Sediment sequence at Muhos, western Finland – a window to the Pleistocene history of the Scandinavian Ice Sheet

In this study, we analysed a ~54‐m sediment core consisting of Quaternary sediments overlying the Neoproterozoic Muhos Formation in central western Finland, adjacent to the Gulf of Bothnia. The sediments recovered were logged, and their sedimentological characteristics defined. Two fine‐grained sediment units were subjected to biostratigraphical studies using pollen and diatom analyses. In addition, two sand‐rich units and a wooden stick were dated by the optically stimulated luminescence (OSL) and 14C‐AMS methods. The core sediments were divided into six units, where several diamicton, sand and gravel, and silt‐and‐clay‐dominated beds were studied. The results indicate that the sediment succession of the core beneath the Holocene Litorina Sea and the Ancylus Lake sediments of the Baltic Basin were deposited in glacial and lacustrine environments that existed in the Oulu River valley during the time period between the Saalian glaciation (MIS 6) and the Holocene. The stratigraphical evidence, supported by the OSL ages, suggests that the Scandinavian Ice Sheet (SIS) entered the Muhos area during the Saalian glaciation, and at least during three separate time intervals in the Weichselian stage. Stratigraphically controlled and age‐bracketed evidence shows that the first Weichselian SIS advance extended further south in the eastern part of Fennoscandia than previously estimated, and that this ice growth phase occurred during the Early Weichselian Herning Stadial (MIS 5d). The subsequent ice growth phases occurred during the Middle (MIS 4) and Late (MIS 2) Weichselian substages. The lacustrine and littoral sediments of the Muhos core were correlated with the late Eemian interglacial (MIS 5e) and two Weichselian interstadials (MIS 5c and MIS 3).

The eastern part of Fennoscandia ( Fig. 1) was covered by the Scandinavian Ice Sheet (SIS) several times during the Middle and Late Pleistocene Svendsen et al. 2004;Johansson et al. 2011;Hughes et al. 2016). However, the exact number and timing of the SIS growth phases from the ice dome areas in the Scandinavian Mountains across Finland to northwestern Russia are not precisely known.
The most comprehensive study on the Pleistocene stratigraphy of northern Finland (i.e. Finnish Lapland) was conducted by Hirvas (1991). The study identified six stratigraphically significant till units overlying the Precambrian bedrock, thereby indicating six SIS growth phases across the area. Indeed, there are several sites in Finnish Lapland where till units are interbedded with sorted sands, gravels, organic-rich silt, clay/gyttja, and/or peat (Korpela 1969;Hirvas 1991;Saarnisto et al. 1999;Helmens et al. 2000;M€ akinen 2005;Lunkka et al. 2014;Salonen et al. 2014;Sarala et al. 2016). However, at several sites, only partial sedimentary sections or core materials have been preserved owing to erosion and glaciotectonic deformation during different glacial growth phases and the intervening interglacials.
Several studies have reconstructed the evolution of the SIS based on empirical data and modelling of the palaeoice sheet (Svendsen et al. 2004;Johnsen 2010;Olsen et al. 2013;Hughes et al. 2016;Batchelor et al. 2019). The different reconstruction techniques have often produced discrepant results, especially in terms of the extent and timing of the SIS coverage in its eastern flank (Sutinen 1992;Nenonen 1995;Saarnisto & Salonen 1995;Lunkka et al. 2004;Svendsen et al. 2004;Sarala 2005;Johnsen 2010;Johansson et al. 2011;Olsen et al. 2013;Hughes et al. 2016;Batchelor et al. 2019; see also Fig. 2), which is mostly because of the lack of empirical data used in reconstructions. Upon comparing the present stratigraphical evidence of the Pleistocene history of northeastern Fennoscandia (i.e. Swedish and Finnish Lapland and the adjacent areas in northwest Russia) with the southern and central parts of Finland and northwest Russia (Figs 1, 2), it is evident that the correlation between these two areas is not straightforward. The application of the Eemian sediments as a marker horizon for correlating Late Pleistocene sediments from these areas is complicated because parts of the Eemian sedimentary successions are often eroded, reworked and /or glaciotectonized. However, the best preserved Eemian sites can be used for stratigraphical correlation to provide evidence if the SIS overran northern Finland during the Early Weichselian substages while southern Finland remained ice-free (Johansson et al. 2011).
The aims of this work are (i) to establish the Pleistocene stratigraphy and palaeoenvironmental history of the Oulu River valley area, where exceptionally thick Quaternary deposits rest on the Precambrian bedrock and (ii) to correlate the glacial events in the study area to the adjacent areas, and through that shed light on the SIS ice extent in the eastern part of Fennoscandia.

Study area
The Muhos core was drilled in the Oulu River valley (latitude 64.81°N, longitude 26.04°E, altitude~29 m a.s.l.), situated at~30 km from the shoreline of the Baltic Sea in western central Finland (Fig. 1). The bedrock of the Muhos drill-core site is composed of low-lying Neoto Mesoproterozoic siltstone (the Muhos Formation; Tynni & Donner 1980;Tynni & Uutela 1984;Kesola 1985). The Muhos Formation is surrounded by higher ground where the bedrock is composed of Palaeoproterozoic granites, gneisses and schists (Fig. 3;Kesola 1985;Honkamo 1988). Late Pleistocene glacial deposits in the study area occur mostly as ground moraine.
