The Little Ice Age : evidence from a sediment record in Gullmar Fjord , Swedish west coast

Introduction Conclusions References

The Little Ice Age: evidence from a sediment record in Gullmar Fjord, Swedish west coast I. Polovodova Asteman 1 , K. Nordberg 1 , and H. L. Filipsson

Introduction
A number of proxy-based reconstructions show that over the last millennium several intervals with anomalously cold summers were characteristic of the period, well-known as the Little Ice Age or LIA (among others: Hughes and Diaz, 1994;Fricke et al., 1995;Bianchi and McCave, 1999;McDermott et al., 2001;Nordli, 2001;Xoplaki et al., 2005;Figures Back Close  Lund et al., 2006;Leijonhufvud et al., 2010;Trouet et al., 2009;review in Ljungqvist, 2009review in Ljungqvist, , 2010;;Cage and Austin, 2010;Sicre et al., 2011).The LIA was associated with a widespread expansion and subsequent fluctuation of glaciers in the Arctic and alpine regions at lower latitudes, which happened in response to climatic changes at ∼ AD 1300-1850 (Porter, 1986;Miller et al., 2012).This climatic deterioration coincided with reduced solar activity (Mauquoy et al., 2002), regular decline in summer insolation as Earth moved steadily farther from the sun during the Northern Hemisphere summer (Wanner et al., 2011), increased volcanism (Miller et al., 2012) and weakening of the North Atlantic Oscillation after its strongly positive state during the Medieval Warm Period (Trouet et al., 2009).Some studies also associated the cooling with a decreased volume transport of the Gulf Stream (Lund et al., 2006).
There is growing evidence that the Little Ice Age occurred throughout the Northern Hemisphere (Moberg et al., 2005;Ljungqvist, 2009Ljungqvist, , 2010) ) and is distinctly seen as taking place during the last three to four centuries with a termination at ∼ 1900 AD (Grove, 2001).Other studies, however, mentioned that the LIA started earlier, somewhere between ∼ 1300 and 1400 AD (Fricke et al., 1995;Mauquoy et al., 2002;Moberg et al., 2005;Cage and Austin, 2010;Ljungqvist, 2010;Miller et al., 2012).The period 1675-1710 AD has been identified as the coldest phase (climax) of the LIA (Lamb, 1983), which was preceded by an abrupt and rather short-lived warming at 1540-1610 AD, according to a proxy-based reconstruction from the Scottish fjords (Cage and Austin, 2010).
The high-resolution fjord sediment records produce archives, which contain information about variations in marine, terrestrial and atmospheric environments (Howe et al., 2010) and therefore allow studying the air-ocean-land interactions in a great detail.The Skagerrak region with its fjords is a key area for investigation of such complex interactions: it acts as a main depositional basin for the North Sea; it is highly influenced by the North Atlantic Oscillation, a dominant pattern of atmospheric circulation over the North Atlantic; and it has been classified as a coastal area with a high cumulative human impact (Halpern et al., 2008).From the few marine records containing the Little Introduction

Conclusions References
Tables Figures

Back Close
Full Ice Age in the Skagerrak and attempting to link the air-ocean or land-sea interactions, the greatest interest in terms of high-resolution LIA stratigraphy represent studies of Hass (1996Hass ( , 1997)), Br ückner and Mackensen (2006) and Erbs-Hansen et al. (2011) (Fig. 1).Some of the previous studies, however, included 14 C dating obtained from the mixed benthic foraminiferal faunas and therefore may potentially contain some chronological bias.This paper represents the first well-dated and high-resolved study of the Little Ice Age in the sedimentary archives from the Scandinavian coast.It aims to reconstruct environmental conditions during the LIA and contribute to a better understanding of the climate variations during the last millennium in the Northern Europe.Using lithology, bulk sediment geochemistry, benthic foraminifera and carbon stable isotopes we intend to reconstruct the various phases of this cold period; to identify its timing (onset, climax and termination) in the study area and to correlate our marine data with terrestrial records.
The Gullmar fjord has a sill area at a depth of 42 m and a maximum basin depth of 120 m.The water masses are stratified with respect to salinity and density into four layers (Fig. 1c).At the surface occurs a first thin layer (< 1 m) of river water from Örekils älven, which does not influence the hydrography of the fjord in any significant way (Arneborg, 2004).The residence times for the second and third layers are 20-38 days and 29-60 days, respectively (Arneborg, 2004), whereas the deepest layer has a residence time of approximately one year (Nordberg et al., 2000;Erlandsson et al., 2006).The long residence time of the deep water and high oxygen consumption rates in the fjord provide a likelihood of hypoxia (< 2 ml O 2 l −1 ) in the basin (Inall and Gillibrand, 2010).Indeed, during the 20th century oxygen consumption in the fjord increased from 0.21 ml O 2 l −1 month −1  to about 0.35 ml O 2 l −1 month −1 (1970s and the early 1980s) and to about 0.41 ml O 2 l −1 month −1 from 1989 onward (Erlandsson et al., 2006).Given that concentration of dissolved oxygen during a bottom water renewal usually ranges between ∼ 4 and 8 ml O 2 l −1 (Filipsson and Nordberg, 2004) and the residence time of bottom water is 1 yr, the annual consumption rates may therefore remove all the oxygen before the next successive bottom water renewal takes place.After 1973, however, the timing of the renewal of the basin water in Gullmar Fjord occurs approximately 20 days later compared to the period prior to 1973, which is suggested to be a result of climate variations triggered by North Atlantic Oscillation (NAO) (Erlandsson et al., 2006).Therefore, annual stagnation periods and temporary hypoxic events, characteristic features in Gullmar Fjord, became significantly stronger during the second part of the 20th century (Filipsson and Nordberg, 2004;Nordberg et al., 2009 and references therein) but likely were less frequent/severe during the Medieval Warm Period (Polovodova et al., 2011).Introduction

