ice component Various ground-ice-forming pro-cesses during and after sediment deposition increase the in-fluence of frost weathering and cryoturbation processes com-pared to ice-free deposits These processes can principallyaffect all grain sizes with potentially higher contributionsto rEM 4 and 5 (see above) Yet assessing the contribu-tion of secondary cryogenic processes relative to primarysedimentation processes would require further studies thatfor example characterize microstructures on mineral grains



Figure 7 The polygenetic origin of Yedoma Ice Complex including (a) primary accumulation areas (b) sediment formation (c) sedimenttransport and (d) accumulation including postsedimentary alteration (modified after Schirrmeister et al 2013)

(Woronko and Pisarska-Jamrozy 2016) andor compare withpermafrost deposits of low ice content

Indirectly all postdepositional sediment reworking pro-cesses affect rEM interpretation across larger geographic ar-eas adding uncertainties to the already large dating uncer-tainties Still the quantification of the contribution of grain-size EMs to Yedoma samples allows us to assess the relativeimportance of certain sedimentary and postdepositional pro-cesses The four rEM clusters are not related to a certain ge-ographic area Yet they are differentiated by a characteristiclack of certain rEMs in the coarse sand very fine sand andvery fine silt classes which allows for the characterization ofcertain types of Yedoma deposits from a granulometric pointof view These types and their full grain-size distributionsmay help to improve parametrization in a reasonable way infurther permafrost modeling and carbon storage estimates

5 Conclusions

We applied grain-size endmember modeling to a large pan-Arctic sample data set of GSD data to differentiate between

possible sedimentation processes responsible for the forma-tion of Yedoma in Beringia during the late Pleistocene Wecharacterized up to nine robust grain-size endmembers orrEMs within Yedoma Ice Complex deposits We interpretthese rEMs as signals of diverse sedimentation processesfrom local to regional aeolian alluvial fluvial nival andponding water accumulation which occurred in a polygo-nal landscape and likely contributed to the minerogenic pro-cess in different Yedoma regions These deposits were notonly frozen into permafrost over thousands of years but alsoperiglacially altered during this time for example by in situfrost weathering The observed variability in grain-size rEMssupports the hypothesis of a polygenetic Yedoma origin in-volving multiple transport depositional and transformationprocesses Each Yedoma site had a different rEM composi-tion dominated by silty rEMs with cluster analyses revealingfour distinct rEM composition clusters

The diversity of Yedoma deposits results from multiplesediment origins and transport and (post)depositional sedi-mentary processes This has strong implications for assess-ments of the role of Yedoma permafrost in the future Sed-



iment properties such as cryolithologic properties sedimentpore volumes and field capacities are mediated by grain sizesand are important parameters for modeling for example car-bon storage capacities ground ice content and permafrostthaw rates that are very different in sand-dominated com-pared to silt- and clay-dominated deposits (Strauss et al2013 Langer et al 2016) Hence next to ground ice contentfuture assessments of climate change impacts on circum-Arctic permafrost deposits need to consider the current grain-size compositions that are ultimately determined by past sed-imentation histories

Data availability Location descriptions and manydata sets have been submitted to PANGAEAhttpsdoiorg101594PANGAEA877882 (Schirrmeister 2017a)httpsdoiorg101594PANGAEA877886 (Schirrmeister 2017b)httpsdoiorg101594PANGAEA884072 (Schirrmeister et al2017a) httpsdoiorg101594PANGAEA877346 (Ashastina etal 2017a) httpsdoiorg101594PANGAEA877345 (Ashastinaet al 2017b) httpsdoiorg101594PANGAEA884063(Schirrmeister et al 2017b)httpsdoiorg101594PANGAEA611549 (Schirrmeister 2007a)httpsdoiorg101594PANGAEA615798 (Schirrmeister 2007b)httpsdoiorg101594PANGAEA887933 (Schirrmeister et al2018a) httpsdoiorg101594PANGAEA858643 (Schirrmeis-ter et al 2016) httpsdoiorg101594PANGAEA880929(Schirrmeister et al 2017c) tohttpsdoiorg101594PANGAEA880931 (Schirrmeister et al2017d) httpsdoiorg101594PANGAEA880951 (Schirrmeis-ter et al 2017e) httpsdoiorg101594PANGAEA887931(Schirrmeister et al 2018b) andhttpsdoiorg101594PANGAEA884069 (Schirrmeister et al2017f)

Supplement The supplement related to this article is availableonline at httpsdoiorg105194egqsj-69-33-2020-supplement

Author contributions LS collected most of the samples overmore than the last 20 years and carried out the evaluation of manyanalyses ED and HM performed the rEMMA SL was responsiblefor preparing the maps GG JS MU FK and SW were involved inthe sampling during numerous expeditions and evaluated grain-sizedata for different locations LS ED and HM wrote the manuscriptwith contributions from all coauthors

Competing interests The authors declare that they have no con-flict of interest

Acknowledgements This studies are embedded into the ActionGroup ldquoThe Yedoma Regionrdquo of the International Permafrost Asso-ciation (IPA) The studies were supported by several internationaland national funding organizations of Germany Russia the USAand the EU listed in detail in the financial support section

We thank numerous colleagues for supporting our field sam-pling in Russia and Alaska We thank Ute Bastian (Kuschel) andDyke Scheidemann for the laboratory work The manuscript ben-efited from English language corrections by Candace OrsquoConnor(Fairbanks Alaska)

The authors also thank Jef Vandenberghe and two anonymousreviewers for their constructive comments and suggestions

Financial support This research has been supported by the Bun-desministerium fuumlr Bildung und Forschung (grant nos 03G053403G0589 03G0836A 01DM12011 03F0806A) the InternationalAssociation for the Promotion of Cooperation with Scientistsfrom the Independent States of the Former Soviet Union (INTAS(grant no 05-1000008-8133)) the Deutsche Forschungsgemein-schaft (grant nos HE 362216-1 164232461 DI 25441-1419058007 WE43907-1 317774679 UL4261-1 232311661 KI8494-1 247453756) the NASA Carbon Cycle Sciences (grantno NNX08AJ37G) the National Science Foundation Office ofPolar Programs (grant no 0732735) the European CommissionCordis (PETA-CARB (grant no 338335)) and the HelmholtzAssociation (grant no ERC-0013)

The article processing charges for this open-accesspublication were covered by a ResearchCentre of the Helmholtz Association

Review statement This paper was edited by Christian Zeedenand reviewed by Jef Vandenberghe and two anonymous referees

References

Anderberg M R Cluster Analysis for Applications AcademicPress New York p 376 ISBN 0120576503 1973

Andreev A A Schirrmeister L Tarasov P E Ganopol-ski A Brovkin V Siegert C and Hubberten H-WVegetation and climate history in the Laptev Sea re-gion (arctic Siberia) during Late Quaternary inferredfrom pollen records Quaternary Sci Rev 30 2182ndash2199httpsdoiorg101016jquascirev201012026 2011

Ashastina K Schirrmeister L Fuchs M C and Kienast FOSL age determination and sedimentological characteristics ofthe Batagay thaw slump Northeastern Siberia PANGAEAhttpsdoiorg101594PANGAEA877346 2017a

Ashastina K Schirrmeister L Scheidemann D FuchsM C and Kienast F Grain size distribution of theBatagay thaw slump Northeastern Siberia PANGAEAhttpsdoiorg101594PANGAEA877345 2017b

