Stratigraphy, optical dating chronology (IRSL) and depositional model of pre-LGM glacial deposits in the Hope Valley, New Zealand Henrik Rother a, * , James Shulmeister b , Uwe Rieser c a Scottish Universities Environmental Research Centre (SUERC), Rankine Avenue, East Kilbride, G75 0QF Scotland, UK b School of Geography, Planning and Environmental Management, University of Queensland, St Lucia 4072, Queensland, Australia c Luminescence Dating Laboratory, Victoria University of Wellington, New Zealand article info Article history: Received 19 November 2007 Received in revised form 29 October 2009 Accepted 1 November 2009 abstract A 110m thick succession of glacial valley fill is described from Poplars Gully, central South Island, New Zealand. The section consists of eight lithofacies assemblages that represent different stages of ice occupation in the valley. Basal sediments record an ice retreat phase followed by a glacial re-advance which deposited mass flow diamictons and till. A subsequent ice retreat from the site is indicated by the stratigraphic transition from till to thick glacio-fluvial gravels. This is followed by a probably short-lived glacier re-advance that caused folding and thrusting of proglacial sediments. Final glacial retreat from the valley led to the formation of a large proglacial lake. In total, Poplars Gully holds evidence for two major ice advances, separated by a glacial retreat that resulted in complete ice evacuation from the lower Hope Valley. Infrared stimulated luminescence (IRSL) dating on ice-proximal sediments from Poplars Gully yielded six ages between 181 and 115ka BP. Our stratigraphic logging and dating results show that the fill sequence was not, as previously thought, deposited in association with ice advances during the Last Glacial Maximum (LGM) nor indeed during the last glacial cycle. LGM glaciers later overran the fill but we find that the older glacial sequences are considerably more voluminous than those deposited during the last glacial cycle. We also show that the mid-Pleistocene glaciers carved a much deeper valley trough than did glaciers during the LGM. Taken together these features are likely to reflect a significant difference in the magnitude of successive Pleistocene glaciations in the valley, with the mid-Pleistocene ice advances having been considerably larger than those of the last glacial cycle. The recognition of the in-situ survival of extensive pre-MIS 5 (Marine Isotope Stage) deposits in valley troughs that were later occupied by LGM glaciers represents a new feature in the Quaternary stratigraphy of the Southern Alps. The results demonstrate that New Zealand’s commonly very large soft-sedimentary valley fills provide a valuable, yet largely unexploited, terrestrial sedimentary archive of successive glaciations in the region. Ó 2009 Elsevier Ltd. All rights reserved. 1. Introduction During the Pleistocene mountain glaciers in New Zealand repeatedly advanced beyond the limits of the Southern Alps reaching the coast of the Tasman Sea in the west and extending into the eastern alpine forelands (Suggate, 1990). Glacial sediments and landforms left behind cover extensive areas in New Zealand and have in recent years gained recognition as important mid-latitude Southern Hemisphere records for the interhemispheric correlation of Late Quaternary glacial events (Denton and Hendy, 1994; Broecker, 1997; Ivy-Ochs et al., 1999; Turney et al., 2003; Shulmeister et al., 2005; Barrows et al., 2007). However, major uncertainties regarding timing of glaciations and underlying climatic forcing mechanisms still exist even within different parts of the Southern Hemisphere (e.g. Kaplan et al., 2004; Glasser et al., 2006; Schaefer et al., 2006). The geographic setting of the Southern Alps provides a case where glaciations have occurred under dominantly perhumid conditions. This is due to the orographic interception of moist westerly air masses arriving from the Tasman Sea, which commonly delivers >10,000 mm annual precipitation into the central Alps (Henderson and Thompson, 1999, Fig. 1). As a result New Zealand’s past and present temperate-maritime glacier systems are characterized by a very high ice-flux and glacier tongues that descend to low elevations above sea level. Super- imposed on this glacial pattern is the region’s tectonic setting along * Corresponding author. Scottish Universities Environmental Research Centre (SUERC), Rankine Avenue, East Kilbride, G75 0QF Scotland, UK. E-mail address: [email protected](H. Rother). Contents lists available at ScienceDirect Quaternary Science Reviews journal homepage: www.elsevier.com/locate/quascirev 0277-3791/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.quascirev.2009.11.001 Quaternary Science Reviews 29 (2010) 576–592
Stratigraphy, optical dating chronology (IRSL) and depositional modelof pre-LGM glacial deposits in the Hope Valley, New Zealand
Henrik Rother a,*, James Shulmeister b, Uwe Rieser c
a Scottish Universities Environmental Research Centre (SUERC), Rankine Avenue, East Kilbride, G75 0QF Scotland, UKb School of Geography, Planning and Environmental Management, University of Queensland, St Lucia 4072, Queensland, Australiac Luminescence Dating Laboratory, Victoria University of Wellington, New Zealand
a r t i c l e i n f o
Article history:Received 19 November 2007Received in revised form29 October 2009Accepted 1 November 2009
0277-3791/$ – see front matter � 2009 Elsevier Ltd.doi:10.1016/j.quascirev.2009.11.001
a b s t r a c t
A 110 m thick succession of glacial valley fill is described from Poplars Gully, central South Island, NewZealand. The section consists of eight lithofacies assemblages that represent different stages of iceoccupation in the valley. Basal sediments record an ice retreat phase followed by a glacial re-advancewhich deposited mass flow diamictons and till. A subsequent ice retreat from the site is indicated by thestratigraphic transition from till to thick glacio-fluvial gravels. This is followed by a probably short-livedglacier re-advance that caused folding and thrusting of proglacial sediments. Final glacial retreat fromthe valley led to the formation of a large proglacial lake. In total, Poplars Gully holds evidence for twomajor ice advances, separated by a glacial retreat that resulted in complete ice evacuation from the lowerHope Valley.
Infrared stimulated luminescence (IRSL) dating on ice-proximal sediments from Poplars Gully yieldedsix ages between 181 and 115 ka BP. Our stratigraphic logging and dating results show that the fillsequence was not, as previously thought, deposited in association with ice advances during the LastGlacial Maximum (LGM) nor indeed during the last glacial cycle. LGM glaciers later overran the fill butwe find that the older glacial sequences are considerably more voluminous than those deposited duringthe last glacial cycle. We also show that the mid-Pleistocene glaciers carved a much deeper valley troughthan did glaciers during the LGM. Taken together these features are likely to reflect a significantdifference in the magnitude of successive Pleistocene glaciations in the valley, with the mid-Pleistoceneice advances having been considerably larger than those of the last glacial cycle. The recognition of thein-situ survival of extensive pre-MIS 5 (Marine Isotope Stage) deposits in valley troughs that were lateroccupied by LGM glaciers represents a new feature in the Quaternary stratigraphy of the Southern Alps.The results demonstrate that New Zealand’s commonly very large soft-sedimentary valley fills providea valuable, yet largely unexploited, terrestrial sedimentary archive of successive glaciations in the region.
� 2009 Elsevier Ltd. All rights reserved.
1. Introduction
During the Pleistocene mountain glaciers in New Zealandrepeatedly advanced beyond the limits of the Southern Alpsreaching the coast of the Tasman Sea in the west and extending intothe eastern alpine forelands (Suggate, 1990). Glacial sediments andlandforms left behind cover extensive areas in New Zealand andhave in recent years gained recognition as important mid-latitudeSouthern Hemisphere records for the interhemispheric correlationof Late Quaternary glacial events (Denton and Hendy, 1994;Broecker, 1997; Ivy-Ochs et al., 1999; Turney et al., 2003;
ironmental Research Centretland, UK.er).
All rights reserved.
Shulmeister et al., 2005; Barrows et al., 2007). However, majoruncertainties regarding timing of glaciations and underlyingclimatic forcing mechanisms still exist even within different partsof the Southern Hemisphere (e.g. Kaplan et al., 2004; Glasser et al.,2006; Schaefer et al., 2006).