Streamlined glacial landforms such as drumlins and ribbed moraines oriented WNW to ESE are also commonly found in the area (Lunkka et al. 2013;Fig. 3). In addition, discontinuous, WNW to ESE-oriented esker chains also occur in the Oulu River valley. The western side of the Muhos area is characterized by low-relief modern topography that resulted from Holocene sedimentation of varved clays, silts, and organic-rich silts in the Oulu River basin during the Holocene Ancylus and Litorina stages of the Baltic Basin (Fig. 3). To the north and east of the Muhos site, the topography is relatively steep and rises by 70 m within 6 km (Fig. 3). As a result of the postglacial uplift, glacifluvial landforms were gradually elevated to the water level of the Baltic Basin, and affected by Holocene littoral processes forming for example, beach ridges (Fig. 3).

Material and methods
Drilling, OSL sampling, and sediment logging The Geological Survey of Finland drilled a 54-m-deep sediment core at Muhos in 2006. The drilling site is located on the Muhos Formation~1 km southwest from the contact to granite bedrock (Fig. 3). The core (Ø  50 mm) was drilled using a GM 200 GT pneumatic drilling platform with 1-m-long opaque sample tubes (plastic HDPV liners, Ø 40 mm) inside stainless-steel cover tubes. Continuous sampling was initiated from the ground surface. At 53.6 m below the ground surface (b.g.s.), the Muhos Formation siltstone was reached, and drilling was terminated at 54.0 m b.g.s. After successful core extraction, the sediment type at the top and bottom of each tube was recorded on site, and subsequently, the sample tubes were sealed with opaque plastic caps. Sample tubes from sand units for OSL dating were wrapped in aluminium foil and placed in black plastic bags.
The entire sediment sequence from the ground surface to the bedrock was cored with a 77% recovery. This percentage includes sediments obtained intact in cores (61% of the 54 m depth), unrecovered diamicton inter-vals, and sediments that were too fluid to remain intact. Fluid sediments were collected in sample bags. During drilling, it was observed that thick diamicton units occur below 30 m. Diamicton was easy to recognize because of its significantly higher drilling resistance compared to the sorted sediments. As a result, samples from only certain parts of the thick diamicton units were extracted.
The core tubes were split longitudinally in the Oulu University laboratory. The OSL sample tubes were opened under a low-level orange light and grain size of sediments estimated. Four OSL samples were taken from the bedded sand units at 24.30, 24.70, 39.70 and 39.30 m levels. After OSL sampling, a detailed sediment logging was made and samples for grain-size analysis were collected (Fig. 4) ig. 3. DEM-imagery and bedrock map of the study area. The Muhos drilling site is located on the Muhos Formation siltstone. The Muhos Formation siltstone is surrounded by granitic-rocks, mica-rich schists and gneisses to the north, south and east, and granitoids to the west. Subglacial lineations and glacially abraded bedrock striations (Hakku service, digital data set of the Geological Survey of Finland, available at hakku.gtk.fi) are from two separate ice-flow directions; the older is from the NNW (dashed arrow) and the younger is from the WNW (solid arrow). The N-S and W-E land surface profiles crossing the drilling site (black dot on the map and in the profile) are shown in the top right corner of the figure. The low-relief topography rises relatively steeply to the north and east of the Muhos site (~70 m in 6 km). Bedrock map is modified from DigiKP, the digital bedrock database of the Geological Survey of Finland (available at gtkdata.gtk.fi/Kalliopera/index.html, DEM hillshade, National Land Survey of Finland).   OSL and 14 C dating The determination of OSL and 14 C AMS ages of the samples was carried out at the Nordic Laboratory for Luminescence Dating, Risø, Denmark (OSL samples) and the Poznan Radiocarbon AMS Laboratory in Poznan, Poland ( 14 C AMS samples).
In the OSL laboratory, the ends of the samples were retained for water content and dose rate analysis. The remaining portion was wet sieved to recover grains of 180-250 mm in diameter. This fraction was etched with HCl, H 2 O 2 , HF, and again with HCl to obtain a quartzrich separate (Wintle 1997). After chemical separation, the samples exhibited a detectable IR-stimulated signal, indicating residual feldspar contamination. As a result, during routine measurement, all blue light stimulation was preceded by infrared stimulation at 125°C (Wallinga et al. 2002).