Ashastina K Kuzmina S Rudaya N Troeva E Schoch W HRoumlmermann C Reinecke J Otte V Savvinov G WescheK and Kienast F Woodlands and steppes Pleistocene vege-tation in Yakutiarsquos most continental part recorded in the Bata-gay permafrost sequence Quaternary Sci Rev 196 38ndash61httpsdoiorg101016jquascirev201807032 2018

Bartholdy J Christiansen C and Pedersen J B T Com-paring spatial grain-size trends inferred from textural param-eters using percentile statistical parameters and those based



on the log-hyperbolic method Sediment Geol 202 436ndash452httpsdoiorg101016jsedgeo200703008 2007

Bateman M D Aeolian processes in periglacial environments inTreatise on Geomorphology edited by Shroder J San DiegoCA Academic Press 416ndash429 httpsdoiorg101016B978-0-12-374739-600219-0 2013

Blott S J and Pye K A GRADISTAT grain size distri-bution and statistics package for the analysis of unconsoli-dated sediments Earth Surf Processes Landf 26 1237ndash1248httpsdoiorg101002esp261 2001

Cockburn J M H and Lamoureux S F Inflow and lake con-trols on short-term mass accumulation and sedimentary particlesize in a High Arctic lake implications for interpreting varvedlacustrine sedimentary records J Paleolimnol 40 923ndash942httpsdoiorg101007s10933-008-9207-5 2008

DallrsquoAmico M Endrizzi S Gruber S and Rigon R A robustand energy-conserving model of freezing variably-saturated soilThe Cryosphere 5 469ndash484 httpsdoiorg105194tc-5-469-2011 2011

Dietrich W E Settling velocity of natural par-ticles Water Resour Res 18 1615ndash1626httpsdoiorg101029WR018i006p01615 1982

Dietze E and Dietze M Grain-size distribution unmixing usingthe R package EMMAgeo EampG Quaternary Sci J 68 29ndash46httpsdoiorg105194egqsj-68-29-2019 2019

Dietze E Hartmann K Diekmann B IJmker J LehmkuhlF Opitz S Stauch G Wuumlnnemann B and Borchers AAn end-member algorithm for deciphering modern detrital pro-cesses from lake sediments of Lake Donggi Cona NE Ti-betan Plateau China Sediment Geol 243ndash244 169ndash180httpsdoiorg101016jsedgeo201109014 2012

Dietze E Maussion F Ahlborn M Diekmann B HartmannK Henkel K Kasper T Lockot G Opitz S and HaberzettlT Sediment transport processes across the Tibetan Plateau in-ferred from robust grain-size end members in lake sedimentsClim Past 10 91ndash106 httpsdoiorg105194cp-10-91-20142014

Ehlers J Gibbard P L and Hughes P D Quaternary Glacia-tions ndash Extent and Chronology a Closer Look Developments inQuaternary Science vol 15 Elsevier Amsterdam available athttpsbooksiteelseviercom9780444534477 last access 2011

ETOPO2 National Geophysical Data Center NESDIS NOAAUS Department of Commerce 2-minute Gridded Global ReliefData (ETOPO2) v2 httpsdoiorg107289v5j1012q 2006

Folk R L and Ward W C A study in the signifi-cance of grain-size parameters J Sediment Petrol27 3ndash26 httpsdoiorg10130674D70646-2B21-11D7-8648000102C1865D 1957

Fradkina A F Alekseev M N Andreev A A and KlimanovV A East Siberia in Cenozoic Climatic and EnvironmentalChanges in Russia edited by Velichko A A and Nechaev V PThe Geological Society of America Special Paper 382 89ndash1032005a

Fradkina A F Grinenko O V Laukhin S A Nechaev V P An-dreev A A and Klimanov V A North-eastern Asia in Ceno-zoic Climatic and Environmental Changes in Russia edited byVelichko A A and Nechaev V P The Geological Society ofAmerica Special Paper 382 105ndash120 2005b

Francus P Bradley R Lewis T Abbott M Retelle Mand Stoner J Limnological and sedimentary processesat Sawtooth Lake Canadian High Arctic and their in-fluence on varve formation J Paleolimnol 40 963ndash985httpsdoiorg101007s10933-008-9210-x 2008

Franke D Hinz K Block M Drachev S S Neben S KosrsquokoM K Reichert C and Roeser H A Tectonics of the LaptevSea Region in Northeastern Siberia Polarforschung 68 51ndash582000

French H M The Periglacial Environment 4th Edition WileyChichester UK and Hoboken New Jersey 544 pp 2018

Galabala R O Pereletki and the initiation of glaciation in SiberiaQuaternary Int 41ndash42 27ndash32 httpsdoiorg101016S1040-6182(96)00033-X 1997

Goossens D Quantification of the dry aeolian depositionof dust on horizontal surfaces an experimental comparisonof theory and measurements Sedimentology 52 859ndash873httpsdoiorg101111j1365-3091200500719x 2005

Grigoriev M N Imaev V S Kozrsquomin B M Kunitski V V Lar-ionov A G Mikulenko K I Skryabin R M and TimirshinK V Geology seismicity and cryogenic processes in the arcticareas of western Yakutia 80 Scientific Center SD RAS Yakutsk1996 (in Russian)

Hulteacuten E Outline of the History of Arctic and Boreal Biota dur-ing the Quaternary Period Bokfoumlrlags aktiebolaget Thule Stock-holm 168 pp 1937

Kanevskiy M Shur Y Fortier D Jorgenson M T and StephaniE Cryostratigraphy of late Pleistocene syngenetic permafrost(yedoma) in northern Alaska Itkillik River exposure QuaternaryRes 75 584ndash596 httpsdoiorg101016jyqres2010120032011

Kanevskiy M Shur Y L Strauss J Jorgenson M TFortier D Stephani E and Vasiliev A Patterns andrates of riverbank erosion in the area of ice-rich permafrost(yedoma) in northern Alaska Geomorphology 253 370ndash384httpsdoiorg101016jgeomorph201510023 2016

Kaplina T P History of the frozen strata of northern Yakutia in thelate Cenozoic in History of permafrost development in EurasialdquoNaukardquo Publishing House Moscow 153ndash181 1981 (in Rus-sian)

Kienast F Schirrmeister L Siegert C and Tarasov P Palaeob-otanical evidence for warm summers in the East Siberian Arc-tic during the last cold stage Quaternary Res 63 283ndash300httpsdoiorg101016jyqres200501003 2005

Kunitsky V Schirrmeister L Grosse G and Kienast F Snowpatches in nival landscapes and their role for the Ice Complexformation in the Laptev Sea coastal lowlands Polarforschung70 53ndash67 httpsdoiorg102312polarforschung7053 2002

Langer M Westermann S Boike J Kirillin G GrosseG Peng S and Krinner G Rapid degradation of per-mafrost underneath waterbodies in tundra landscapes-Toward a representation of thermokarst in land surfacemodels J Geophys Res-Earth Surf 121 2446ndash2470httpsdoiorg1010022016JF003956 2016

Lenton T M Arctic Climate Tipping Points Ambio 41 10ndash22httpsdoiorg101007s13280-011-0221-x 2012

Macumber A L Patterson R T Galloway J M Falck Hand Swindles G T Reconstruction of Holocene hydrocli-matic variability in subarctic treeline lakes using lake sed-



iment grain-size end-members The Holocene 28 845ndash857httpsdoiorg1011770959683617752836 2018