The geographic setting of the Southern Alps provides a casewhere glaciations have occurred under dominantly perhumidconditions. This is due to the orographic interception of moistwesterly air masses arriving from the Tasman Sea, whichcommonly delivers >10,000 mm annual precipitation into thecentral Alps (Henderson and Thompson, 1999, Fig. 1). As a resultNew Zealand’s past and present temperate-maritime glaciersystems are characterized by a very high ice-flux and glaciertongues that descend to low elevations above sea level. Super-imposed on this glacial pattern is the region’s tectonic setting along
Fig. 1. New Zealand’s climatic and glacial setting. Rapid tectonic uplift since the Pliocene formed the up to 3.7 km high Southern Alps along the plate margin of the Indo-Australianand Pacific Plates. The mountain range constitutes a large orographic barrier in the pathway of the mid-latitude Westerlies resulting in very high precipitation yields, particularly inthe central part of the Alps. The sketch on the right shows New Zealand’s approximate last glacial ice extent (LGM), the glacial shoreline and the location of the study area.
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592 577
an active plate boundary, which is associated with very high ratesof uplift, erosion and sediment supply (Andrews,1973; Kamp, 1986;Norris et al., 1990).
The basic framework for discerning Late-Pleistocene mountainglaciations in the Southern Alps was established during the period1958–1970. Work during this pioneering phase focussed on themapping of moraine positions and the geomorphological differ-entiation of generations of glacial landform associations (e.g. Gage,1958; Soons, 1963; Speight, 1963; Suggate, 1965; Clayton, 1968).Based on limited radiocarbon dating and various relative datingtechniques a general chronology for Quaternary glaciations wasformulated (e.g. Suggate, 1990), but major uncertainties remainregarding the absolute timings of most Quaternary glacial events.
A commonly noted feature in the Southern Alps is the presenceof very voluminous valley fills. Across the Alps postglacial fluvialdowncutting produced spectacular flights of terraces within thesefills that often feature impressively large sedimentary exposures.Despite this, comparatively little work has been undertaken onthese valley fills and glacial-depositional processes in the regionalsetting are poorly understood. In this paper we present strati-graphic and geochronological data from glacial sediments in theHope Valley of North Canterbury in the South Island. Our inter-pretation lays emphasis on reconstructing depositional style,sequence architecture and geochronology of past ice advances andassociated phases of valley aggradation.
1.1. Quaternary geology and study site
The Hope River is a tributary to the Waiau River, which drainsroughly 2000 km2 of the Southern Alps eastwards into the PacificOcean. The investigated valley portion west of Hanmer Basin is part
of the active Marlborough Fault Zone that comprises a system ofdominantly NE trending dextral strike-slip faults with averageannual Quaternary slip rates in the order of 5–15 mm (Cowan,1990;Wood et al., 1994). Present glaciation in the area is limited to a smallnumber of cirque glaciers in the highest part of the catchment(w2300 m a.s.l.), but during the Pleistocene large valley glaciersextended 30–40 km from the headwaters. Glacial landforms in thearea were first systematically investigated by Clayton (1968) whodifferentiated six Mid–Late-Pleistocene glacial advances. Usingrelative criteria these ice advances were subsequently correlated toglaciations ranging from MIS 8 to MIS 2 (Clayton, 1968; Suggate,1990; Fig. 2).
The conceptual model based on which Pleistocene glaciationshave traditionally been differentiated in New Zealand postulatesthat geologically recent uplift caused an increasing verticaldisplacement of glacial landforms through time (e.g. Gage, 1958;Suggate, 1965, 1990). Successively older glacial deposits are thusoften preserved at successively higher elevations in the valley.Stacked and buried sequences of glacio-fluvial deposits are alsoknown from many non-glaciated lower valley portions andsubsiding alpine foreland basins such as the Canterbury Plains eastof the Southern Alps (Moar and Gage, 1973; Brown et al., 1988;Burrows and Moar, 1996). In this paper we deal with an extensivesedimentary fill section located in the LGM glaciated portion of theHope Valley. Because valley fill surfaces like this are often con-nected to LGM moraine systems, it has generally been inferred thatthe fills were predominantly deposited in association with the lastmajor ice advance phase (Gage, 1958; Speight, 1963; Clayton, 1968;Maizels, 1989; Hart, 1996; Mager and Fitzsimons, 2007). This notionis also expressed in Suggate’s (1965) seminal and still influentialwork on New Zealand glaciations where it is stated that ‘‘owing to
Fig. 2. Views into the study area showing glacially truncated spurs in the upper Hope Valley (a) and the terraced lower valley portion (b). The shaded DEM (c) shows the location ofPoplars Gully and the principal moraine features in the area: LGM moraines are associated with the main aggradational terrace on the southern valley side and are situated w160 m(610 m a.s.l.) above the modern Hope River. Two higher glacial surfaces are located at w300 m (Horseshoe Hill at 700 m a.s.l.) and w430 m (Kakapo Hill at 840 m a.s.l.) above riverlevel and are believed to represent earlier Late-Pleistocene glaciations (Clayton, 1968; Suggate, 1990). The photo in (d) shows a view of the Poplars Gully exposure, which formedduring a landslide in 1994. Also shown are the varying orientations of the individual sections of the logged western gully face.
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592578
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592 579
subsequent re-advances of the ice, deposits formed during the retreatfrom each [earlier] advance are rarely preserved .’’ (p. 9). Here wepresent a case study from the Hope Valley where we show thatsuch earlier (pre-LGM and pre-last glacial cycle) deposits havesurvived in substantial thickness despite subsequent ice advancesinto the valley during the LGM. The results are likely to apply tosimilar fill sequences in other valleys of the Southern Alps.
2. Methods
For this study we used standard sedimentary logging techniquesand a modified facies code from Eyles et al. (1983), Table 1. Depositsinterpreted to have formed under similar conditions of sedimen-tation were grouped into lithofacies assemblages (Lfa) on the basisof composition, texture, grain size, sorting, structures and strati-graphic context. Clast fabric and clast roundness data werecollected from most units identified as diamictons. These data aredisplayed using equal-area stereographic scatter- and contour plots(Stereo32 software) and were statistically analysed for eigenvectorsand eigenvalues. Paleo-flow directions from aqueous units wereobtained from cross-stratifications and imbricated clasts. Grain sizedistributions of fine-grained units were measured via laser analyser(Saturn DigiSizer).
2.1. IRSL dating methods
Four samples for infrared stimulated luminescence dating (IRSL)were analysed via a multiple aliquot additive-dose method (MAAD)and two additional control samples were analysed via a singlealiquot regenerative method (SAR). The samples were taken fromwell sorted lacustrine and fluvial sand/silt units of at least 1.5 mthickness by forcing a 220� 75 mm steel tube into the sediment.The cylinder was then sealed and carefully excavated to preventlight exposure and sediment mixing. All samples contained ampleamounts of silt and sand, with both quartz and feldspars present.We decided to apply a luminescence method based on feldspars, asNew Zealand quartz has proven to be very difficult to date (e.g.Preusser et al., 2006). Also, feldspar with its higher saturation doseis likely to be a superior material in the age range above 100 ka. Wedecided to date the silt fraction, as the data usually results in lessscatter than for sand sized samples.
Table 1Facies codes used for sedimentary logging at Poplars Gully.
Facies codes
Fines (<0.063 mm) Granules (2–8 mm)Fl – laminated mud, silt and fine sand GRo – openwork granulesFld – laminated silts and clays with
dropstonesGRmc – massive granules, with isolatedclasts
Sample preparation was conducted under subdued orange lightand included the removal of carbonates with HCl, organic matterwith H2O2 and iron oxide coatings by treatment with a solution ofsodium citrate, sodium bicarbonate and sodium dithionate. The4–11 mm grain size polymineralic fraction was extracted. Thesamples were then deposited evenly in a thin layer on 10 mmdiameter aluminium discs. All luminescence measurements werecarried out using a RISO TL-DA15 measurement system, equippedwith Kopp 5-58 and Schott BG39 optical filters to select the lumi-nescence blue band around 410 nm. This wavelength band isknown to minimize potential fading problems in feldspars (Langand Wagner, 1997). Optical stimulation was carried out atw30 mW/cm2 using infrared diodes. For beta irradiations theinternal 90Sr, 90Y source of the Riso instrument and an externalDaybreak 801E 90Sr, 90Y irradiator was used. Alpha irradiationswere done on an ELSEC 241Am irradiator. The palaeodose wasdetermined by measuring the blue light output during infraredoptical stimulation, which selectively stimulates the feldspar frac-tion of the polymineral sample.