Conclusions References
Tables Figures

Back Close
Full

Methods
This study is based on three sediment cores: GA113-2Aa, 9004 and G113-091, which were all taken at 116 m water depth in Gullmar Fjord (58 • 17.57N, 11 • 23.06 E) (Fig. 1).The cores 9004 (731-cm-long) and G113-091 (195-cm-long) were sampled with a piston corer during the expeditions of the R/V Svanic and R/V Skagerrak in 1990 and 2009, respectively.The core GA113-2Aa (60-cm-long) was taken with a Gemini corer aboard of the R/V Skagerrak in 1999.Together cores GA113-2Aa and 9004 represented a continuous sediment record with a small gap in-between due to difficulties with the piston corer methodology in relatively soft sediments.In order to find out the size of the gap, core G113-091 was used.
In the laboratory, all cores were split in two halves.One of the halves was saved for dinoflagellate cyst analysis (for the cores G113-091 and 9004 -to be presented elsewhere; for core GA113-2A see Harland et al., 2006).The other half was used for foraminiferal and geochemical studies.Twelve mollusc shells (Table 1, Fig. 2) were found in life position in cores G113-091 and 9004 and were subjected to radiometric 14 C AMS dating.The shells were dated at the Ångstr öm laboratory (Uppsala University, Sweden) using the marine model calibration curve (Reimer et al., 2004;Bronk Ramsey, 2005).The half-life used is 5568 yr, and the margin of error is 1σ.Ages are normalized to δ 13 C of −25 ‰ according to Stuiver and Polach (1977), and a correction corresponding to δ 13 C = 0 ‰ (not measured) versus PDB has been applied.Based on the latest complementary datings radiocarbon dates were corrected using a reservoir age of 500 yr (∆R = 0) (Nordberg and Possnert, unpubl. data;Polovodova et al., 2011).For dating of core GA113Aa and details regarding the 210 Pb dating method, see Filipsson and Nordberg (2004) and Nordberg et al. (2001), respectively.Prior to foraminiferal studies, core 9004 was sliced into 1 cm intervals down to a depth of 28 cm and thereafter into 2 cm intervals.Core G113-091 was cut into 1 cm slices until 19 cm, and thereafter 2 cm slices were taken.The same preparation technique was used for all samples: they were weighed, freeze-dried and their water content was Introduction