Muhs D R Ager T A Skipp G Beann J Budahn J andMcGeehin J P Paleoclimatic significance of chemical weather-ing in loess-derived paleosols of subarctic central Alaska ArctAntarct Alp Res 40 396ndash411 httpsdoiorg1016571523-0430(07-022)[MUHS]20CO2 2008

Munroe J S and Bockheim J G Soil development in low-arctic tundra of the northern Brooks Range Alaska USA ArctAntarct Alp Res 33 78ndash87 httpsdoiorg10230715522802001

Murton J B Goslar T Edwards M E Bateman M D DanilovP P Savvinov G N and Gubin S V Palaeoenvironmen-tal interpretation of Yedoma silt (Ice Complex) deposition ascold-climate loess Duvanny Yar northeast Siberia PermafrostPeriglac 26 208ndash288 httpsdoiorg101002ppp1843 2015

Murton J B Edwards M E Lozhkin A V Anderson P MSavvinov G N Bakulina N Bondarenko O V CherepanovaM Danilov P P Boeskorov V Goslar T Grigoriev S Gu-bin S V Korzun J Lupachev A V Tikhonov A Tsy-gankova V I and Zanina O G Preliminary paleoenviron-mental analysis of permafrost deposits at Batagaika megaslumpYana Uplands northeast Siberia Quaternary Res 87 314ndash330httpsdoiorg101017qua201615 2017

North C P and Davidson S K Unconfined alluvial flow pro-cesses Recognition and interpretation of their deposits andthe significance for palaeogeographic reconstruction Earth-SciRev 111 199ndash223 2012

Palmtag J and Kuhry P Grain size controls on cryotur-bation and soil organic carbon density in permafrost-affected soils Permafrost Periglac 29 112ndash120httpsdoiorg101002ppp1975 2018

Pendea I F Gray J T Ghaleb B Tantau I Badarau A S andNicorici C Episodic build-up of alluvial fan deposits duringthe Weichselian Pleniglacial in the western Transylvanian BasinRomania and their paleoenvironmental significance QuaternaryInt 198 98ndash112 httpsdoiorg101016jquaint2008050022009

Peacuteweacute T L Origin of the upland silt near Fairbanks Alaska GeolSoc Am Bull 66 699ndash724 1955

Peacuteweacute T L Quaternary geology of Alaska US Geological SurveyProfessional Paper 835 p 143 1975

Peacuteweacute T L and Journaux A Origin and character of loess-like siltin unglaciated south-central Yakutia Siberia USSR US Geolog-ical Survey Professional Paper 1262 p 46 1983

Ping C L Jastrow J D Jorgenson M T Michaelson G J andShur Y L Permafrost soils and carbon cycling SOIL 1 147ndash171 httpsdoiorg105194soil-1-147-2015 2015

Schirrmeister L Documentation of sediment profile Kha-3 PAN-GAEA httpsdoiorg101594PANGAEA611549 2007a

Schirrmeister L Documentation of outcrop Mamontovy KhayataPANGAEA httpsdoiorg101594PANGAEA615798 2007b

Schirrmeister L Cryolitholgical biogeochemical and geochrono-logical data from Byk_98_Mkh Bykovsky Peninsula in 1998Alfred Wegener Institute ndash Research Unit Potsdam PANGAEAhttpsdoiorg101594PANGAEA877882 2017a

Schirrmeister L Cryolitholgical biogeochemical and geochrono-logical data from Byk_99_Mkh Bykovsky Peninsula in 1999

Alfred Wegener Institute ndash Research Unit Potsdam PANGAEAhttpsdoiorg101594PANGAEA877886 2017b

Schirrmeister L Kunitsky V V Grosse G Wetterich S MeyerH Schwamborn G Babiy O Derevyagin A Y and SiegertC Sedimentary characteristics and origin of the Late Pleis-tocene Ice Complex on North-East Siberian Arctic coastallowlands and islands ndash a review Quaternary Int 241 3ndash25httpsdoiorg101016jquaint201004004 2011

Schirrmeister L Froese D Tumskoy V Grosse G and Wet-terich S Yedoma Late Pleistocene ice-rich syngenetic per-mafrost of Beringia in The Encyclopedia of Quaternary Science2nd Edition vol 3 edited by Elias S A Elsevier Amsterdam542ndash552 2013

Schirrmeister L Pestryakova L A Schneider Aand Wetterich S Characteristics of samples ob-tained during Pokhodsk 2012-2013 campaigns in thejoint Russian-German POLYGON Project PANGAEAhttpsdoiorg101594PANGAEA858643 2016

Schirrmeister L Grosse G Kunitsky V V and SiegertC Sedimentological biogeochemical and geochrono-logical data from the Lena Delta 2000 PANGAEAhttpsdoiorg101594PANGAEA884072 2017a

Schirrmeister L Grosse G Kunitsky V V and SiegertC Sedimentological biogeochemical and geochronolog-ical data from permafrost deposit Nagym PANGAEAhttpsdoiorg101594PANGAEA884063 2017b

Schirrmeister L Grosse G Kunitsky V V and SiegertC Sedimentological biogeochemical and geochrono-logical data from permafrost exposures of the BolrsquoshoyLyakhovsky Island (Expedition 1999) site 1TZ AlfredWegener Institute ndash Research Unit Potsdam PANGAEAhttpsdoiorg101594PANGAEA880929 2017c

Schirrmeister L Grosse G Kunitsky V V and SiegertC Sedimentological biogeochemical and geochrono-logical data from permafrost exposures of the BolrsquoshoyLyakhovsky Island (Expedition 1999) site 3TZ AlfredWegener Institute ndash Research Unit Potsdam PANGAEAhttpsdoiorg101594PANGAEA880931 2017d

Schirrmeister L Grosse G Kunitsky V V and SiegertC Sedimentological biogeochemical and geochrono-logical data from permafrost exposures of the BolrsquoshoyLyakhovsky Island (Expedition 1999) site R8+50 AlfredWegener Institute ndash Research Unit Potsdam PANGAEAhttpsdoiorg101594PANGAEA880951 2017e

Schirrmeister L Grosse G Kunitsky V V and SiegertC Sedimentological biogeochemical and geochronologi-cal data from permafrost deposit Kurungnakh PANGAEAhttpsdoiorg101594PANGAEA884069 2017f

Schirrmeister L Bobrov A A Raschke E and WetterichS Sediment ground ice geochronological and paleoecologi-cal data from polygon cores in the Siberian Arctic PANGAEAhttpsdoiorg101594PANGAEA887933 2018a

Schirrmeister L Bobrov A A Raschke E and Wetterich SSediment data from polygon core in the Siberian Arctic Al-fred Wegener Institute ndash Research Unit Potsdam PANGAEAhttpsdoiorg101594PANGAEA887931 2018b

Schwamborn G Schirrmeister L Fruumltsch F and Diekmann BQuartz weathering in freeze-thaw cycles experiment and appli-cation to the Elrsquogygytgyn Crater lake record for tracing Siberian



permafrost history Geografiska annaler Series A Phys Geogr94 481ndash499 httpsdoiorg101111j1468-0459201200472x2012

Sher A Yedoma as a store of paleoenvironmental records inBeringida in Beringian Paleoenvironmental Workshop (ab-stracts and program) edited by Elias S and Brigham-GretteJ Ohana Productions Nepean ON Canada 92ndash94 1997

Sher A V Kuzmina S A Kuznetsova T V and Sulerzhit-sky L D New insights into the Weichselian environment andclimate of the East Siberian Arctic derived from fossil in-sects plants and mammals Quaternary Sci Rev 24 533ndash569httpsdoiorg101016jquascirev200409007 2005