All samples were initially measured using a multiple aliquotadditive-dose method (MAAD). Following shortshine normal-isation and irradiation the aliquots were stored in the dark for fourweeks and then preheated to 220 �C for 5 min to remove unstablesignal components. Subsequently, infrared stimulated lumines-cence (IRSL) measurements were done for 100s at room tempera-ture. Growth curves were then constructed from natural (9aliquots) plus 6 additive dose points (5 aliquots each). Fading testswere carried out on subsets of natural and irradiated discs (5 each),which were stored for 6 months.
When it became apparent that some samples were close tosaturation, and the mathematical extrapolation needed to calculatean MAAD age became potentially untrustworthy, two of thesamples were re-measured using a single aliquot regenerative(SAR) approach. The reasoning is that SAR uses interpolation andthus potentially achieves higher precision in the high dose region,provided there is still sufficient signal growth with dose. Certainly,if MAAD and SAR results match within error it increases theconfidence in the resulting age. The SAR measurements were (likeMAAD) conducted at room temperature, after a preheat to 270 �C(for 2 min). Apart from using fine-grained feldspars, our SARapproach was broadly similar to the one described by Murray andWintle (2000). Four regenerative dose points (including zero dose)plus one recycled point were used. As fine-grained samples havegood reproducibility, eight single aliquot discs per sample weresufficient to establish a narrow dose distribution.
After 4 weeks storage in an airtight perspex container theradionuclide contents for dose rate calculation were measured. Forthis the dry homogenised samples were counted on a broad energyHPGe gamma spectrometer for a minimum time of 24 h. The doserate calculation was based on the activity concentration of thenuclides 40K, 208Tl, 212Pb, 228Ac, 214Bi, 214Pb, 226Ra.
3. Lithofacies descriptions and interpretations
Sedimentological and stratigraphic data are presented for fourstacked sediment columns that total 120 m in vertical fill thickness(see Fig. 3). Eight lithofacies assemblages (Lfa A–H) were identifiedand are described below.
3.1. Lfa A: rhythmically laminated clay, silt and fine sand withdropstones, matrix supported and stratified diamictons:Fl, Fe, Fld, Dms
Lfa A rests above 1.4 m of plane bedded sand to granules withinterbedded silt and stratified gravel that form the base of the
Fig. 3. Simplified stratigraphic overview and location of lithofacies assemblages that are described in the text. Inset at top left shows the logged sedimentary fill section in the valleycontext. The circled numbers show the positions from where MAAD and SAR luminescence ages were obtained (note that samples 1þ2 and 4þ 5 are from the same unit and thatthe unit of sample 3 stratigraphically underlies samples 1þ2). The presence of two bedrock strath terraces below the Poplars Gully section indicates a fluvial incision phase prior tothe onset of deposition.
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592580
Poplars Gully exposure. Bedding in these units is heavily contortedwith local mobilization of silt, which was injected into overlyingunits (Fig. 4). Above this Lfa A consists of 2.1 m of laminated silt andclay containing isolated clasts with associated impact structures insurrounding fines (Fig. 4b) and a 0.2 m thick cobbly diamicton. Thisis overlain by 2.2 m of normally graded, banded sand and silt withcm-scale flame structures at silt-sand contacts and dm-scaleconvoluted bedding (Fig. 4d). Normal faulting is pervasivethroughout the fines and in some cases offsets the convoluted beds(Fig. 4c). The normal faults are noted for an incremental decrease ofdisplacement in the younging direction of the sediment. All units ofLfa A dip uniformly by w20� (SSW) towards the middle of the HopeValley (see dipping silty clinoforms in Fig. 4c).
Interpretation: Units directly below Lfa A are contorted throughloading which caused injection of material from these beds intohigher units. Above this follow laminated fines of Lfa A which area slack water deposit representing sedimentation into a small lake.A proximal, and probably calving glacier margin is indicated by theoccurrence of dropstones and a mass flow diamicton with associ-ated dump structures. A generally high sediment flux is demon-strated by hydroplastic soft-sediment deformation and small scaleflame structures between beds of varying density. In the context ofa glaciolacustrine setting and a nearby glacier margin, the probablecause for the observed severe sediment loading of units below Lfa Ais a glacial advance to near or over the site. This view is supportedby dead-ice deformation in the overlying glaciolacustrine sedi-ments where normal faults show a decrease of fault displacementin the younging direction of the sediment. The probable cause forthis is syndepositional removal of volume in underlying units,
which is consistent with the gradual melting of buried dead ice leftfrom the earlier glacier advance. The noted relatively steep dip(20�) of the laminated lacustrine fines is highly unlikely to repre-sent the original bedding angle of these units. The same bedorientations were also found for all other units in Lfa A (and Lfa B,see below) and we suggest postdepositional tilting of these heter-ogenous units en-block as the most likely explanation for theobserved bed orientations in Lfa A.
3.2. Lfa B: stratified sand, normally and inversely gradedmatrix supported gravel, massive and stratified gravel:Sx, Sh, Gs, Gm, Gms, Gni
Lfa B comprises a w10 m thick sand and gravel package (Fig. 4)which is exposed 6–16 m above ground where deposits were onlypartially accessible for logging at the face. Gravel and sand of Lfa Boccupy a w15 m wide channel cut that is exposed in cross-section.The basal 1.5 m of this fill comprise cross-bedded sand which isunconformably overlain by sheets of matrix to clast supportedcobble gravels. Individual gravel beds are 0.3–0.5 m thick, normallyand inversely graded, with sandy interbeds and patches of open-work gravel. Most clasts are subrounded to rounded. All units of LfaB dip concordantly with Lfa A at about 20� to SSW.
Interpretation: Lfa B represents a channel fill of fluvial deposits.Matrix supported gravels within the fluvial channel were laid downas sheet flows probably during flood events. Sedimentation of Lfa Bfollowed the termination of the prior ice-marginal lake (Lfa A)either through infilling or rapid drainage. The overall orientation ofthe exposed channel cut indicates a roughly eastward flow (down
Fig. 4. Stratigraphy, depositional context and summary log for lithofacies assemblages A and B. Units directly below Lfa A are heavily contorted with injection of material into higherunits (4a). Above this Lfa A includes mainly laminated lacustrine silt/sand with dropstones and associated impact structures in surrounding fines (4b). Convolute bedding iscommon in Lfa A (4d) as is small to medium (cm–m) scale normal faulting (4c). Photos and IRSL sample locations are marked on the stratigraphic column.
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592 581
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592582
the modern valley). Lfa B units contain maturely rounded clasts(subrounded to rounded) and deposition occurred on a braidfloodplain, probably in a more ice distal or paraglacial context. Thisimplies that the stratigraphic transition from the glaciolacustrineLfa A, to the more ice distal gravel and sands of Lfa B was accom-panied by glacial retreat from the site. All units of Lfa B dip at 20�
concordantly with the underlying beds of Lfa A which we see asa result of postdepositional en-block tilting of Lfa A and B. In thecontext of the stratigraphic transition from Lfa A to Lfa B whichindicates glacial retreat from the site a likely cause for the tilting isremoval of structural support associated with ice retreat.
3.3. Lfa C: stratified and massive diamictons, stratified gravel, planebedded and deformed sand: Dms, Dm, Gs, Sh, Sd
Lithofacies assemblage C comprises 7 m of diamictons andgravel that truncate Lfa A and B (Fig. 5). Towards the valley side LfaC rests directly on bedrock whereas it unconformably overlies softsediments of Lfa B further to the central valley. Based on differencesin composition and texture five diamict types can be distinguished,which are briefly described below.