Conclusions References
Tables Figures

Back Close
Full determined.The measurements of organic carbon (C org ), total nitrogen (TN) in the bulk sediment and stable carbon isotopes on tests of benthic foraminifer Cassidulina laevigata were performed for the cores GA113-2Aa and 9004.All organic carbon samples from the core GA113-2Aa were analysed at University of Gothenburg (Sweden) using a Carlo Erba NA 1500 CHN analyser.Samples from the core 9004 were run at Bremen University (Germany) using a Vario EL III CHN analyser.In core G113-091 we ran δ 13 C measurements on tests of C. laevigata and conducted a benthic foraminiferal analysis in order to determine the gap between the cores GA113-2Aa and 9004.All δ 13 C measurements were run at the Department of Geosciences, University of Bremen using a Finnigan Mat 251 mass spectrometer equipped with an automatic carbonate preparation device.All foraminiferal samples were washed over 63 µm and 1 mm sieves and treated with sodiumdiphosphate (Na 2 P 2 O 7 ), where necessary, in order to disintegrate sediment aggregates.In order to obtain a rough estimate of the sand content, the size fractions > 63 µm were dried at 50 • C and weighed.The foraminifera-rich samples were split and at least 300 specimens were counted in each sample from the dried > 63 µm fraction.Inner organic linings were counted separately.Planktonic individuals were counted but not identified at the species level.The partially destroyed foraminiferal tests lacking more than one chamber were counted separately in order to get a rough estimate of shell loss/carbonate dissolution.In total, 262 samples were analysed for foraminiferal fauna, here we report 72 samples.Both relative (%) and absolute abundances (ind.g −1 dry sed.) for the benthic species were determined.Faunal diversity is estimated as Fisher α-diversity (Murray, 2006), whereas species number represents the total amount of species recovered per sample.In order to distinguish the foraminiferal units with different species dominance we performed a simple CABFAC factor analysis with Varimax rotation using the software PAST, University of Oslo (Hammer et al., 2001).The analysis has been run on raw foraminiferal data with relative abundances > 5 %.
Results of C org and δ 13 C measurements from the cores GA113-2Aa and 9004 are given in Filipsson andNordberg (2004, 2010).For the results on benthic foraminifera Figures

Back Close
Full and dinoflagellate cysts analyses from the core GA113-2Aa see Filipsson and Nordberg (2004) and Harland et al. (2006), respectively.The part of the benthic foraminiferal record covering the Medieval Warm Period (core 9004: 350-500 cm) was previously published in Polovodova et al. (2011).In the current paper we concentrate upon the last 1000 yr of the high-resolution record with a special emphasis on the Little Ice Age.

Chronology
The radiocarbon (AMS) dates suggest a time interval for core 9004 from at least 350 BC to ca.AD 1850-1900, and it comprises most of the well-known climatic periods: the Roman Warm Period (RWP), the Viking Age/Medieval Warm Period (MWP) and the Little Ice Age (LIA) (Filipsson and Nordberg, 2010).Radiocarbon AMS dates obtained for core 9004 between 410 and 750 cm indicate sedimentation rates of 2.8 mm yr −1 during the first 1250 yr, which then increases to 6.6 mm yr −1 up core around AD 1350-1500 (from 410 to 305 cm).From 305 cm and upwards, core sedimentation rate drops again to 3.1 mm yr −1 .The lower sedimentation rates in the deeper part of the core are regarded as an effect of sediment compaction.For the core G113-091 the only intact mollusc shell of Nuculana pernula yields an age of 1665±85 AD (Table 1; Fig. 2), which indicates that core represents the time interval from 2009 to the middle part of the Little Ice Age.The 210 Pb analyses for core GA113-2Aa reveal a sedimentation rate of 7 mm yr and represent the time interval between 1915 and 1999, which, therefore, encompasses the recent warming of the 1900s (Filipsson andNordberg 2004, 2010).Introduction

Conclusions References
Tables Figures

Back Close
Full

Geochemistry of bulk sediment
The organic carbon (C org ) data for the sediment cores GA113-2Aa and 9004 are presented in Filipsson and Nordberg (2010) and show a range from 0.6 to 3 % with 2 % as a long-term average.The C/N ratio measurements are available only for core 9004 and therefore C/N data for the last 100 yr are missing.In core 9004 C/N ratio ranges from 5 to 10, indicating a organic matter of primarily marine origin (C/N ratio of 5-7: Redfield et al., 1963).The long-term average of the ratio is 8.3.Around 170 cm depth (ca.1650 AD) a slight offset occurs in the C/N values, which divides the Little Ice Age into two periods: (1) before 1650 AD with C/N values higher than the long-term average and (2) after 1650 AD with C/N ratio below the average.Around 1900 AD, C/N ratio increases again and reaches 9.5 on top of the core.

Core GA113-2Aa
The sediment column in general was olive-green-grey and contained mainly organicrich silt and clay.The sediment surface was light brown, which indicated oxic conditions.An increased amount (9 %) of sand-sized fraction (> 63 µm) was found between 51 and 43 cm (∼ AD 1915).For more detailed lithology of a Gemini core GA113-2Aa see Filipsson and Nordberg (2004).