Siegert C Schirrmeister L and Babiy O The sedimen-tological mineralogical and geochemical composition oflate Pleistocene deposits from the ice complex on theBykovsky peninsula northern Siberia Polarforschung 70 3ndash11httpsdoiorg102312polarforschung703 2002

Stauch G Ijmkera J Poumltsch S Zhao H Hilgers ADiekmann B Dietze E Hartmann K Opitz S Wuumln-nemann B and Lehmkuhl F Aeolian sediments on thenorth-eastern Tibetan Plateau Quaternary Sci Rev 57 71ndash74httpsdoiorg101016jquascirev201210001 2012

Strauss J Schirrmeister L Wetterich S Borchers A and Davy-dov S P Grain-size properties and organic-carbon stock ofYedoma Ice Complex permafrost from the Kolyma lowlandnortheastern Siberia Global Biogeochem Cyclesbdquo 26 GB3003httpsdoiorg1010292011GB004104 2012

Strauss J Schirrmeister L Grosse G Wetterich S Ul-rich M Herzschuh U and Hubberten H-W Thedeep permafrost carbon pool of the Yedoma region inSiberia and Alaska Geophys Res Lett 40 6165ndash6170httpsdoiorg1010022013GL058088 2013

Strauss J Schirrmeister L Grosse G Fortier D Hugelius GKnoblauch C Romanovsky V Schaumldel C Schneider vonDeimling T Schuur EAG Shmelev D Ulrich M and Vere-meeva A Deep Yedoma permafrost A synthesis of depositionalcharacteristics and carbon vulnerability Earth-Sci Rev 17275ndash86 httpsdoiorg101016jearscirev201707007 2017

Sun D Bloemendal J Rea D K Vandenberghe J Jiang FAn Z and Su R Grain-size distribution function of polymodalsediments in hydraulic and aeolian environments and numeri-cal partitioning of the sedimentary components Sediment Geol152 263ndash277 httpsdoiorg101016S0037-0738(02)00082-92002

Suzuki R and Shimodaira H Pvclust an R package for assess-ing the uncertainty in hierarchical clustering Bioinformatics22 1540ndash1542 httpsdoiorg101093bioinformaticsbtl1172006

Tomirdiaro S V Palaeogeography of Beringia and Arctida inAmerican Beginnings The Prehistory and Palaeoecology ofBeringia edited by West C F University of Chicago PressChicago and London 58ndash69 1996

Tsoar H and Pye K Dust transport and the questionof desert loess formation Sedimentology 34 139ndash153httpsdoiorg101111j1365-30911987tb00566x 1987

Ulrich M Grosse G Strauss J and Schirrmeister L Quan-tifying wedge-ice volumes in Yedoma and thermokarstbasin deposits Permafrost Periglac 25 151ndash161httpsdoiorg101002ppp1810 2014

Ulrich M Matthes H Schmidt J Fedorov A N Schirrmeis-ter L Siegert C Schneider B Strauss J and Ziel-hofer C Holocene thermokarst dynamics in CentralYakutia ndash A multi-core and robust grain-size endmem-ber modeling approach Quaternary Sci Rev 218 10ndash33httpsdoiorg101016jquascirev201906010 2019

Vandenberghe J Grain size of fine-grained windblown sedimentA powerful proxy for process identification Earth-Sci Rev 12118ndash30 httpsdoiorg101016jearscirev201303001 2013

Vandenberghe J Sun Y Wang X Abels H A andLiu X Grain-size characterization of reworked fine-grained aeolian deposits Earth-Sci Rev 177 43ndash52httpsdoiorg101016jearscirev201711005 2018

van Huissteden J Vandenberghe J Gibbard P L and Lewin JPeriglacial rivers in The Encyclopedia of Quaternary Science2nd edition edited by Elias A E and Mock C J ElsevierAmsterdam 490ndash499 2013

Viran P A G and Binal A Effects of repeated freezendashthaw cy-cles on physico-mechanical properties of cohesive soils ArabJ Geosci 11 250 httpsdoiorg101007s12517-018-3592-52018

Visher G S Grain size distributions and deposi-tional processes J Sediment Res 39 1074ndash1106httpsdoiorg104236ijg2016712099 1969

Walter Anthony K M Zimov S A Grosse G Jones M C An-thony P M Chapin III F S Finlay J C Mack M C Davy-dov S Frenzel P and Frolking S A shift of thermokarst lakesfrom carbon sources to sinks during the Holocene epoch Nature511 452ndash456 httpsdoiorg101038nature13560 2014

Weltje G J and Prins M A Genetically meaningful decomposi-tion of grain-size distributions Sediment Geol 202 409ndash424httpsdoiorg101016jsedgeo200703007 2007

Woronko B and Pisarska-Jamrozy M Micro-Scale Frost Weath-ering of Sand-Sized Quartz Grains Permafrost Periglac 27109ndash122 httpsdoiorg101002ppp1855 2016


Kurzfassung

Abstract

Introduction

Material and methods

Study region

Analytical methods

Results

Results of classical grain-size analyses

Results of site-specific endmember modeling analyses

Results of regional and Arctic-wide endmember analyses

Results of cluster analysis

Discussion

Interpretation of endmember modeling analyses

Yedoma grain-size endmembers and associated processes

Synthesis

Conclusions

Data availability

Supplement

Author contributions

Competing interests

Acknowledgements

Financial support

Review statement

References


mafrost models and soil carbon budgets are considered Our results show the characteristic propertiesof ice-rich Yedoma deposits in the terrestrial Arctic Characterizing and quantifying site-specific pastdepositional processes is crucial for elucidating and understanding the trajectories of this unique kindof ice-rich permafrost in a warmer future

Kurzfassung Der spaumltpleistozaumlne Yedoma Eiskomplex ist ein eisreicher und organikhaltiger Permafrosttyp der inBeringia weit verbreitet ist Durch den hohen Eisanteil wird der Yedoma Eiskomplex im Zuge desKlimawandels als besonders anfaumlllig fuumlr tiefgreifende Stoumlrungen betrachtet und damit ein potentiellerKipppunkt des Klimasystems Um seine Entstehung die lokalen Eigenschaften und die regionale sed-imentologische Zusammensetzung besser zu verstehen haben wir die Korngroumlszligenverteilung von 771Proben an 23 Yedoma-Standorten in der Arktis zusammengestellt raumlumlich eng zusammenhaumlngendeProbenserien wurden zu 17 Untersuchungsstandorten zusammengefasst Daruumlber hinaus wurden 160Proben aus nicht Yedoma-Ablagerungen von drei Eiskeilpolygon- und Uumlberschwemmungsgebietenals holozaumlne Referenzen untersucht Die multimodalen Korngroumlszligenverteilungen zeigen dass eineVielzahl von Sedimentbildungs- Transport- und Ablagerungsprozessen an der Yedoma-Entstehungbeteiligt waren Um diese Prozesse zu erkennen wurde eine robuste Endmembermodellierungs-analyse (rEMMA) durchgefuumlhrt Neun robuste Endmember (rEM) charakterisieren die Yedoma-Ablagerungen uumlber ganz Beringia Die untersuchten Standorte der Yedoma-Ablagerungen wurdenanschlieszligend mittels Clusteranalyse klassifiziert Die daraus resultierenden vier Cluster umfassenzwei bis fuumlnf Untersuchungsstandorte die unregelmaumlszligig uumlber den Nordosten Sibiriens und Alaskaverteilt sind Die breite Palette von rEMs zeigt dass nicht allein aumlolische Sedimentationsprozessefuumlr die Variabilitaumlt in den Korngroumlszligenverteilungen von Yedoma-Ablagerungen verantwortlich sindVielmehr wird die Sedimentation der Yedoma-Ablagerungen eher durch lokale Bedingungen wieAusgangsgesteine ehemalige Topographie und multiple Transportprozesse gesteuert Das stuumltzt dieHypothese einer polygenetischen Yedoma-Entstehung die alluvialen fluvialen und nival-aumlolischenTransport und Akkumulation in polygonalen Tuumlmpeln und in-situ Frostverwitterung sowie postsedi-mentaumlre Frostverwitterung Solifluktion Kryoturbation und Pedogenese beinhaltet Die charakteristis-che rEM Zusammensetzung der Yedoma Cluster kann auch helfen korngroumlszligenspezifische Parameterbesser in der Kohlenstoffbudgetierung und Permafrostmodellierung zu beruumlcksichtigen Damit traumlgtdie Charakterisierung und Quantifizierung standortspezifischer Ablagerungsprozesse in der Vergan-genheit dazu bei die charakteristischen Eigenschaften eisreicher Yedoma-Ablagerungen in der ter-restrischen Arktis aufzuklaumlren Dies ist entscheidend fuumlr das Verstaumlndnis der Fortentwicklung diesesbesonderen Permafrosttyps in einer waumlrmeren Zukunft