OSL measurements were carried out using Risø OSL/ TL readers, each equipped with a calibrated 90 Sr/ 90 Y beta radiation source and a blue (470AE20 nm) light source (Bøtter-Jensen et al. 2000). The prepared quartz grains were attached in a monolayer to 10-mm-diameter stainless steel discs (with approximately a few thousand grains) using silicone oil. A SAR-protocol (Murray & Wintle 2000 was used to estimate all equivalent doses at 260°C for 10 s and a cut heat of 160°C. Dose rates were derived from radionuclide concentrations and were measured using high-resolution gamma spectrometry (Murray et al. 1987) considering conversion factors from Olley et al. (1996). These were modified by attenuation factors based on the observed sediment water content of 22-33% (Table 1). Finally, the internal alpha radiation contribution (assumed to be 0.06AE0.03 Gy; Prescott & Hutton 1994) was added to the dose rates. The results of the OSL measurements are listed in Table 1. One 14 C-AMS age determination was performed on a wooden twig found in the sand unit at 25.6 m ( Fig. 4, Table 1).
Eight samples collected from 38.74-38.82 m and 31 samples collected from 29.50-29.80 m depths were subjected to pollen analysis. Approximately 1-2 g of sediment was used per pollen sample, and sampling was performed continuously at 1-cm intervals. The samples were treated with heavy liquids using LST Fastfloat, lowtoxicity heteropolytungstates, at 1.95 g cm -3 (Eskola et al. 2021), boiled in 10% KOH, and were subsequently subjected to standard acetolysis. The concentrate was then mounted on silicone oil. Lycopodium tablets were added to the samples to calculate pollen concentrations. More than 500 terrestrial vascular pollen grains and spores were counted from each sample level whenever possible, at 4009 magnification. At least half of each slide was counted to account for the uneven distribution of pollen and spores in the sample (Hicks 2001). Pollen identification was mainly based on Beug (2004), Reille (1998Reille ( , 1999 and Moore et al. (1991). The terrestrial pollen sum (ΣP) was calculated based on trees, shrubs, dwarf shrubs and herbs. The percentage of aquatic plants, spores, and the green alga Pediastrum were calculated from the terrestrial pollen sum (ΣP) plus the sum of each group separately, for example, ΣP+aquatic plants. Unknown pollen were excluded from the pollen sum, and were counted separately (ΣP+unknown). The pollen diagram was drawn using Tilia software version 2.0.41 (Grimm 2018). Although pollen in most samples were well preserved, samples with broken and crumbled Table 1. List of the OSL and 14 C-AMS dating results from the Muhos sediment core where laboratory codes (OSL Risø and 14 C-AMS Poznan), dated sample codes and their material aswell as sample depth levels and sediment units are indicated. In the OSL data, dose and dose rate (equivalent dose, D e ) are listed. n = the number of individual aliquots contributing to D e ; w.c. = the saturated water content of the sediment samples. pollen and also thin-walled pollen were counted separately. However, they were included in ΣP.
An attempt was made to distinguish the Betula tree and Betula nana based on their size and morphological characteristics (Birks 1968;M€ akel€ a 1996). It is known that the age and laboratory treatment methods affect the pollen grain size (Reitsma 1969;Faegri et al. 1989). All samples included a portion of distinctly smaller-sized pollen than that reported in the literature, which in some cases hindered reliable identification.

Lithostratigraphy and dating results
The sediment succession above the bedrock is 53.6 m thick and can be divided into six lithostratigraphical units (Fig. 4).
Unit 1 (53.6-40.4 m b.g.s.) is a 13-m-thick massive diamicton (Fig. 4). Its colour varies between grey, red, and dark brown (Fig. 5A). Clasts are abundant in the red diamicton-type. The matrix of the upper, grey diamicton is coarser than that of the brown and red diamictontypes. Several superimposed diamicton beds are interpreted as lodgement and/or deformation till. However, it is impossible to evaluate whether each mineralogically and texturally distinct till bed corresponds to deposition during one or several glaciations.
Unit 2 (40.4-37.4 m b.g.s.) is mainly composed of parallel-bedded, coarse, medium and fine sands, including pebble gravel horizons, layers of laminated silt, and organic-rich silt (Fig. 4). The lower contact of Unit 2 is sharp. The lower part consists of coarse to medium sand, which fines upwards into laminated and massive organicrich silt with a subordinate amount of clay (Fig. 5B). Parallel-bedded sand layers are 0.5-1.0 cm and the average silt laminae 0.2 cm thick. There was a massive (at places faintly laminated and deformed) fine to medium, poorly sorted sand layer above the organic-rich layer (Fig. 5B). The OSL ages of the medium and coarse sand (including fine gravel horizon) are 137AE9 and 146AE10 ka, respectively (Fig. 4, Table 1). Unit 2 was deposited in a lake basin. Parallel-bedded, coarse and medium sand could represent littoral, beach, or distal deltaic sands deposited in the basin, whereas fine sediments were deposited in deeper water. There is a massive and faintly stratified, fine to medium sand layer with subordinate amounts of silt overlying the laminated and organic-bearing silt layer, which could represent a massflow deposit owing to its massive structure, or a subglacially deformed sand and silt because it occurs below the diamicton unit. The latter interpretation is considered more likely.
Unit 3 (37.4-29.8 m b.g.s.) is more than 7 m thick and consists of massive sand diamicton (Figs 4 and 5C). Its basal contact was not preserved in the drill-core; however, the contact was defined by the changes in drilling resistance at 37.4 m. Matrix supported and massive diamicton with striated clasts is interpreted as lodgement or deformation till. It is most likely that the massive sand layer (in the upper part of Unit 2) beneath the till layer (Unit 3) is glaciotectonically homogenized and faintly deformed.