Core 9004
The sediments generally consisted of mud and clay of an olive-green-grey colour up to 369 cm depth.Around 369-367 cm, a prominent brown layer was found and contained plant remnants.A distinct light grey horizon of clay, silt and fine sand at 367-364 cm was interpreted as a turbidite related to a landslide (Polovodova et al., 2011).From 364 cm up core, grey colour gradually turned brownish to 350 cm.The brownish layer Introduction

Conclusions References
Tables Figures

Back Close
Full continued until 330 cm and included visible organic particles.After 330 cm up core, the brown colour faded away and sediments became olive-green-grey again.
The sand-sized fraction (> 63 µm) had a long-term average of 1.89 % for the whole record with a distinct maximum of 6.7 % at ca. AD 1200 (370 cm).Within depth interval 350-60 cm, indicative of the Little Ice Age (AD 1350(AD -1900)), the sand-sized fraction showed three different units (Figs. 4,5).A first coarser unit at ca. AD 1350-1600 (350-190 cm) can be identified based on a high variability of sandy fraction.A second unit between AD 1600 and 1850 (190-100 cm) was characterized by less variability and finer sediment.A third unit at ca. AD 1850-1900 (100-60 cm) was also characterised by coarser sediment.

Core G113-091
In general, the sediment column consisted of dark-olive-grey mud (gyttja clay) with mottled (containing black dots) sections in places.However, there were some distinct changes in colour within the interval 185-129 cm.At 185-181 cm the generally darkgrey mud contained horizontal layers of darker sediment, which gradually faded away at 174 cm.An interval of a light-grey mud was encountered at a depth of 174-171 cm and had a distinctly lighter 5-6 mm-thick horizon in the middle.After this layer sediments became dark-grey again and stayed so until 152 cm.Another light horizon was found at 152-129 cm, after which and towards the core top sediments column had a dark-grey colour.The surface of the core G113-091 was oxidised and had a light-brown colour down to 5 cm.The redox-cline was encountered at a depth of 9-5 cm.
The sand-sized fraction showed two distinct maxima (5 and 3 %) at depths of 36 and 64 cm and had a long-term average of 0.9 %.Introduction

Conclusions References
Tables Figures

Back Close
Full The full record of δ 13 C values from the cores GA113-2Aa and 9004 covering the last ca.two millennia was discussed in detail by Filipsson and Nordberg (2010).In the current paper we concentrate on the part relevant to the Little Ice Age.
The average δ 13 C value for the GA113-2Aa-9004 record is −0.65 ‰.The δ 13 C record of the LIA begins around 1350 AD with less negative values than the longterm average (Fig. 3) and the heavier values persist until ca.1500 AD.After that δ 13 C values reach their minimum of −1.15 ‰ and have a distinctly more negative phase, which lasts until almost 1850 AD.Afterwards the trend of the δ 13 C record changed again towards the less negative values, reached a maximum of −0.2 ‰ and remained less negative until the termination of the LIA.

Core G113-091
The stable carbon isotope record for G113-091 shows a fairly good correlation with the δ 13 C record derived from GA113-2Aa and 9004 down to a depth of ca.90 cm (Fig. 3).Below 90 cm the G113-091 curve has a slight displacement of 0.2-0.4‰ from the GA113-2Aa and 9004 record.Nevertheless, the good correlation of both isotopic records until 90 cm indicates that there is no gap between the cores GA113-2Aa and 9004.This is also confirmed by analysis of the dominant foraminiferal species in both records (Fig. 3).

350 BC-2001 AD
For foraminiferal data from the GA113-2Aa and 9004 record, we performed a factor analysis, which resulted in 4 factors, explaining as much as 89 % of variance ( Full Fig. 4).According to Klovan and Imbrie (1971), the minimum absolute factor value is a zero, which indicates that the variable contributes nothing to the factor.In our study eight foraminiferal species had high (> 1.0) absolute values of factor scores (Table 3, Fig. 4), which resulted in four foraminiferal assemblage units.Each foraminiferal unit was named after the most important species explaining each factor.Thus, Factor 1 (62 % of variance) corresponded to the Nonionella iridea-Cassidulina laevigata unit.
Factor 2, which corresponded to Stainforthia fusiformis unit, accounted for 14 % of variance and also included species Textularia earlandi, Bolivina pseudopunctata and Bulimina marginata.The Factors 3 and 4 accounted for 7 and 5 % of variance and corresponded to units of Adercotryma glomerata and Hyalinea balthica, respectively.
The fourth unit also included species C. laevigata and T. earlandi.The "N. iridea-C.laevigata" unit dominated the record during ca.350 BC-1150 AD and 1350-1650 AD (Fig. 4).The Adercotryma glomerata unit interrupted its dominance at ∼ 1300 AD.From ∼ 1650 to 1900 AD the record alternated between foraminiferal units 3 and 4 (A.glomerata versus H. balthica).After 1900 AD, however, Hyalinea balthica unit dominated for ca.80 yr and was replaced by unit 2 represented by S. fusiformis as the main species.