1 Introduction

The formation and distribution of late Pleistocene YedomaIce Complex deposits located in western (Siberia) andeastern (Alaska and northwest Canada) Beringia are stillwidely debated (Table S1 in the Supplement) These perma-nently frozen (permafrost) deposits are of silt- and sand-richorganic-bearing sediments up to tens of meters thick inter-spersed with large syngenetic ice wedges that contain highamounts of excess ground ice making them highly sensitiveto degradation in a warming climate Syngenetic ice wedgesand segregated intrasedimental ice (ice lenses and bands)constitute the largest portion (50 ndash95 ) of this type of de-posit by volume in most Yedoma regions (Kanevskiy et al2011 2016 Strauss et al 2013 Ulrich et al 2014) next toclastic and organic components (Schirrmeister et al 2013)Thus ground ice aggradation is clearly one of the most crit-ical factors in Yedoma Ice Complex genesis (hence also the

name ldquoIce Complexrdquo) and ice forms a main component ofthe entire deposit in clear contrast to the accumulation ofsedimentary deposits in temperate regions where ice does notplay a role either in the formation of deposits or as a struc-tural and stratigraphic component

In terms of depositional and stratigraphic characteristicsvarious Yedoma types seem to exist (Kaplina 1981 Sher etal 2005 Strauss et al 2012 Murton et al 2015) acrossthe area of 14times 106 km2 where Yedoma deposits currentlyoccur (Strauss et al 2017) Because of their carbon storageand high ice content Yedoma deposits have been suggestedas a potential ldquotipping elementrdquo for future climate warming(Lenton 2012) Permafrost models require the parametriza-tion of the types of Yedoma deposits for example in terms oftheir grain-size composition to better constrain factors suchas the hydraulic conductivity and pore space volume wherewater can freeze (DallrsquoAmico et al 2011) The estimation of








21 Study region
















Alaska


Western Laptev Sea


Lena Delta










Yakutian inland



















dij =

sum7k=1





3 Results











2(a)Main

modes

ofrobustgrain-sizeE

Ms


variances(

belowm

ode)andthe


()explained

bythe

averagerobustm

odelforeach

studysite

(seealso

Figs14aandS4)(b)M

ainm


EM

s(microm


(below

mode)and

thetotalgrain-size

variability(

)explained

bythe

averagerobustm

odelfornon-Y

edoma

sitesof

modern

ice-wedge

polygons(see

alsoFigs1S45)(c)

Robustgrain-size

EM

s(in

microm

etersof

main

modebold)

andrespective

explainedvariances

(below

mode)in

regionalandA

rctic-wide

combinations

(seealso

FigS61)

(a)

Locno

Location

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1Total

clayvery

finesilt

finesilt

medium

siltcoarse

siltvery

finesand

finesand

medium

sandcoarse

sandexplained

lt4

microm(4ndash8

microm)

(79ndash119microm

)(12ndash27

microm)

(28ndash50microm

)(51ndash100

microm)

(101ndash280microm

)(281ndash750

microm)

(750ndash865microm

)variability

Alaska

1C

olvilleR

iver39

172330

234182

249278

340133

2Itkillik

River

29172

362634

213273

271260

173147

1483

Seward

Peninsula(K

itlukR

iver)32

157398

5782341

75227

250258

32233

4V

aultCreek

tunnel172

479764

339874

476267

17286

Western

Laptev

Sea

5C

apeM

amontov

Klyk

74300

4791768

786982

409340

71150

31

Lena

Delta

67E

beB

asynSise

andK

hardangSise

islands578

19424096

86213

423153

634211

8K

urungnakhSise

Island39

4792821

786973

347239

276139

Centraland

easternL

aptevSea

9B

ykovskyPeninsula

47250

16123731

69128

249313

31010

Muostakh

Island89

3981612

786974

214115

178384

8639109

11B

uorKhaya

Peninsula32

250839

121876

95320

316269

New

SiberianIslands

andthe

Dm

itryL

aptevStrait



otelnysouthwestern

Kotelnyand

Maly

29108

43768

Lyakhovskyislands

116332

55217

Bolrsquoshoy

LyakhovskyIsland

35108

330696

75339

135254

27218

Oyogos

Yarcoast

35108

2741768

72232

128250

23710

153

Yakutian

inland

19D

uvannyY

ar32

330634

66306

183511

20K

ytalyk62

3982132

54378

269353

21B

atagayM

ega-slump

301634

213276

115200

152839

2568433

1002223

Tabagaand

Yukechi

39362

6961612

69124

448167

260



Tabl

e2

Con

tinue

d

(b)

Loc

no

Loc

atio

nrE

M9

rEM

8rE

M7

rEM

6rE

M5

rEM

4rE

M3

rEM

2rE

M1

Tota

lcl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndex

plai

ned

lt4

microm(4

ndash8microm

)(7

9ndash1

19

microm)

(12ndash

27microm

)(2

8ndash50

microm)

(51ndash

100

microm)

(101

ndash280

microm)

(281

ndash750

microm)

(750

ndash865

microm)

vari

abili

ty

Non

-Yed

oma

site

s(m

oder

nic

e-w

edge

poly

gons

)

Pokh

odsk

poly

gon

core

s3

917

263

413

37

7934

68

823

033

5Po

khod

skpo

lygo

nbo

ttom

32

250

121

871

69

7438

216

614

530

7K

ytal

ykpo

lygo

nco

res

32

156

398

194

270

168

309

351

171

Kyt

alyk

poly

gon

botto

m3

218

969

630

96

7326

413

412

647

5K

olym

aan

dB

erel

ekh

flood

plai

ns4

247

917

68

7944

239

516

3

(c)

Reg

ion

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1cl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndlt

4microm

(4ndash8

microm)

(79

ndash11

9microm

)(1

2ndash27

)microm

(28ndash

50microm

)(5

1ndash10

0microm

)(1

01ndash2

80microm

)(2

81ndash7

50microm

)(7

50ndash8

65microm

)