Type I diamictons are clast supported, crudely stratified anddirectly overlie an erosional unconformity to Lfa B (see Fig. 5e).Above this follows a massive matrix dominated diamicton (type II)that contains few mainly rounded to subrounded clasts which insome cases are striated. This diamicton also shows an undulatingbasal contact and is associated with distinct horizontal planes in theunderlying beds along which material was displaced (Fig. 5a). Dia-micton type III is matrix supported, crudely stratified and includesup to 50 cm wide lenses of stratified sand which are extensionallydeformed by normal faulting (see Fig. 5b). Above this, type IV dia-mictons are well stratified with beds ranging in thickness between 3and 10 cm. The deposit contains lodged bullet-shaped cobble clastswhich are draped by cm-thick bands of sand as shown in Fig. 5d.Finally, type V diamictons are matrix supported and stratified andare distinguished because they are dominated by clinoform bedsthat dip at 30–33� and contain ductilly deformed sand beds (Fig. 5c).We collected fabric measurements (a-axis orientation and dip) fromall diamictons together with clast roundness data for 30–60 stonesper diamicton (except for diamicton type I which was inaccessible).The resulting preferred long axis eigenvectors (V1) and eigenvalues(S1 and S3) are shown in Fig. 5.
Interpretation: Based on composition, texture and fabricstrength we differentiate between mass flows and till in Lfa C. Massflows (diamictons I and V) were recognized based on their crudestratification and sediment lenses that were ductilly deformedduring remobilisation (see sandy boudinage structure at top left inFig. 5c). Calculated eigenvalues indicate a relatively weak fabric(S1¼0.526, diamicton V) and a larger average particle dip than in allother Lfa C diamictons. Reduced fabric strength for mass flowdeposits (compared to till) is a common observation from compar-ative fabric studies on modern sediment flows in glacio-deposi-tional environments (Lawson, 1979; Dowdeswell and Sharp, 1986).
The massive type II diamicton near the base of Lfa C has a loadedundulating basal contact and is associated with shear planes indirectly underlying beds. The deposit is identified as a subglacialtraction till where shearing of the ice-overan unit below causeddisplacement of pods of silty material. Fabric measurements indi-cate an eigenvalue (S1) of 0.635 (S3¼ 0.061) and a principaleigenvector (V1) of 204� suggesting ice movement towards NNE.Above this, stratified diamictons type III and IV show draping oflodged clasts by sorted sediment bands (Fig. 5d) and the presenceof intra-diamict lenses of fluvial deposits that indicate episodes ofmeltwater deposition during diamict accumulation. The fluvialsediment lenses show extensional deformation (Fig. 5b). Together,
we interpret these features (draping; lenses of fluvial sediments) asindicating subglacial diamict deposition (e.g. Lawson, 1979;Dreimanis, 1988) during basal meltout from debris rich basal icewith associated loss of volume that caused normal faulting. Meltouttills are usually associated with strong fabric signals (Lawson, 1979;Kruger, 1982; Dowdeswell and Sharp, 1986) and we record corre-sponding eigenvalues (S1) of 0.745 (type III) and 0.728 (type VI)with low accompanying S3 values of 0.028 and 0.026. In summary,Lfa C comprises a stratigraphic stack of mass flows, subglacialtraction till and meltout till, that truncate Lfa A & B and indicatea glacier re-advance into the lower Hope Valley.
This 23 m thick facies assemblage consists of multiple graveland sand units with a bed thickness of 0.4–1.0 m (Fig. 5). Individualunits comprise conformably bedded clast supported pebble andcobble gravels and cross-stratified sand. Interbedded betweenthese units occur several stratified and sandy diamictons. Intensedeformation at the base of Lfa D includes disruption of bedding andthe apparent rotation of m-scale sediment blocks that dip at steepangles into various directions. Higher in the sequence beddingplanes are intact but display gentle downward bending witha central deflection estimated to be about 1.5 m.
Interpretation: Lfa D was deposited as proglacial fluvial outwashand flood deposits. In its stratigraphic context the sequencerepresents the transition from ice-proximal mass flows and till (LfaC) to collapsed glacio-fluvial gravel (basal portion Lfa D) to lessdeformed fluvial units in the upper part of Lfa D. Sediment collapseand the downwarping of originally planar beds was presumablycaused by melting of buried ice left from the preceding glacieradvance (Lfa C). Fluvial palaeo-current directions taken fromundeformed ripple cross-stratified sand in the upper portion of LfaD indicate flow towards ENE matching the modern drainageorientation in the valley. The limited thickness of most units (<1 m)and very rapid changes in grain size between units suggest frequentand intense fluctuations in flow conditions, consistent with domi-nantly melt driven drainage on a proglacial outwash plain.
Lfa E consists of two repeated sequences of conformably beddeddiamicton, gravel and sand which are separated by a 1.5 m thickunit of laminated fines (Fig. 6). The diamictons contain few pebbles,are matrix supported, stratified and include 20–50 mm thick intra-diamict layers of medium sand (Fig. 6d). The basal 6 m of Lfa E areheavily deformed by folding and thrusting as shown in Fig. 7. Wealso observe small scale (mm–cm) deformation in the form oflaminae distortions within the fines of Lfa E (basal unit in Fig. 6aand detail in Fig. 6e).
Interpretation: Based on their stratification diamictons in Lfa Eare interpreted as mass flow deposits that alternate with units oflaterally discontinuous lacustrine mud and coarse fluvial gravel(see Fig. 6a). We interpret the depositional environment as pro-glacial. Small scale deformation within the fines are interpreted aswater escape structures either due to loading or rapid sedimenta-tion. Following deposition, basal sediments of Lfa E were deformedcausing folding of matrix supported materials and low angle thrustfaulting in adjacent, probably better drained, sand and gravel units(Fig. 7). The deformed beds are overlain by mass flow diamictonsand we consider two possible scenarios to explain the deformation:(1) glaciotectonic deformation associated with an advancing glacierto near or over the site and (2) deformation due to a down slope
Fig. 5. Stratigraphy, depositional context and summary log for lithofacies assemblages C and D. Lfa C truncates underlying units of Lfa B (see 5e and section sketch at top right offigure) and comprises a series of stratified and massive diamictons. The massive unit with associated shear planes in the underlying sand shown in 5a is a subglacial traction till.Other diamict units show draping of lodged clasts by sand bands (5d) and include deformed intra-diamict lenses of fluvial sand (5b). The overlying Lfa D comprises dominantlygravel and sand interbedded with stratified diamictons (see text).
Fig. 6. Stratigraphy, depositional context and summary log of lithofacies assemblages E, F, G and H. Lfa E consists of deformed lacustrine mud (see 6e), clast supported fluvial gravel(central unit in 6a) and stratified diamictons (6d). The diamicton at the base of Lfa E overlies folded and thrusted sediments, which we interpret as glaciotectonic deformation (seeFig. 7). Lfa F comprises lacustrine mud containing facetted dropstones (see 6c). Alluvial and fluvial gravel of Lfa G and loess of Lfa H (see 6b) cap the sequence.
Fig. 7. Deformation of deposits at the base of Lfa E. The structures combine ductile folding of matrix dominated units (7a) and brittle thrust faulting of sand and gravel units (7b).Note the 2-m high measuring staff in center of photo (a), and knife in photo (b) for scale. The local face cuts obliquely across the asymmetrical fold in (a) (N to NNW axial plane).
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592 585
mass failure. There is no till in Lfa E but it is possible that defor-mation occurred from proglacial shortening due to advancing icewithout actual ice-overrun. Measurements on fold limbs and thrustplanes indicate that compression occurred in an E – ENE (down-valley) direction. This is roughly consistent with the fabric signalrecorded from the overlying diamicton (NE, Fig. 6). Because theadjacent mountain side trends WSW–ENE we consider a potentialdown slope mass failure as an unlikely cause as this would producecompression towards the south and not in the observed (mountainparallel) easterly direction. We therefore prefer a scenario whereunits in Lfa E were deformed proglacially by advancing ice whichdeposited mass flow diamictons but stopped short of overrunningLfa E.
This 56 m thick dominantly fine-grained lithofacies assemblagecomprises alternating units of mud and sand (Fig. 6). Dispersedstones with impact structures in surrounding laminated fines arepresent throughout the mud and sand units but increase infrequency below a 0.35 m thick diamicton near the middle of Lfa F.About half of the stones within the diamicton are angular in shape(rest subrounded to rounded) and some of the stones dispersed
throughout the fines show facetted surfaces (Fig. 6c). Between thelaminated fines in the middle section occurs a 5 m thick unit ofcross-stratified sand. Above this follow more laminated and beddedfines that contain dispersed stones.