Unit 4 (29.8-18.6 m b.g.s.) is~10 m thick (Fig. 4). It conformably overlies the diamicton (Unit 3). The basal part of the unit (29.83-29.55 m) is characterized by laminated clay, silt, and parallel-bedded fine sand layers with organic contents of less than 2% (Fig. 5D). A few small grains (3-5 mm in diameter) were also observed in the laminated basal unit. This laminated subunit grades upward into a 3.5-m-thick (29.5-26.0 m) massive and/or deformed (clay) silt and sand layer with occasional pebbles (Fig. 5E). Wood fragments are present in the deformed fine sand and silt between 25.5-25.6 m. Ripple and parallel-bedded sands and laminated silt occur between 25.0 and 19.0 m ( Fig. 5F) above the deformed sand/silt unit. Unit 4 terminates with coarse sand, including two gravel layers occurring at 20.1 and 20.4 m that fine upwards into parallel-bedded sand between 20.1 and 18.6 m. A wood fragment at 25.5 m yields an infinite 14 C-AMS age (>50 ka) and two OSL ages determined at 24.2 m and 24.5 m yield ages of 107AE6 and 110AE6 ka, respectively (Fig. 4, Table 1). The laminated clay silt layer with parallel-bedded fine sand and occasional small clasts at the base of Unit 4 were deposited in a freshwater basin, probably in a glacial lake. The laminated structures, grain size, and low organic content suggest that the sediments were probably deposited in a relatively deep basin. The upward-fining massive and deformed sand and silt sets, including sporadic clasts overlying the laminated sediments, probably represent syndepositionally deformed lacustrine/ glaciolacustrine sediments. Two cycles of parallelbedded and ripple-bedded sands fining upward into deformed sands and silts indicate two sediment pulses into the lake basin. Overall, the upward-coarsening trend in the sediment succession indicates a lake basin infill and water level regression.
Unit 5 (18.6-13.2 m b.g.s.) is composed of diamicton, interbedded with 0.6-m-thick parallel-bedded sand and laminated silt (Fig. 4). The lower contact of the unit is gradational, and the underlying sand and silt in the top 0.3 m of Unit 4 deformed. The base (18.6-17.2 m) consists of weakly stratified sandy diamicton (subunit 5a in Figs 4, 5G), which grades into parallel-and ripple-bedded sand and laminated fines (17.2-16.6 m). This unit is overlain by massive sandy diamicton (16.6-13.3 m). The diamicton beds (subunits 5a and 5c in Fig. 4) are interpreted as lodgement tills. The contact between the sand beneath (top of Unit 4) and the lower diamicton in Unit 5 exhibits deformation similar to the contact between the upper till (subunit 5c) and the underlying sand (subunit 5b). Such deformed structures might have a glaciotectonic origin caused by ice overriding the area twice, although it is difficult to deduce this from core exposure only. The genesis of partially tectonized, originally parallel-and ripple-bedded sands and fine silt lamina interbedded between two till units is difficult to interpret. Since sand layers (<1 cm thick) and silt laminae (a few millimetres thick) seem to alternate quite regularly, it is plausible that the sandsilt unit between the tills was deposited in a lacustrine rather than in a fluvial or aeolian environment, although the layer might have undergone deformation at a later stage.
Unit 6 (13.3-1.0 m b.g.s.) consists of an upwardfining sequence from gravel to laminated organic and clay-rich silt (Fig. 4). The lower contact was sharp. The base contains stratified gravel (13.3-12.9 m) overlain by a stratified mixture of deformed and laminated sand and silt with occasional clasts and clusters of clasts within a sand-silt matrix (12.9-11.0 m). Sand and silt with a few pebbles occur between 11.0-9.9 m overlain by deformed, laminated, parallel-and ripple-bedded sand and silt with isolated and clustered clasts (9.5-5.7 m) and dropstones (Fig. 5H). The top part of the unit is composed of laminated silt with occasional sulphide-  rich layers, and the organic content increased towards the top of the core. The upward-fining sediment sequence (Unit 6) from gravel and sand at the base to organic-bearing silt/clay represent a typical succession indicating deglaciation in an ice-contact glaciolacustrine basin, which subsequently became ice-free and part of the Baltic Basin waterbody (Ignatius et al. 1981). These sediments are thought to have been deposited in the Ancylus Lake and Litorina Sea of the Baltic Basin during the Holocene. The top 1 m of the core is composed of modern soil.