Ala

ska

(site

nos

1ndash4)

35

172

437

282

178

69

289

170

355

130

56

Lap

tev

Sea

and

Eas

tSi

beri

anSe

aco

asts

in-

clud

ing

the

Len

aD

elta

(site

nos

5ndash18

)

35

300

111

086

39

194

346

153

107

213

220

0

Yak

utia

nin

land

(site

nos

19ndash2

3)25

047

992

125

68

386

212

396

07

Arc

tic-w

ide

35

330

921

234

178

69

220

346

209

158

68















Figure 4















4 Discussion





















43 Synthesis











5 Conclusions

















References






























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References







21 Study region
















Alaska


Western Laptev Sea


Lena Delta










Yakutian inland



















dij =

sum7k=1





3 Results











2(a)Main

modes

ofrobustgrain-sizeE

Ms


variances(

belowm

ode)andthe


()explained

bythe

averagerobustm

odelforeach

studysite

(seealso

Figs14aandS4)(b)M

ainm


EM

s(microm


(below

mode)and

thetotalgrain-size

variability(

)explained

bythe

averagerobustm

odelfornon-Y

edoma

sitesof

modern

ice-wedge

polygons(see

alsoFigs1S45)(c)

Robustgrain-size

EM

s(in

microm

etersof

main

modebold)

andrespective

explainedvariances

(below

mode)in

regionalandA

rctic-wide

combinations

(seealso

FigS61)

(a)

Locno

Location

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1Total

clayvery

finesilt

finesilt

medium

siltcoarse

siltvery

finesand

finesand

medium

sandcoarse

sandexplained

lt4

microm(4ndash8

microm)

(79ndash119microm

)(12ndash27

microm)

(28ndash50microm

)(51ndash100

microm)

(101ndash280microm

)(281ndash750

microm)

(750ndash865microm

)variability

Alaska

1C

olvilleR

iver39

172330

234182

249278

340133

2Itkillik

River

29172

362634

213273

271260

173147

1483

Seward

Peninsula(K

itlukR

iver)32

157398

5782341

75227

250258

32233

4V

aultCreek

tunnel172

479764

339874

476267

17286

Western

Laptev

Sea

5C

apeM

amontov

Klyk

74300

4791768

786982

409340

71150

31

Lena

Delta

67E

beB

asynSise

andK

hardangSise

islands578

19424096

86213

423153

634211

8K

urungnakhSise

Island39

4792821

786973

347239

276139

Centraland

easternL

aptevSea

9B

ykovskyPeninsula

47250

16123731

69128

249313

31010

Muostakh

Island89

3981612

786974

214115

178384

8639109

11B

uorKhaya

Peninsula32

250839

121876

95320

316269

New

SiberianIslands

andthe

Dm

itryL

aptevStrait



otelnysouthwestern

Kotelnyand

Maly

29108

43768

Lyakhovskyislands

116332

55217

Bolrsquoshoy

LyakhovskyIsland

35108

330696

75339

135254

27218

Oyogos

Yarcoast

35108

2741768

72232

128250

23710

153

Yakutian

inland

19D

uvannyY

ar32

330634

66306

183511

20K

ytalyk62

3982132

54378

269353

21B

atagayM

ega-slump

301634

213276

115200

152839

2568433

1002223

Tabagaand

Yukechi

39362

6961612

69124

448167

260



Tabl

e2

Con

tinue

d

(b)

Loc

no

Loc

atio

nrE

M9

rEM

8rE

M7

rEM

6rE

M5

rEM

4rE

M3

rEM

2rE

M1

Tota

lcl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndex

plai

ned

lt4

microm(4

ndash8microm

)(7

9ndash1

19

microm)

(12ndash

27microm

)(2

8ndash50

microm)

(51ndash

100

microm)

(101

ndash280

microm)

(281

ndash750

microm)

(750

ndash865

microm)

vari

abili

ty

Non

-Yed

oma

site

s(m

oder

nic

e-w

edge

poly

gons

)

Pokh

odsk

poly

gon

core

s3

917

263

413

37

7934

68

823

033

5Po

khod

skpo

lygo

nbo

ttom

32

250

121

871

69

7438

216

614

530

7K

ytal

ykpo

lygo

nco

res

32

156

398

194

270

168

309

351

171

Kyt

alyk

poly

gon

botto

m3

218

969

630

96

7326

413

412

647

5K

olym

aan

dB

erel

ekh

flood

plai

ns4

247

917

68

7944

239

516

3

(c)

Reg

ion

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1cl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndlt

4microm

(4ndash8

microm)

(79

ndash11

9microm

)(1

2ndash27

)microm

(28ndash

50microm

)(5

1ndash10

0microm

)(1

01ndash2

80microm

)(2

81ndash7

50microm

)(7

50ndash8

65microm

)

Ala

ska

(site

nos

1ndash4)

35

172

437

282

178

69

289

170

355

130

56

Lap

tev

Sea

and

Eas

tSi

beri

anSe

aco

asts

in-

clud

ing

the

Len

aD

elta

(site

nos

5ndash18

)

35

300

111

086

39

194

346

153

107

213

220

0

Yak

utia

nin

land

(site

nos

19ndash2

3)25

047

992

125

68

386

212

396

07

Arc

tic-w

ide

35

330

921

234

178

69

220

346

209

158

68















Figure 4















4 Discussion





















43 Synthesis











5 Conclusions

















References






























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References











Alaska


Western Laptev Sea


Lena Delta










Yakutian inland



















dij =

sum7k=1





3 Results











2(a)Main

modes

ofrobustgrain-sizeE

Ms


variances(

belowm

ode)andthe


()explained

bythe

averagerobustm

odelforeach

studysite

(seealso

Figs14aandS4)(b)M

ainm


EM

s(microm


(below

mode)and

thetotalgrain-size

variability(

)explained

bythe

averagerobustm

odelfornon-Y

edoma

sitesof

modern

ice-wedge

polygons(see

alsoFigs1S45)(c)

Robustgrain-size

EM

s(in

microm

etersof

main

modebold)

andrespective

explainedvariances

(below

mode)in

regionalandA

rctic-wide

combinations

(seealso

FigS61)

(a)

Locno

Location

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1Total

clayvery

finesilt

finesilt

medium

siltcoarse

siltvery

finesand

finesand

medium

sandcoarse

sandexplained

lt4

microm(4ndash8

microm)

(79ndash119microm

)(12ndash27

microm)

(28ndash50microm

)(51ndash100

microm)

(101ndash280microm

)(281ndash750

microm)

(750ndash865microm

)variability

Alaska

1C

olvilleR

iver39

172330

234182

249278

340133

2Itkillik

River

29172

362634

213273

271260

173147

1483

Seward

Peninsula(K

itlukR

iver)32

157398

5782341

75227

250258

32233

4V

aultCreek

tunnel172

479764

339874

476267

17286

Western

Laptev

Sea

5C

apeM

amontov

Klyk

74300

4791768

786982

409340

71150

31

Lena

Delta

67E

beB

asynSise

andK

hardangSise

islands578

19424096

86213

423153

634211

8K

urungnakhSise

Island39

4792821

786973

347239

276139

Centraland

easternL

aptevSea

9B

ykovskyPeninsula

47250

16123731

69128

249313

31010

Muostakh

Island89

3981612

786974

214115

178384

8639109

11B

uorKhaya

Peninsula32

250839

121876

95320

316269

New

SiberianIslands

andthe

Dm

itryL

aptevStrait



otelnysouthwestern

Kotelnyand

Maly

29108

43768

Lyakhovskyislands

116332

55217

Bolrsquoshoy

LyakhovskyIsland

35108

330696

75339

135254

27218

Oyogos

Yarcoast

35108

2741768

72232

128250

23710

153

Yakutian

inland

19D

uvannyY

ar32

330634

66306

183511

20K

ytalyk62

3982132

54378

269353

21B

atagayM

ega-slump

301634

213276

115200

152839

2568433

1002223

Tabagaand

Yukechi

39362

6961612

69124

448167

260



Tabl

e2

Con

tinue

d

(b)