Interpretation: We interpret Lfa F deposits as lacustrine finesthat are interbedded with fluvial sand. The substantial thickness ofthe lacustrine fines (>45 m) and frequent occurrence of angulardropstones suggest that these units accumulated in a deep pro-glacial lake which was connected to a calving glacier margin. In thestratigraphic context the lake formed following the retreat from theearlier ice advance (Lfa E). Frequent granulometric changes withinthe facies assemblage suggest varying lake levels, where units ofcross-stratified sand represent lake low stands. Our measurementsof crest orientations on the sinuous-crested climbing ripples withinthe cross-stratified sands indicate a NNE paleo-flow direction(down the modern valley).
Lfa G has a total thickness of 14.5 m and consists of three distinctgravel units. The lower and upper bed (G1 and G3 in Fig. 6b) areclast supported cobble to pebble gravels with a matrix of grittysand. Clasts within these units are noted for being dominantlysubangular to subrounded in shape. An intermediary unit (G2)
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592586
comprises 4 m of coarse, well sorted, clast supported and domi-nantly well rounded bouldery gravel.
Interpretation: The lower and upper gravel beds (G1 and G3) arelocally derived alluvial fan sediments that were transported overa short distance. The interbedded gravel unit (G2) representsa coarse fluvial gravel. Deposition of G1 occurred when a fan wasdeposited from the adjoining hill onto the floodplain. As the fanaccumulated, channel avulsion from the main floodplain tempo-rarily relocated a braid channel to this section of the valley where iteroded part of this fan and deposited coarse fluvial gravel (G2).Subsequently, as the channel avulsed away again, fan buildingrecommenced burying the fluvial deposit (G3). The mature sortingand roundness of boulders in the G2 gravel suggest medium to longdistance transport. Lfa G clearly represents a depositional envi-ronment that differs from all previously described facies at PoplarsGully, which are dominated by ice-proximal and proglacialdeposits. G2 gravels closely resemble the modern floodplain sedi-ments in this portion of the Hope Valley and we suggest depositionin a paraglacial or non-glacial environment. The deposits cap thePoplars Gully sequence and relate directly to the terrace surfacemorphology. Lithologically very similar fan/fluvial terrace gravels(at equal elevation and identical stratigraphic position) are exposedon the opposite (southern) side of the lower Hope Valley. A lumi-nescence age of 32.1�2.6 ka BP (WLL 351) was obtained by Rotheret al. (2007) from those gravels and a similar age of Lfa G is likely.
3.8. Lfa H: massive silt and fine sand: Fm
This is a massive sandy silt to fine sand unit of 0.5 m thicknessthat overlies Lfa G. Numerous large angular boulders are scatteredon the surface.
Interpretation: The deposit is identified as loess. The boulders onthe surface are rockfall debris derived from the adjoining hill side.
4. Luminescence dating results
Four samples for infrared stimulated luminescence dating (IRSL)were analysed using a multiple aliquot additive-dose method(MAAD) while two additional samples were measured usinga single aliquot regenerative technique (SAR). The sampled unitscomprise well sorted laminated and cross-stratified fluvial-
Table 2Radionuclide and water contents for luminescence samples from Poplars Gully. The wateris calculated from the error weighted mean of the isotope equivalent contents between 2
dose rate dDc/dt, total dose rate dD/dt and resulting OSL ages. Samples 1/2 and 4/5 werecosmic dose rate was calculated as per Prescott and Hutton (1994). The a-irradiated subsaSample 6 showed a minor radioactive disequilibrium. The dose rate and age given were atime is unknown, this age is only a better estimate and cannot be seen as the ‘true’ age. Thcalculated under the invalid assumption that the samples were in radioactive equilibriu
Sample no. Water content d U (mg/g) from 234Th U (mg/g) from 226Ra,214Pb, 214Bi
lacustrine sands which were judged to have had a good chance ofsufficient light exposure during transport. Our sampling targetedstratigraphically widely spaced units (vertical distance betweensamples from Lfa A and F is 75 m, see sample positions in Fig. 3) andincludes various soft-sediment lithologies to maximize chancesthat different sediment sources and depositional pathways wererepresented. Dating yielded MAAD IRSL ages of 115.0� 9.9 ka (1),127.5�10.7 ka (3), 181.3�16.4 ka (4) and 157.8� 14.8 ka (6). Tocheck the ages for consistency two repeat measurements onsamples 1 and 4 were undertaken using the SAR method. Theresulting two SAR ages 119.8� 10.5 ka (2) and 164.2�17.4 ka (5)agree within error with the MAAD ages for the same units(115.0� 9.9 ka and 181.3�16.4 ka). Details on water content,radionuclide data, dose rates, measured a-values and the resultingluminescence ages are shown in Table 2, all errors are quoted at 1sigma. A fading test performed after 6 months storage showed nosignificant anomalous fading in any of the samples and no correc-tive measures were required.
Fig. 8A shows the growth curves of all MAAD samples. As is to beexpected in this dose range, the samples show the onset of satu-ration. We were still able to fit the data with a saturating expo-nential curve, and have some confidence in the mathematicalextrapolation that yields the final OSL age. The two SAR repeatmeasurements show strong signal growth (Fig. 8B and C) andconfirm the MAAD IRSL results for the same units within error,which gives confidence to our assessment that these samples are atthe onset of saturation but are otherwise robust. Still, we notea partial age reversal as some of the older samples stratigraphicallyoverlie younger samples (see sample positions in Fig. 3). Althoughthis provides limits in the interpretability of the data we note thatour logging shows conclusively that the Poplars Gully sequenceconstitutes a continuous sequence of related sediments (glacial–proglacial). The sequence contains no major hiatus and it is quitelikely that the glacial sediments there are contemporaneous (theexception being the younger non-glacial cover beds of Lfa G & Hwhich over-topped the sequence). An important result of our agedating is that all IRSL ages are >115 ka which we regard to be theminimum age of the Poplars Gully sequence. This interpretation issupported by additional stratigraphic and geochronologicalconstraints from other sections of the Hope Valley fill (see discus-sion, Section 5.2).
content d is calculated as a weight ratio of wet sample/dry sample. U- and Th-content26Ra and 210Pb. The lower table shows measured a-value, equivalent dose De, cosmic
measured by both MAAD and SAR methods, using finegrain-feldspar material. Themple of sample 6 was saturated, so the a-value given in the table had to be estimated.djusted to that, using all U-chain data. However, as the level of disequilibrium over
e effect appears minor, and the uncorrected dose rates and ages are given in brackets,m.
Fig. 8. MAAD growth curves for samples 1, 3, 4 and 6 (8A) all of which approach saturation. Results for the SAR repeat measurements on two finegrain-feldspar subsamples areshown in 8B (sample 2) and 8C (sample 5). We show representative natural shinedown curves (I), and growth curves (II), and dose distributions (III) for all 8 aliquots that weremeasured.
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592 587
The four MAAD and two SAR IRSL ages from Poplars Gully fallbetween 115 and 181 ka, suggesting that deposition of all or most ofthe glacial materials at Poplars Gully is likely to have occurredduring the MIS 6 glaciation. However, because the MAAD IRSLsamples were approaching saturation we cannot entirely rule outthe possibility that at least some materials are pre-MIS 6 in age andwe note that such scenario would in fact strengthen our point thatextensive pre-LGM deposits survived in the valley. But the pre-MIS6 scenario would imply an age around or in excess of 200 ka for thePoplars Gully sequence, which appears to conflict with the resultsof the two SAR IRSL samples. These samples showed strong growthcurves and yielded ages of 164.2 ka and 119.8 ka, respectively. Insummary, although we hesitate to entirely rule out a pre-MIS 6scenario for Poplars Gully, on balance, we believe that glacialdeposition there during MIS 6 is the likelier and simplerinterpretation.