Diatom and pollen analyses and their environmental interpretation
Diatom assemblage. -Diatoms in Unit 2 (38.77 and 38.79 m) were heavily fragmented; however, the species or genus level could be still identified. Only the small taxa were found to be intact, and most of the other taxa were fractured. The diatom assemblage was composed mostly of planktonic freshwater taxa, dominated by Aulacoseira islandica and its resting spores (48-53%). Other Aulacoseira species, such as A. granulata, A. subarctica, A. ambigua and A. lirata, were also present. Several species of small Fragilariaceae, such as Staurosira construens, S. venter, Staurosirella pinnata, S. lapponica, Pseudostaurosira brevistriata, Stephanodiscus rotula, Cyclotella schumannii and Achnanthes oestrupii, were preserved. Sporadic frustules of Fragilaria leptostauron var. martyi, Eunotia faba, Cocconeis spp., Cyclotella iris, Cymbella sinuata, Cymbella spp., Stephanodiscus minutulus and Tetracyclus glans were identified. Several frustules and fragments of planktonic freshwater species Cyclotella omarensis and Ellerbeckia arenaria, Epithemia spp., Diploneis spp., Didymosphenia geminata, Pinnularia spp., Stauroneis spp., Stephanodiscus spp. and Gomphonema spp. fragments were also identified ( Table 2). The identified diatom assemblage entirely belongs to freshwater taxa, with the exception of a single frustule of the brackish water species, Mastogloia smithii. The diatom taxa are mostly planktonic, suggesting a deepwater environmental setting, and are comparable to the Holocene Ancylus Lake-phase diatom taxa of the Baltic Basin (Gr€ onlund 1991a). The considerable presence of the cold-water species A. islandica and its resting spores indicate unfavourable conditions, such as limited light and temperature (Round et al. 1990;McQuoid & Hobson 1996). The small planktonic taxon C. schumannii has been reported from low-conductive lakes in northern Finland and Sweden, large lakes, mountain lakes, rivers (Cleve-Euler 1951; H akansson 2002) and cold and deep lakes of Finnish Lapland (Sorvari et al. 2002). Small benthic Fragilariaceae (Staurosira construens, S. venter, Staurosirella pinnata, S. lapponica, and Pseudostaurosira brevistriata) are common species in alpine, sub-arctic and arctic lakes (Pienitz et al. 1995;Weckstr€ om et al. 1997;Laing et al. 1999;Lotter & Biegler 2000;R€ uhland et al. 2003). They could be also associated with colder and less productive conditions or lowering of the water level (R€ uhland et al. 2003), although small benthic Fragilariaceae are known to be environmentally tolerant taxa.
No diatom fragments or valves were found in the reddish clayey silt and silt sediment samples in Unit 4 at depths of 29.58, 29.66 and 29.79 m. The absence of diatoms might be associated with unfavourable conditions for diatom growth.
Since the laminated sediments at the base of Unit 4 were devoid of diatoms, we could not interpret the water quality into which laminated silt/clay/fine sand sediments were deposited. However, the laminated sediments normally accumulate in large basins with water depths exceeding 40 m (Ojala et al. 2013). Since the laminated sediments at the base of Unit 4 rest conformably on diamicton, it is likely that they were laid down in a glaciolacustrine setting.
An in-depth interpretation of the vegetation history of the region is difficult based on only an 8-cm sediment interval at the base of Unit 2. However, the pollen assemblage suggests that mixed forest with deciduous (birch and some alder) and coniferous trees (spruce and pine) existed around the freshwater basin. The sporadic occurrence of thermophilous tree pollen further complicated the interpretation, although it is most likely that their occurrence might indicate either long-distance transport or redeposition. The presence of heavy spores of Osmunda indicated temperate climate because it grows in temperate and tropical areas worldwide (Birks & Paus 1991). Wetland taxa such as Cyperaceae, Poaceae and cryptogams were well preserved and were   s   680  641  571  347  621  602  580  256  181  619  606  522  376  596  565  660  638  590  642  551  596  648  761  665  521  568  342  539  508  694  The pollen assemblage in Unit 4 at sample depths of 29.55-29.83 m consists of trees (74-91%), shrubs (1-9%), dwarf shrubs (1-4%) and herbs (3-15%) (Fig. 6B). The major tree taxa are Betula (60-79%), Alnus (5-14%) and Pinus (2-9%). Picea is constantly present with low percentages (0.2-1.7%), and other trees such as Carpinus, Quercus, Sorbus, Tilia and Ulmus occur as single grains. Betula nana is the most common shrub with a value ranging from 0 to 7.6% and Salix has an occurrence of up to 3%. Vaccinium and Ledum are the most common dwarf shrubswith low percentages, and Empetrum, Calluna and Bruckenthalia spiculifolia occur as single grains. Cyperaceae (0-8%), Poaceae (0-6%), Artemisia (0-2%), Filipendula and Chenopodiaceae are the most common herb taxa. The most common aquatic species are Potamogeton and Myriophyllum, with sporadic presence of Sparganium. The freshwater alga, Pediastrum, is low in abundance. Sphagnum, Polypodiaceae, Equisetum, Lycopodium and the aquatic plant Isoetes comprise the majority of the spore-forming taxa. The percentage of worn and degraded pollen and spore grains ranged from 1-26%, and the maximum percentage of thin-walled grains was 21%. The majorityof these degraded and thinwalled pollen are Betula-type pollen grains, and it may also include Betula nana pollen. The total concentration of pollen and spores varies from 7000-51 000 grains g -1 , which is merely a tenth of the concentration at sampling depth of 38.74-38.81 m.