Loc

no

Loc

atio

nrE

M9

rEM

8rE

M7

rEM

6rE

M5

rEM

4rE

M3

rEM

2rE

M1

Tota

lcl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndex

plai

ned

lt4

microm(4

ndash8microm

)(7

9ndash1

19

microm)

(12ndash

27microm

)(2

8ndash50

microm)

(51ndash

100

microm)

(101

ndash280

microm)

(281

ndash750

microm)

(750

ndash865

microm)

vari

abili

ty

Non

-Yed

oma

site

s(m

oder

nic

e-w

edge

poly

gons

)

Pokh

odsk

poly

gon

core

s3

917

263

413

37

7934

68

823

033

5Po

khod

skpo

lygo

nbo

ttom

32

250

121

871

69

7438

216

614

530

7K

ytal

ykpo

lygo

nco

res

32

156

398

194

270

168

309

351

171

Kyt

alyk

poly

gon

botto

m3

218

969

630

96

7326

413

412

647

5K

olym

aan

dB

erel

ekh

flood

plai

ns4

247

917

68

7944

239

516

3

(c)

Reg

ion

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1cl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndlt

4microm

(4ndash8

microm)

(79

ndash11

9microm

)(1

2ndash27

)microm

(28ndash

50microm

)(5

1ndash10

0microm

)(1

01ndash2

80microm

)(2

81ndash7

50microm

)(7

50ndash8

65microm

)

Ala

ska

(site

nos

1ndash4)

35

172

437

282

178

69

289

170

355

130

56

Lap

tev

Sea

and

Eas

tSi

beri

anSe

aco

asts

in-

clud

ing

the

Len

aD

elta

(site

nos

5ndash18

)

35

300

111

086

39

194

346

153

107

213

220

0

Yak

utia

nin

land

(site

nos

19ndash2

3)25

047

992

125

68

386

212

396

07

Arc

tic-w

ide

35

330

921

234

178

69

220

346

209

158

68















Figure 4















4 Discussion





















43 Synthesis











5 Conclusions

















References






























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References












dij =

sum7k=1





3 Results











2(a)Main

modes

ofrobustgrain-sizeE

Ms


variances(

belowm

ode)andthe


()explained

bythe

averagerobustm

odelforeach

studysite

(seealso

Figs14aandS4)(b)M

ainm


EM

s(microm


(below

mode)and

thetotalgrain-size

variability(

)explained

bythe

averagerobustm

odelfornon-Y

edoma

sitesof

modern

ice-wedge

polygons(see

alsoFigs1S45)(c)

Robustgrain-size

EM

s(in

microm

etersof

main

modebold)

andrespective

explainedvariances

(below

mode)in

regionalandA

rctic-wide

combinations

(seealso

FigS61)

(a)

Locno

Location

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1Total

clayvery

finesilt

finesilt

medium

siltcoarse

siltvery

finesand

finesand

medium

sandcoarse

sandexplained

lt4

microm(4ndash8

microm)

(79ndash119microm

)(12ndash27

microm)

(28ndash50microm

)(51ndash100

microm)

(101ndash280microm

)(281ndash750

microm)

(750ndash865microm

)variability

Alaska

1C

olvilleR

iver39

172330

234182

249278

340133

2Itkillik

River

29172

362634

213273

271260

173147

1483

Seward

Peninsula(K

itlukR

iver)32

157398

5782341

75227

250258

32233

4V

aultCreek

tunnel172

479764

339874

476267

17286

Western

Laptev

Sea

5C

apeM

amontov

Klyk

74300

4791768

786982

409340

71150

31

Lena

Delta

67E

beB

asynSise

andK

hardangSise

islands578

19424096

86213

423153

634211

8K

urungnakhSise

Island39

4792821

786973

347239

276139

Centraland

easternL

aptevSea

9B

ykovskyPeninsula

47250

16123731

69128

249313

31010

Muostakh

Island89

3981612

786974

214115

178384

8639109

11B

uorKhaya

Peninsula32

250839

121876

95320

316269

New

SiberianIslands

andthe

Dm

itryL

aptevStrait



otelnysouthwestern

Kotelnyand

Maly

29108

43768

Lyakhovskyislands

116332

55217

Bolrsquoshoy

LyakhovskyIsland

35108

330696

75339

135254

27218

Oyogos

Yarcoast

35108

2741768

72232

128250

23710

153

Yakutian

inland

19D

uvannyY

ar32

330634

66306

183511

20K

ytalyk62

3982132

54378

269353

21B

atagayM

ega-slump

301634

213276

115200

152839

2568433

1002223

Tabagaand

Yukechi

39362

6961612

69124

448167

260



Tabl

e2

Con

tinue

d

(b)

Loc

no

Loc

atio

nrE

M9

rEM

8rE

M7

rEM

6rE

M5

rEM

4rE

M3

rEM

2rE

M1

Tota

lcl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndex

plai

ned

lt4

microm(4

ndash8microm

)(7

9ndash1

19

microm)

(12ndash

27microm

)(2

8ndash50

microm)

(51ndash

100

microm)

(101

ndash280

microm)

(281

ndash750

microm)

(750

ndash865

microm)

vari

abili

ty

Non

-Yed

oma

site

s(m

oder

nic

e-w

edge

poly

gons

)

Pokh

odsk

poly

gon

core

s3

917

263

413

37

7934

68

823

033

5Po

khod

skpo

lygo

nbo

ttom

32

250

121

871

69

7438

216

614

530

7K

ytal

ykpo

lygo

nco

res

32

156

398

194

270

168

309

351

171

Kyt

alyk

poly

gon

botto

m3

218

969

630

96

7326

413

412

647

5K

olym

aan

dB

erel

ekh

flood

plai

ns4

247

917

68

7944

239

516

3

(c)

Reg

ion

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1cl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndlt

4microm

(4ndash8

microm)

(79

ndash11

9microm

)(1

2ndash27

)microm

(28ndash

50microm

)(5

1ndash10

0microm

)(1

01ndash2

80microm

)(2

81ndash7

50microm

)(7

50ndash8

65microm

)

Ala

ska

(site

nos

1ndash4)

35

172

437

282

178

69

289

170

355

130

56

Lap

tev

Sea

and

Eas

tSi

beri

anSe

aco

asts

in-

clud

ing

the

Len

aD

elta

(site

nos

5ndash18

)

35

300

111

086

39

194

346

153

107

213

220

0

Yak

utia

nin

land

(site

nos

19ndash2

3)25

047

992

125

68

386

212

396

07

Arc

tic-w

ide

35

330

921

234

178

69

220

346

209

158

68















Figure 4















4 Discussion





















43 Synthesis











5 Conclusions

















References






























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References


2(a)Main

modes

ofrobustgrain-sizeE

Ms


variances(

belowm

ode)andthe


()explained

bythe

averagerobustm

odelforeach

studysite

(seealso

Figs14aandS4)(b)M

ainm


EM

s(microm


(below

mode)and

thetotalgrain-size

variability(

)explained

bythe

averagerobustm

odelfornon-Y

edoma

sitesof

modern

ice-wedge

polygons(see

alsoFigs1S45)(c)