5. Discussion
5.1. Depositional model
Poplars Gully preserves a stratigraphically intact depositionalsuccession that allows for the reconstruction of the site history. Weidentify nine phases, which place deposition in the context oflargely ice-marginal/pro- and paraglacial sedimentation duringmultiple phases of ice incursion, stagnation and ice withdrawalfrom the lower Hope Valley. Common deposits in the sequenceinclude deformed and undeformed, glacio-fluvial gravels, glacio-lacustrine fines, fan sediments, debris flow and glacial diamictons(Fig. 9).
Preceding the onset of sedimentation, a fluvial incision phase inthe valley is indicated by two bedrock strath terraces whichstratigraphically underlie Poplars Gully deposits (Figs. 9a and 3).
Fig. 9. Site history of Poplars Gully showing key depositional events (a–h) and the successive build-up of the exposed stratigraphy (i). The initial glacial advance (b) left ice-marginalsediments from the retreat phase of that advance (c). The deposits display meltout deformation and were postdepositionally tilted due to structural ice collapse (d). This wasfollowed by a glacial re-advance (e) that truncated deposits laid down during the previous retreat. Glacio-fluvial deposits above this indicate ice retreat from the site (f) before a last,probably short-lived glacier advance deposited mass flow diamictons and glaciotectonized the underlying proglacial sequence (g). Final glacial retreat resulted in the formation ofa large proglacial lake (h) and eventually by the deposition of coarse floodplain gravels (i). Details are discussed in the text.
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592588
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592 589
This phase of valley erosion is likely to have removed pre-existingvalley fill deposits, though geophysical data show that olderdeposits may still be preserved below the modern floodplain(Rother et al., 2007). Following an initial ice advance (Fig. 9b), basalunits at Poplars Gully record sedimentation associated with theretreat of that advance. This includes glaciolacustrine sedimentswith dead-ice meltout deformation and overlying proglacial fluvialgravels (Lfa A and B; Fig. 9c). The retreat appears to have resulted ingeneral ice withdrawal from the lower Hope Valley because it isaccompanied by large scale collapse deformation such as the tiltingof blocks of heterogeneous lacustrine and fluvial deposits (Lfa A andB) probably due to the removal of structural ice support (Fig. 9d).
The ice retreat was followed by an ice re-advance (Fig. 9e). Massflows and tills from this advance (Lfa C) unconformably truncateunderlying sediments. The limited occurrence of till deposited fromactively moving ice together with the absence of significant gla-ciotectonic deformation in this part of the sequence suggestsdeposition occurred mainly during ice stagnation via passivemeltout. Sedimentation in the context of progressive ice down-wasting would also be consistent with the observed stratigraphictransition from basal meltout till to mass flows (Lfa C) to domi-nantly glacio-fluvial sediments of the overlying Lfa D. During thedeposition of the latter facies assemblage the site may either havebeen proglacial or formed part of a kame terrace system (Fig. 9f).
Sediments of Lfa E record a renewed ice advance (Fig. 9g) to nearor over the site leaving ice-proximal mass flows and causing gla-ciotectonic folding and thrusting. The subsequent ice retreat fromthe valley led to the formation of a proglacial lake in which 45 m ofglaciolacustrine sediments accumulated (Fig. 9h). An oscillatingand calving glacier margin during deposition of Lfa F is indicated bythe input of varying amounts of ice rafted debris. Interbedded unitsof cross-stratified sand within the lacustrine fines reflect repeatedlake low stands. In total, stratigraphic data from Poplars Gullyprovide evidence for three individual glacial advances thatoccurred during two major ice incursions into the lower HopeValley. We consider this a significant finding because Poplars Gullyis located 30–40 km down valley from the ice accumulation areasand only large glacial advances reach the lower portion of thevalley.
5.2. Glacial and geochronological implications
Until recently Late Quaternary age control in the Hope Valleywas limited to a small number of 14C ages that primarily con-strained the end of paraglacial aggradation (17.4 ka cal BP, Suggate,1965) and the timing of last glacial ice retreat (16.2 ka cal BP,Clayton, 1968). In addition, Knuepfer (1988) reported a relative clastweathering rind date of 18.7�3.8 ka for a presumed LGM morainecomplex at Glynn Wye. Based on a clear geomorphological asso-ciation between the Glynn Wye moraine and extensive outwashdeposits (the main terraced outwash surface in the valley is shownin Fig. 2b) it was generally assumed that most or all of the fill therewas deposited in the context of LGM aggradation and subsequentice incursions (Suggate, 1965; Clayton, 1968). However, recent workby our group on the south side of the lower Hope Valley challengedthis notion by showing that the majority of the local fill was notdeposited during the LGM but instead was laid down during theperiod 95.7–32.1 ka (MIS 5–MIS 3, see Rother, 2006; Rother et al.,2007). These sediments were then overrun by an LGM ice advance,which left a comparably minor volume of sediments but shaped theglacial topography of the present valley fill surface (Fig. 10).
Our logging at Poplars Gully found mainly ice-proximal sedi-ments, which are very different from the dominantly non-glacialfluvial and lacustrine-deltaic units exposed in identical strati-graphic position on the opposing southern valley side (see Fig. 10B,
C; Rother et al., 2007). Apart from a veneer of capping gravels(which form the surface of the terrace on both valley sides) none ofthe glacial sediments at Poplars Gully can be projected across thevalley for stratigraphic correlation. This stratigraphic mismatch isconsistent with our luminescence data, which show that the sedi-ments at Poplars Gully (�MIS 6) pre-date the non-glacial fillexposed on the southern valley side.
The key finding of this and our earlier studies from the lowerHope Valley is the recognition that penultimate glaciation depositsand early-mid last glacial cycle sediments dominate within thelocal fill stratigraphy. This was unexpected insofar as that LGM iceadvances are known to have overran this section of the valley andmost of the present fill has previously been ascribed to this lastglacial advance phase (e.g. Suggate, 1965, 1990; Clayton, 1968).Although LGM deposits are documented in the valley (Clayton,1968; Knuepfer, 1988; Rother, 2006) we can show that the asso-ciated sedimentary volume is comparably small as these depositsonly cap much thicker pre-LGM sediments, which survived in situin the valley trough. The limited LGM depositional volume couldbe explained if the LGM ice advances were to have been largelyerosive in this part of the valley, but this can confidently be ruledout as we demonstrate the extensive survival of pre-existing fillsequences.
Pre-last glacial cycle terminal moraines from which to recon-struct the relative ice extent of mid-Pleistocene glaciations(particularly MIS 6 and 8) are extremely scarce in the SouthernAlps and entirely missing in the case of the Hope Valley. In thissituation a stratigraphic approach such as presented in this papermay be the next best proxy to compare the relative magnitude ofsuccessive Pleistocene glaciations. It has been argued that over thepast 350 ka glaciations in the New Zealand region were of verysimilar scale (e.g. Suggate, 1990; p. 182). In contrast to this, ourresults from the Hope Valley indicate that ice advances during theLGM were significantly smaller than during previous glaciations(particularly MIS 6), which we base on three key observations:(1) aggradation during the penultimate glaciation filled the valleyto much higher levels than during the last glacial cycle (Suggate,1965; Clayton, 1968; Rother, 2006), (2) a direct comparison ofsedimentary volumes of glacial deposits associated with the LGMand penultimate glaciation shows the LGM advances to be signif-icantly smaller, and (3) MIS 6 glaciers reaching the lower HopeValley eroded a much deeper glacial trough than during the LGM.The latter observation is based on reconstructed base levels of therespective glacial troughs from exposed basal contacts of MIS 6 andLGM sequences to older deposits or to bedrock. The upper verticallimit of the sedimentary column was established from exposedconformable contacts to overlying non-glacial deposits (loess andalluvium). If directly compared, the MIS 6 trough extends at least120 m below that of the LGM for the same valley segment. Takentogether these observations suggest a marked difference in thescale of successive glaciation in the Hope Valley. Alternatively,a comparably shorter duration of ice occupation during the LGMmay also account for the observed phenomena in the Hope Valley,but this too would indicate that MIS 6 was the more significantglacial event.