The pollen assemblage suggests taxa typical of interstadial phase vegetation around the basin where sediments were deposited. A relatively high abundance of Betula tree and Alnus, a relatively low percentage of nonarboreal pollen (NAP), and low abundance of Betula nana, Salix, Artemisia and Poaceae suggest a forested environment with birch, alder, and perhaps scattered pine. It is likely that the sporadic pollen-grain occurrences of thermophilous trees such as Carpinus, Quercus, Tilia and Ulmus reflect their long-distance origin or redeposition. The presence of Salix, Cyperaceae, Filipendula, Poaceae and cryptogams reflects the type of vegetation that existed on the shore of the basin.

Discussion
The sedimentary succession of the Muhos core exhibits four till units interbedded with sands and less abundant silt, clay, gravel, and two organic-rich intervals. Four till units (Units 1, 3, 5a and 5b; Fig. 4) indicate that the Muhos and adjacent areas were covered by the SIS four times prior to the Holocene, and the Bothnian Bay sediments (Unit 6) were deposited during this period.
The lithological and mineralogical composition of the lowermost till complex (i.e. Unit 1) has been previously studied by Lunkka et al. (2013). The study shows that the dark brown till beds containing abundant mica schists, mica gneisses, and mafic volcanic clasts reflect a northern provenance of the clasts, i.e. ice movement was from the north. In contrast, the grey and red till beds containing abundant granite and siltstone clasts indicate a western provenance, i.e. the ice flow was from the west to the Muhos area. Two OSL samples from Unit 2 situated above the till complex (Unit 1) yielded ages of 137AE9 and 146AE10 ka, suggesting that the till complex beneath Unit 2 was deposited during the late Saalian (MIS 6) glaciation (Fig. 7). However, it is also plausible that the till complex (Unit 1) represents till beds deposited not only during the late Saalian glaciation but also during earlier glacial phases.
Unit 2, with littoral sands at the base overlain by deepwater organic fine-grained sediments and deformed sands above (Fig. 4), was deposited in a lake. The OSL results indicate that the maximum age for the deposition of littoral sands is 137AE9 ka. The presence of freshwater diatom taxa in the laminated silt supports the environmental interpretation. In addition, the presence of a green alga, Pediastrum, throughout the laminated silt layer reflects sedimentation in a lacustrine basin. Pediastrum, along with the abundant planktonic taxon A. islandica and small benthic Fragilariaceae taxa, indicate that a cool, non-saline lake existed when the laminated sediments were deposited. Although the dominance of resting spores of A. islandica might not necessarily indicate a deep-water environment (Peltoniemi et al. 1989), the presence of laminated silt rhythmites and planktonic diatom taxa indicate deposition in a relatively deep lacustrine environment.
Western Finland, including the Muhos area, was inundated by the Eemian Sea during the Eemian interglacial Gr€ onlund 1991b). The estimated water level of the Eemian Sea in central western Finland adjacent to the Gulf of Bothnia was 110 m a.s.l. ; Fig. 1). Since there are no marine diatom taxa apart from a single frustule of a brackish water taxon (i.e. Mastogloia sp.), the silt and clay interval in Unit 2 cannot be correlated with the high water level Eemian Sea phase. Therefore, it is suggested that the laminated silt and clay were deposited during the early or late Eemian freshwater phase, which either preceded or post-dated the Eemian Sea phase. However, the pollen data contradict the correlation with the early Eemian freshwater phase. The pollen assemblage includes abundant conifers and grains of thermophilous trees with sporadic occurrence of Osmunda. However, the pollen taxa are not dominated by birch, as is the case for the pollen taxa in Haapavesi (Vesiper€ a site), Per€ asein€ ajoki (Viitala site) (Nenonen et al. 1991) and Ylivieska (Mertuanoja site) ; Fig. 1), which represent early Eemian freshwater lacus-BOREAS trine sediments with diatom taxa similar to those of Unit 2 at Muhos.
Pollen and diatom evidence from the Ukonkangas site, K€ ars€ am€ aki (Fig. 1), only 100 km south of the Muhos site, suggests that the upper sediment units at the Ukonkangas site were deposited during the end of the temperate Eemian interglacial (Gr€ onlund 1991b;Eriksson 1993). The pollen data from the silt-rich sediments show similar characteristics as those studied at Muhos, with the exception of slightly lower spruce and pine and greater abundance of alder. Quercus and Corylus are ubiquitously present at the Ukonkangas site, whereas their occurrence in Muhos is sporadic. The diatom taxa from the sediments of the Ukonkangas site show a transition from the brackish Eemian Sea to freshwater conditions (Gr€ onlund 1991b). Therefore, the lake sediments in Unit 2 at Muhos can be correlated to the late Eemian lacustrine sediments of the Ukonkangas site.