Robustgrain-size

EM

s(in

microm

etersof

main

modebold)

andrespective

explainedvariances

(below

mode)in

regionalandA

rctic-wide

combinations

(seealso

FigS61)

(a)

Locno

Location

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1Total

clayvery

finesilt

finesilt

medium

siltcoarse

siltvery

finesand

finesand

medium

sandcoarse

sandexplained

lt4

microm(4ndash8

microm)

(79ndash119microm

)(12ndash27

microm)

(28ndash50microm

)(51ndash100

microm)

(101ndash280microm

)(281ndash750

microm)

(750ndash865microm

)variability

Alaska

1C

olvilleR

iver39

172330

234182

249278

340133

2Itkillik

River

29172

362634

213273

271260

173147

1483

Seward

Peninsula(K

itlukR

iver)32

157398

5782341

75227

250258

32233

4V

aultCreek

tunnel172

479764

339874

476267

17286

Western

Laptev

Sea

5C

apeM

amontov

Klyk

74300

4791768

786982

409340

71150

31

Lena

Delta

67E

beB

asynSise

andK

hardangSise

islands578

19424096

86213

423153

634211

8K

urungnakhSise

Island39

4792821

786973

347239

276139

Centraland

easternL

aptevSea

9B

ykovskyPeninsula

47250

16123731

69128

249313

31010

Muostakh

Island89

3981612

786974

214115

178384

8639109

11B

uorKhaya

Peninsula32

250839

121876

95320

316269

New

SiberianIslands

andthe

Dm

itryL

aptevStrait



otelnysouthwestern

Kotelnyand

Maly

29108

43768

Lyakhovskyislands

116332

55217

Bolrsquoshoy

LyakhovskyIsland

35108

330696

75339

135254

27218

Oyogos

Yarcoast

35108

2741768

72232

128250

23710

153

Yakutian

inland

19D

uvannyY

ar32

330634

66306

183511

20K

ytalyk62

3982132

54378

269353

21B

atagayM

ega-slump

301634

213276

115200

152839

2568433

1002223

Tabagaand

Yukechi

39362

6961612

69124

448167

260



Tabl

e2

Con

tinue

d

(b)

Loc

no

Loc

atio

nrE

M9

rEM

8rE

M7

rEM

6rE

M5

rEM

4rE

M3

rEM

2rE

M1

Tota

lcl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndex

plai

ned

lt4

microm(4

ndash8microm

)(7

9ndash1

19

microm)

(12ndash

27microm

)(2

8ndash50

microm)

(51ndash

100

microm)

(101

ndash280

microm)

(281

ndash750

microm)

(750

ndash865

microm)

vari

abili

ty

Non

-Yed

oma

site

s(m

oder

nic

e-w

edge

poly

gons

)

Pokh

odsk

poly

gon

core

s3

917

263

413

37

7934

68

823

033

5Po

khod

skpo

lygo

nbo

ttom

32

250

121

871

69

7438

216

614

530

7K

ytal

ykpo

lygo

nco

res

32

156

398

194

270

168

309

351

171

Kyt

alyk

poly

gon

botto

m3

218

969

630

96

7326

413

412

647

5K

olym

aan

dB

erel

ekh

flood

plai

ns4

247

917

68

7944

239

516

3

(c)

Reg

ion

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1cl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndlt

4microm

(4ndash8

microm)

(79

ndash11

9microm

)(1

2ndash27

)microm

(28ndash

50microm

)(5

1ndash10

0microm

)(1

01ndash2

80microm

)(2

81ndash7

50microm

)(7

50ndash8

65microm

)

Ala

ska

(site

nos

1ndash4)

35

172

437

282

178

69

289

170

355

130

56

Lap

tev

Sea

and

Eas

tSi

beri

anSe

aco

asts

in-

clud

ing

the

Len

aD

elta

(site

nos

5ndash18

)

35

300

111

086

39

194

346

153

107

213

220

0

Yak

utia

nin

land

(site

nos

19ndash2

3)25

047

992

125

68

386

212

396

07

Arc

tic-w

ide

35

330

921

234

178

69

220

346

209

158

68















Figure 4















4 Discussion





















43 Synthesis











5 Conclusions

















References






























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References


Tabl

e2

Con

tinue

d

(b)

Loc

no

Loc

atio

nrE

M9

rEM

8rE

M7

rEM

6rE

M5

rEM

4rE

M3

rEM

2rE

M1

Tota

lcl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndex

plai

ned

lt4

microm(4

ndash8microm

)(7

9ndash1

19

microm)

(12ndash

27microm

)(2

8ndash50

microm)

(51ndash

100

microm)

(101

ndash280

microm)

(281

ndash750

microm)

(750

ndash865

microm)

vari

abili

ty

Non

-Yed

oma

site

s(m

oder

nic

e-w

edge

poly

gons

)

Pokh

odsk

poly

gon

core

s3

917

263

413

37

7934

68

823

033

5Po

khod

skpo

lygo

nbo

ttom

32

250

121

871

69

7438

216

614

530

7K

ytal

ykpo

lygo

nco

res

32

156

398

194

270

168

309

351

171

Kyt

alyk

poly

gon

botto

m3

218

969

630

96

7326

413

412

647

5K

olym

aan

dB

erel

ekh

flood

plai

ns4

247

917

68

7944

239

516

3

(c)

Reg

ion

rEM

9rE

M8

rEM

7rE

M6

rEM

5rE

M4

rEM

3rE

M2

rEM

1cl

ayve

ryfin

esi

ltfin

esi

ltm

ediu

msi

ltco

arse

silt

very

fine

sand

fine

sand

med

ium

sand

coar

sesa

ndlt

4microm

(4ndash8

microm)

(79

ndash11

9microm

)(1

2ndash27

)microm

(28ndash

50microm

)(5

1ndash10

0microm

)(1

01ndash2

80microm

)(2

81ndash7

50microm

)(7

50ndash8

65microm

)

Ala

ska

(site

nos

1ndash4)

35

172

437

282

178

69

289

170

355

130

56

Lap

tev

Sea

and

Eas

tSi

beri

anSe

aco

asts

in-

clud

ing

the

Len

aD

elta

(site

nos

5ndash18

)

35

300

111

086

39

194

346

153

107

213

220

0

Yak

utia

nin

land

(site

nos

19ndash2

3)25

047

992

125

68

386

212

396

07

Arc

tic-w

ide

35

330

921

234

178

69

220

346

209

158

68















Figure 4















4 Discussion





















43 Synthesis











5 Conclusions

















References






























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References














Figure 4















4 Discussion





















43 Synthesis











5 Conclusions

















References






























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References














4 Discussion





















43 Synthesis











5 Conclusions

















References






























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References












43 Synthesis











5 Conclusions

















References






























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References





5 Conclusions

















References






























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References













References






























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References























































































Kurzfassung

Abstract

Introduction


Study region

Analytical methods

Results





Discussion



Synthesis

Conclusions

Data availability

Supplement


Competing interests

Acknowledgements

Financial support

Review statement

References

the genesis of yedoma ice complex permafrost – grain-size

Documents