Similar results have recently been reported from the RakaiaValley in the central Southern Alps (Shulmeister et al., 2007), whereearly-mid last glacial cycle and penultimate glacial sediments alsodominate despite later LGM ice occupation of the valley. Togetherthese findings draw into question the common notion in NewZealand that the sedimentary fills widely preserved in valleytroughs that were ice occupied during the LGM, are necessarily orpredominantly the product of the last glacial advance phase. Inaddition to the known preservation of tectonically uplifted sedi-ments of earlier Late-Pleistocene glaciations at higher elevations,
Fig. 10. Schematic cross-section through the lower Hope Valley showing the composite architecture of the local fill sequence (10A). The level to which aggradation during the penultimateglaciation filled the valley can be reconstructed from terrace remnants preserved 100–120 m above the LGM surfaces on the southern valley side (see 10A, Suggate,1965; Clayton,1968).Rother (2006) obtained an IRSL luminescence age of 110.5�10.6 ka from a degradational terrace that is part of this higher terrace sequence. Our stratigraphic and geochronological resultsindicate that last interglacial fluvial incision (similar in scale to Holocene incision) excavated the valley before renewed aggradation commenced at around 95 ka BP (see 10B, Rother, 2006;Rother et al., 2007). However, a large remnant of on older glacial fill sequence survived on the northern valley side in a sheltered position behind a bedrock spur (Poplars Gully). Youngeraggradation deposits (last glacial cycle) over-topped the older sediments at Poplars Gully and incorporated them into the paired aggradational terrace of the modern valley (10A).
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592 591
we highlight that such sequences also survived in the main valleytroughs where they form part of the modern valley fill.
A probable explanation for the preservation of thick soft-sedi-mentary pre-LGM sequences despite later ice incursions is that theLGM glaciers (which reached their maximum extent in the lowerHope Valley) were too thin to cause effective bed erosion. Inaddition, we note that these glaciers advanced primarily overproglacial outwash, which in turn was aggraded over older fillsequences. The typically coarse outwash deposits would haveconstituted a well drained glacier bed, thus largely preventingsubglacial erosion by dissipating meltwaters. Similarly, a relativelydry glacier–bed interface would have reduced basal ice sliding andthe associated plucking of sediment. Thus, although New Zealand’stemperate glaciers are noted for very high erosion rates overbedrock (Hallet et al., 1996), in a situation where these glaciersexpand into the sediment dominated middle and lower valleyreaches their erosive capacity is limited.
A number of depositional and structural features observed inthe Hope Valley are generally typical for glacial valley fills and havebeen described from various other mountain regions (Owen, 1989;Fitzsimons, 1992; Aitken, 1998; Ward and Rutter, 2000; Benn et al.,2003). Some less common characteristics of the valley fills from theSouthern Alps include the very substantial sedimentary thicknessand the near complete dominance of meltwater derived deposits inboth advance and recessional fill sequences. We see these featuresas inherent characteristics of New Zealand’s perhumid maritimeglacier systems, which generate larges volumes of free meltwaterand operate in an active tectonic zone with very high sedimentsupply. Fluvial reworking of glacial deposits is pervasive andprimary subglacial sediments (i.e. subglacial traction tills) havea very low chance of preservation in this setting. Glaciolacustrinefacies types appear to be the dominant component in glacio-depositional environments, particularly in the less steeplydescending valleys of the eastern Alps. Widespread stratified andweakly compacted diamictons, often described as tills in NewZealand (e.g. Gage, 1958; Speight, 1963; Soons, 1963; McGregor,1981; McCalpin, 1992) represent the type sediment for this setting.From our experience many of these diamict deposits are betterdescribed as ice-proximal mass flows or remobilized/glacio-tectonized sediments.
6. Conclusions
(1) Sediments at Poplars Gully record three ice advances in thecontext of two major ice incursions into the lower Hope Valley.Deposition occurred predominantly along a stagnating glaciermargin in conjunction with short-lived glacial lakes that weredammed behind terminal moraines or thick outwash heads. Iceadvances typically produced large glacio-fluvial sequences,while retreat phases are identified by the deposition ofglaciolacustrine sediments, mass flow diamictons and poorlysorted proglacial fan deposits.
(2) Based on stratigraphic and geochronological data from PoplarsGully and other sites throughout the valley (this paper, Rother,2006; Rother et al., 2007) we show that the majority of the soft-sedimentary fill in the lower Hope Valley was deposited duringthe penultimate and early-mid last glacial cycle (MIS 6, MIS5–3). It is possible that some of the glacial materials at PoplarsGully even pre-date MIS 6. We conclude that deposition ofmost of the fill was not, as previously thought, associated withdeposition in the context of LGM aggradation or ice incursions.Although we document LGM deposits in the valley their sedi-mentary volume is comparably small as we find that thesedeposits only cap much thicker older deposits, which survivedin situ in the valley trough.
(3) The penultimate glaciation deposits in the Hope Valley aresubstantially more voluminous than those of the last glacialcycle. MIS 6 valley glaciers excavated a larger and deepererosional valley trough than during either MIS 4 or the LGM,respectively. In addition, the MIS 6 glaciation aggradationsurface is located at least 120 m above the LGM glacial surface.We suggest these features reflect a significant difference in thescale of successive Mid- to Late-Pleistocene glaciations in thevalley, where particularly the MIS 6 glacial advances wereconsiderably larger than those of the last glacial cycle.
(4) Pleistocene valley fills in New Zealand are a product ofa geologic setting that is dominated by very high sedimentfluxes coupled with high-throughput/high-capacity glacial andfluvial transport systems. Proglacial fluvial reworking ofprimary glacial deposits was pervasive and led to a low chanceof preservation for tills and related subglacial facies. Thegeneral dominance and great thickness of meltwater deriveddeposits in Pleistocene glacio-depositional environments ofNew Zealand indicates that free water was abundantly avail-able probably for most parts of the year. This is consistent witha model of maritime glaciers that extended far into the ablationzone to low elevations above sea level. The resulting deposi-tional sequences, albeit not entirely unique to New Zealand,differ in composition and sedimentary architecture from thoseof drier, higher latitude/higher elevation glacier systems inother alpine regions.
Acknowledgements
We would like to thank Chris Smart and Phil Tonkin for theircomments in the field and acknowledge Mrs. Ningsheng Wang’scareful work in preparing the OSL samples. H. Rother would like tothank the Henderson family of Poplars Station for providing shelterand access to ‘Henry’s Hole’. This study was financially supportedby a Royal Society of New Zealand Marsden Grant (contract UoC301) and a Mason Trust Award through the Department ofGeological Sciences/University of Canterbury (New Zealand).
References
Aitken, J.F., 1998. Sedimentology of Late Devensian glaciofluvial outwash in the DonValley, Grampian Region. Scottish Journal of Geology 34, 97–117.
Andrews, P.B., 1973. Late Quaternary continental shelf sediments off OtagoPeninsula, New Zealand. New Zealand Journal of Geology and Geophysics 16(4), 793–830.
Barrows, T., Lehman, S., Fifield, L.K., De Decker, P., 2007. Absence of cooling in NewZealand and the adjacent ocean during the Younger Dryas chronozone. Science318, 86–89.
Brown, L.J., Wilson, D.D., Moar, N.T., Mildenhall, D.C., 1988. Stratigraphy of the lateQuaternary deposits of the northern Canterbury Plains, New Zealand. NewZealand Journal of Geology and Geophysics 31, 305–335.
Burrows, C.J., Moar, N.T., 1996. A mid Otira glaciation palaeosol and flora from theCastle Hill Basin, Canterbury, New Zealand. New Zealand Journal of Botany 34,539–545.
Clayton, L.S., 1968. Late Pleistocene glaciations of the Waiau Valleys, North Can-terbury. New Zealand Journal of Geology and Geophysics 11, 753–767.
Cowan, H.A., 1990. Late Quaternary displacements on the Hope Fault at GlynnWye, North Canterbury. New Zealand Journal of Geology and Geophysics 33,285–293.
Denton, G.H., Hendy, C.H., 1994. Younger Dryas age advance of Franz-Josef Glacier inthe Southern Alps of New Zealand. Science 264, 1434–1437.
Dowdeswell, J.A., Sharp, M.J., 1986. Characterization of pebble fabrics in modernterrestrial glacigenic sediments. Sedimentology 33, 699–710.