The littoral sands at the base of Unit 2 with the OSL age of 137AE9 ka (Fig. 4) may represent sands deposited during the transition from the Saalian glaciation to the Eemian interglacial. However, the OSL age uncertainty due to systematic and random errors, such as a spread in equivalent doses and uncertainty in water content fluctuations and burial history, can be significantly large (higher than 10%) (Wallinga & Cunningham 2015). Therefore, it is tentatively suggested that these sands represent littoral deposits that were laid down when the lake basin isolated from the Eemian Sea due to glacioisostatic uplift. The silt and clay above represent deeper freshwater facies of the basin that was subsequently uplifted and filled with sands (Figs 4, 7). Lodgement till (Unit 3) indicates that the SIS overrode the Muhos region. Although the application of till clast lithology to trace the direction of ice movement is ambiguous, a significantly high proportion (>50%) of mica gneiss, quartzite, and mafic volcanic clasts in this till unit (Lunkka et al. 2013) might reflect ice movement from the N or NW. Since the sands in Unit 4 have ages of 107AE6 and 110AE6 ka (Fig. 4, Table 1) and the lake sediments in Unit 2 were deposited during the late Eemian, it is most likely that the till (Unit 3) was deposited by the SIS during the Early Weichselian Herning stadial (MIS 5d). This indicates that the SIS had extended further to the south than previously thought (Sutinen 1992;Nenonen 1995;Saarnisto & Salonen 1995;Lunkka et al. 2004;Svendsen et al. 2004;Sarala 2005;Johansson et al. 2011;Olsen et al. 2013;Batchelor et al. 2019 ; Fig. 7).
The glaciolacustrine sediments in Unit 4 exhibit a succession that represents deep to shallow water facies (Fig. 4) A g e ( k a B P ) N o r t h w e s t E u r o p e a n ( N o r t h A m e r ic a n ) s t a g e s the base most likely represent rain-out from melting icebergs in the glaciolacustrine basin or grains that were transported by lake ice during melting seasons. The occurrence of deformed sediments above the laminated unit indicates that the sediment source became more proximal with intermittent input events of coarse sediments. For example, input from collapsed deltaic sediments or riverine input may have caused syndepositional deformation in finely laminated sediments deposited during quiet periods. Two OSL ages and one infinite 14 C age (Table 1) obtained from the gravel and sand layer above the basal rhythmites, and the deformed beds suggest that the laminated deep-water sediments were deposited during the early part of the Brørup interstadial (MIS 5c) (Fig. 7).
The pollen content of the fine sediment interval at the base of Unit 4 is dominated by birch with subordinate pine and alder (Fig. 6B). The pollen assemblage is monotonous throughout the sediment interval, and therefore it is difficult to correlate with other subtill sediment pollen assemblages in the region. Birchdominated subtill, organic-bearing pollen sites with minor (or absent) pine, spruce and alder have been reported from Pudasj€ arvi (Katosharju site ;Sutinen 1992;Sarala et al. 2006) and Taivalkoski (Kostonniska site; Korpela 1969) in the northern part of the North Ostrobothnian Province and at several sites in southern Lapland (Permantokoski, Ossauskoski and Kauvonkangas sites; Fig. 1; Korpela 1969;M€ akinen 2005). The interstadial-type sediments at these sites represent the Per€ apohjola interstadial (Korpela 1969), which was previously correlated with the Early Weichselian interstadials Brørup (MIS 5c) or Odderade (MIS 5a). However, later studies have suggested its correlation with the Middle Weichselian interstadials (MIS 3; Johansson et al. 2011). The pollen assemblage of Unit 4 at Muhos differs from the pollen assemblage of the Per€ apohjola interstadial sites in terms of higher relative proportions of alder and conifers and a lower proportion of NAP.
South of the Muhos site, the birch-dominated pollen assemblages from the organic-rich sediments of the Oulainen (Vuojalankangas site) (Forsstr€ om 1982(Forsstr€ om , 1988Nenonen 1995), Pyh€ asalmi (Ruotanen site; Nenonen 1995), Vimpeli (Vimpeli I site; Aalto et al. 1983Aalto et al. , 1989 and Teuva regions (Horonkyl€ a site ;Nenonen 1995;Gr€ onlund & Ikonen 1996) (Fig. 1) were deposited during the Early Weichselian Brørup interstadial (Donner 1995;Nenonen 1995). The pollen assemblage of the laminated sediment interval in Unit 4 cannot be directly correlated with the pollen assemblages of the Brørup interstadial sites mentioned above. However, at Sokli (Fig. 1), a composite sediment core covering the last interglacialglacial cycle is preserved (Helmens et al. 2000), and the initiation of the MIS 5c (Brørup) (Sub-zone II a; Helmens et al. 2012) gyttja shows a pollen assemblage similar to that at Muhos. This indicates that a sub-arctic birch forest with alder and possible conifers was present in the Sub-zone II a, i.e. during the beginning of the MIS 5c. The higher proportion of Alnus at Muhos may also indicate moist conditions around the local lake basin.
It is evident that the fine-sediment interval at the base of Unit 4 partially exhibits an interstadial succession. Furthermore, the counts of worn and thin-walled pollen are relatively large, and they might include Betula nana. Therefore, the redeposition of at least some pollen and spores cannot be disregarded. Based on the overall sedimentary succession, pollen, and the OSL age, we suggest that the fine sediment interval and the associated sands and gravels were deposited during the Brørup interstadial.