Dreimanis, A., 1988. Tills: their genetic terminology and classification. In:Goldthwait, R.P., Matsch, C.L. (Eds.), Genetic Classification of GlacigenicDeposits, pp. 17–83. Rotterdam.
Eyles, N., Eyles, C., Miall, A.D., 1983. Lithofacies types and vertical profilemethods; an alternative approach to the description and environmental
H. Rother et al. / Quaternary Science Reviews 29 (2010) 576–592592
interpretation of glacial diamict and diamictite sequences. Sedimentology 30,393–410.
Fitzsimons, S.J., 1992. Sedimentology and depositional model for glaciolacustrinedeposits in an ice-dammed tributary valley, west Tasmania, Australia. Sedi-mentology 39, 393–410.
Gage, M., 1958. Late Pleistocene glaciations of the Waimakariri Valley, Canterbury,New Zealand. New Zealand Journal of Geology and Geophysics 1, 123–155.
Glasser, N.F., Harrison, S., Ivy-Ochs, S., Duller, G.A.T., Kubik, P.W., 2006. Evidencefrom the Rio Bayo valley on the extent of the North Patagonian Icefield duringthe Late Pleistocene-Holocene transition. Quaternary Research 65, 70–77.
Hallet, B., Hunter, L., Bogen, J., 1996. Rates of erosion and sediment evacuation byglaciers: a review of field data and their implications. Global and PlanetaryChange 12, 213–235.
Hart, J., 1996. Proglacial glaciotectonic deformation associated with glaciolacustrinesedimentation, Lake Pukaki, New Zealand. Journal of Quaternary Science 11,149–160.
Henderson, R.D., Thompson, S.M., 1999. Extreme rainfalls in the Southern Alps ofNew Zealand. Journal of Hydrology (N.Z.) 38, 309–330.
Ivy-Ochs, S., Schluechter, C., Kubik, P.W., Denton, G.H., 1999. Moraine exposure datesimply synchronous Younger Dryas glacier advances in the European Alps and inthe Southern Alps of New Zealand. Geografiska Annaler 81A, 313–323.
Kamp, P.J.J., 1986. The mid-Cenozoic Challenger Rift System of western New Zealandand its implications for the age of the Alpine fault inception. Bulletin of theGeological Society of America 97, 255–281.
Kaplan, M.R., Ackert Jr., R.P., Singer, B.S., Douglass, D.C., Kurz, M.D., 2004. Cosmo-genic nuclide chronology of millennial-scale glacial advances during O-isotopestage 2 in Patagonia. Geological Society of America Bulletin 116, 308–321.
Knuepfer, P.L.K., 1988. Estimating ages of late Quaternary stream terraces fromanalysis of weathering rinds and soils. Geological Society of American Bulletin100, 1224–1236.
Kruger, J., 1982. Processer og till fabric i et recent dodislandskap vedH}ofdabrekkujokull, Island. Dansk Geologisk Forening, Årsskrift 1981, 45–56.
Lang, A., Wagner, G.A., 1997. Infrared stimulated luminescence dating of Holocenecolluvial sediments using the 410 nm emission. Quaternary Geochronology(Quaternary Science Reviews) 16, 393–396.
Lawson, D.E., 1979. Sedimentological analysis of the western terminus region of theMatanuska Glacier, Alaska. Journal of Geology 87, 629–645.
Mager, S., Fitzsimons, S., 2007. Formation of glaciolacustrine Late Pleistocene endmoraines in the Tasman Valley, New Zealand. Quaternary Science Reviews 26,743–758.
Maizels, J.K., 1989. Differentiation of late Pleistocene terrace outwash deposits usinggeomorphic criteria: Tekapo Valley, South Island, New Zealand. New ZealandJournal of Geology and Geophysics 32, 225–241.
McCalpin, J.P., 1992. Glacial geology of the upper Wairau Valley, Marlborough, NewZealand. New Zealand Journal of Geology and Geophysics 35, 211–222.
McGregor, G.R., 1981. Weathering characteristics of Late Pleistocene tills. NewZealand Journal of Geology and Geophysics 24, 107–113.
Moar, N.T., Gage, M., 1973. Interglacial deposits in Joyce Stream (S74), WaimakaririValley, Canterbury. New Zealand Journal of Geology and Geophysics 16, 321–331.
Murray, A.S., Wintle, A.G., 2000. Luminescence dating of quartz using an improvedsingle-aliquot regenerative-dose protocol. Radiation Measurements 32, 57–73.
Norris, R.J., Koons, P.O., Cooper, A.F., 1990. The obliquely-convergent plate boundaryin the South Island of New Zealand: implications for ancient collision zones.Journal of Structural Geology 12, 715–725.
Owen, L.A., 1989. Terraces, uplift and climate in the Karakoram Mountains,Northern Pakistan: Karakoram intermontane basin evolution. Zeitschrift fuerGeomorphologie, N.F., suppl.-bd. 76, 117–146.
Prescott, J.R., Hutton, J.T., 1994. Cosmic ray contribution to dose rates for lumines-cence and ESR dating: large depths and long-term time variations. RadiationMeasurements 23 (2–3), 497–500.
Preusser, F., Ramseyer, K., Schluchter, Ch, 2006. Characterisation of low OSLintensity quartz from the New Zealand Alps. Radiation Measurements 41,871–877.
Rother, H., 2006. Late Pleistocene glacial geology of the Waiau Valley system innorth Canterbury, New Zealand, – glacial stratigraphy and implications forreconstructing glacial processes and palaeoclimate. Ph.D. thesis, Departmentof Geological Sciences, University of Canterbury, Christchurch, New Zealand,344 pp.
Rother, H., Jol, H.M., Shulmeister, J., 2007. Stratigraphy and tectonic implications of LatePleistocene valley fill in the Hope Valley, Canterbury, South Island, New Zealand.In: Baker, G.S., Jol, H.M. (Eds.), Stratigraphic Analyses using Ground PenetratingRadar (GPR). Geological Society of America Special Paper 432, 155–168.
Schaefer, J.M., Denton, G.H., Barrell, D.J.A., Ivy-Ochs, S., Kubik, P.W., Andersen, B.G.,Phillips, F.M., Lowell, T.V., Schluchter, C., 2006. Near-synchronous interhemi-spheric termination of the last glacial maximum in mid-latitudes. Science 312,1510–1513.
Shulmeister, J., Rieser, U., Fink, D., Thackray, G., Hyat, O., Rother, H., 2007. Lumi-nescence and surface exposure dating chronology from 5e to YD in the RakaiaValley, Canterbury, New Zealand and insights for the timing and forcing of NZglaciation. Quaternary International 167–168, 383. supplement, INQUACongress Abstracts.
Shulmeister, J., Fink, D., Augustinus, P., 2005. A cosmogenic nuclide chronology ofthe last glacial transition in North-West Nelson, New Zealand – new insights inSouthern Hemisphere climate forcing during the last deglaciation. Earth andPlanetary Science Letters 233, 455–466.
Soons, J.M., 1963. The glacial sequence in part of the Rakaia Valley, Canterbury, NewZealand. New Zealand Journal of Geology and Geophysics 6, 735–756.
Speight, J.G., 1963. Late Pleistocene historical geomorphology of the LakePukaki area, New Zealand. New Zealand Journal of Geology and Geophysics 6,160–188.
Suggate, R.P., 1965. Late Pleistocene geology of the northern part of the South Island,New Zealand. New Zealand Geological Survey Bulletin 77, 90.
Suggate, R.P., 1990. Late Pleistocene and Quaternary glaciations of New Zealand.Quaternary Science Reviews 9, 175–197.
Turney, C.S.M., McGlone, M.S., Wilmhurst, J.M., 2003. Asynchronous climate changebetween New Zealand and the North Atlantic during the last deglaciation.Geology 31 (3), 223–226.
Ward, B., Rutter, N., 2000. Deglacial valley fill sedimentation, Pelly River, YukonTerritory, Canada. Quaternary International 68–71, 309–328.
Wood, R.A., Pettinga, J.R., Bannister, S., Lamarche, G., McMorran, T.J., 1994. Structureof the Hanmer strike-slip basin, Hope Fault, New Zealand. Geological Society ofAmerica Bulletin 106, 1459–1473.