Unit 5 is a composite unit consisting of two till beds (subunits 5a and 5c in Fig. 4) interbedded with paralleland ripple-bedded sand and laminated silt (subunit 5b in Fig. 3). This sedimentary succession indicates that there were possibly two ice advances across the area separated by one ice-free period (Fig. 7). Based on lithological analyses of tills in Unit 5, Lunkka et al. (2013) suggested that the clasts in the lower till bed (subunit 5a) are predominantly mica schists and mica gneiss, while granite clasts dominate the lithological composition of the upper till bed (subunit 5c). This difference, together with ice flow direction data (cross-cutting bedrock striations and cross-cutting landform lineation patterns; see Fig. 3) indicate that there was an older ice advance from the north, and subsequently a younger one from the NW direction. In the Hangaskangas, Katosharju and Ajos sites situated at 15 km NW, 50 km NE and 120 km NW of the Muhos site, respectively (Fig. 1), sands beneath the uppermost till are of Middle Weichselian age (OSL ages 41.6, 49.2 and 46 ka, respectively), and the till clast fabric results indicate that the uppermost till was deposited during awesterly ice-flow phase (Sutinen 1992;Bargel et al. 1999;Pasanen & Lunkka 2008). Therefore, we tentatively correlate the lower till (subunit 5a) with the first Middle Weichselian ice growth phase (MIS 4), the sand and silt (subunit 5b) between the two till beds with the Middle Weichselian interstadial (MIS 3), and the upper till (subunit 5c) with the Late Weichselian ice growth phase.
Unit 6, with its basal conglomerate grading upwards into mass movement and sub-aqueous glacifluvial and glaciolacustrine sediments, represents a typical sedimentary succession deposited during the Ancylus Lake phase of the Baltic basin as the ice retreated from the area (Ignatius et al. 1981). Organic-and sulphide-rich laminated and massive silt above the Ancylus sediments most likely represents the Litorina stage sediments of the Baltic Basin.
Overall, the sediment successions at Muhos and the adjacent sites demonstrate that after the Saalian glaciation, the Eemian Sea inundated the coastal areas of the Gulf of Bothnia. Due to the glacio-isostatic uplift following the Saalian glaciation, the Muhos area was eventually uplifted to the contemporary Eemian Sea level. At that time, littoral processes must have eroded the existing Eemian Sea sediments, and sands and gravels remained as littoral deposits in the area (Nenonen 1995). Subsequently, a lake basin was isolated from the Eemian Sea and the lacustrine sediments in Unit 2 were deposited.
The first ice advance into the Muhos area in the Weichselian took place during the Herning stadial (MIS 5d). According to earlier studies (Lunkka et al. 2013), this ice advanced from the north. However, the extent of this ice south of the Muhos region is uncertain. Since a till unit deposited during the Herning stadial occurs at the Muhos site and in western Lapland (Rautuvaara and Hannukainen sites), but not in eastern Finnish Lapland (Sokli site), the ice margin must have covered large areas in western Lapland and central Finland (Fig. 7). There has been no evidence of a Herning stadial ice advance to the central and southern parts of Finland. Therefore, the till (Unit 3) between the late Eemian lacustrine sediments (Unit 2) and the dated sands (Unit 4) at Muhos is the first stratigraphically controlled and age-bracketed evidence of an ice advance into central Finland during the Early Weichselian Herning stadial. These results suggest that the extent of the ice sheet was underestimated in former studies, at least in the southeastern flank of the SIS during the Early Weichselian (MIS 5d) substage.
A lake, probably a part of a large glacial lake in the Bothnian Bay of the Baltic Basin existed in the Muhos area during the early part of the Brørup interstadial and the SIS covered the area three times after the Brørup interstadial. Although the biostratigraphical and geochronological evidence from the Muhos core sediments is limited, information from adjacent sites such as Kauvonkangas (M€ akinen 2005), Hannukainen (Howett et al. 2015) and Mertuanoja (Nenonen 1995) suggests that the SIS covered the area twice during the Middle Weichselian (MIS 4 and MIS 3) and once during the Late Weichselian (MIS 2) stage (Fig. 7).

Conclusions
• The 54-m-long Muhos sediment core studied in the Oulu River valley, central western Finland, situated adjacent to the centre of the SIS, displays multiple till units interbedded with organic-rich clays, silts, sands and gravels. The core covers a time-span from the Saalian (MIS 6) glaciation to the Holocene. • The results of the Muhos core indicate that there were at least four ice advances across the Muhos area. Based on the OSL results and biostratigraphical evidence, the oldest ice advance took place during the late Saalian stage, and the subsequent advances during the Early Weichselian, Middle Weichselian and Late Weichselian.
• The sediment record indicates that a freshwater basin existed in the Muhos area during the late Eemian interglacial, Brørup interstadial, and possibly also during the Middle Weichselian interstadial as well as during the Holocene Ancylus Lake phase of the Baltic Basin. • In contrast to the previous estimations of the SIS extent during the Early Weichselian, the results of this study show that the SIS overran the Muhos area from the north and advanced into central Finland during the Early Weichselian Herning stadial.