The Little Ice Age: 1350-1900 AD
In general, species composition of the benthic foraminiferal fauna during the Little Ice Age was characterized by three dominant (> 10 %) species: A. glomerata, C. laevigata and H. balthica, whereas Eggerelloides scaber, Elphidium excavatum, N. iridea, S.
Altogether 156 benthic foraminiferal species were found in the part of the record corresponding to the Little Ice Age.Among them 42 species had agglutinated tests and Introduction

Conclusions References
Tables Figures
The absolute abundances showed rather lower values over the time period from ca.
1450 to 1600 AD (Fig. 5).At the same time there was a clear increase of the foraminifer N. iridea, which reached its highest abundances (30 % and 56 ind.g −1 ) during the LIA at ∼ 1600 AD.
Both relative and absolute abundances showed a salient faunal change at 180 cm (ca.1650 AD) when N. iridea became rare in the record.Around the same time A.
glomerata, E. excavatum, E. scaber, S. fusiformis and T. earlandi increased.The foraminiferal absolute abundances showed high values from 1650 to 1850 AD with two minima at ca. 1630 and 1700 AD.
Hyalinea balthica became a prominent species in the record at ∼ 1650-1700 AD (Figs. 4, 5), but experienced several short declines until 1850 AD, after which its absolute abundance increased significantly.After 1850 AD there was also a significant increase in a portion of calcareous species, whereas some species such as A. glomerata, C. laevigata, E. excavatum and E. scaber, declined during that time.
In general, calcareous foraminiferal species dominated for the majority of the LIA.However during the second part of the LIA agglutinated species such as A. glomerata, E. scaber and T. earlandi became important components of the foraminiferal assemblages (Fig. 5).Introduction

Conclusions References
Tables Figures

Back Close
Full In our record, during ca.1300-1650 AD we observed increased scores of Factor 1 represented by N. iridea and C. laevigata (Fig. 4).This agrees with previous studies based on benthic foraminiferal stratigraphy from the Skagerrak (Hass, 1997;Erbs-Hansen et al., 2011), which reported high abundances of C. laevigata after 1450 AD and suggested an increased storminess, enhanced mixing of the water column and more fresh phytodetritus reaching the sea floor.Those studies however did not report any increase in N. iridea, which is probably due to use of the 100-and 125-µm fractions for foraminiferal analysis.A short and abrupt climate warming occurred at 1540-1610 AD as suggested by the benthic foraminiferal δ 18 O record from the Scottish fjords (Cage and Austin, 2010).Nonionella iridea was previously reported as a species indicative of fresh phytodetritus pulses during the Medieval Warm Period (Polovodova et al., 2011).It is likely that this species responded positively to both the warming event (1540-1610 AD) and the increase of phytodetritus, and therefore may be considered as a proxy of climate warming during the LIA.The short warming event around AD 1600 is also seen in the foraminiferal δ 18 O record based on core 9004 (to be published elsewhere).Nonionella iridea disappeared from the record after 1650 AD, which coincides with Maunder Minimum in solar activity between 1645-1715 AD (Mauquoy et al., 2002).The decreased bottom water temperatures due to an abrupt cooling in the second part of LIA could also contribute to dissolution of calcareous tests and favoured a proliferation of some agglutinated species (Adercotryma glomerata, Textularia earlandi and Eggerelloides scaber ), which we will discuss below.The lower temperatures may have been one reason for the disappearance of N. iridea.This faunal change coincides with a unit characterised by less sand-sized fraction (Fig. 4), which implies calmer sedimentation or decreased bottom water currents (Hass, 1996).Theoretically, Introduction

Conclusions References
Tables Figures

Back Close
Full a less-energy sedimentation phase would favour more algae to settle on the sea floor.
On the other hand, the period 1675-1710 AD has been known as the climax of the LIA (Lamb, 1983), which probably would be characterised by a shorter growing season of phytoplankton and therefore by less food available for fresh-phytodetritus feeders like N. iridea.Indeed, according to a tree-ring based climate reconstruction from the West-Central Sweden, the LIA was characterised by a cold climate phase with milder summers during 1350-1600 AD and colder summer conditions prevailed during 1600-1900 AD (Gunnarson et al., 2011).

Adercotryma glomerata, a proxy of climate deterioration
Br ückner and Mackensen ( 2006) use benthic foraminiferal δ 18 O to track a cold climate period in the Skagerrak and link this potentially to a decelerated thermohaline circulation until ca.1630 AD.In addition, they suggest an enhanced circulation and warmer temperatures from ca. 1630 to 1870 AD.The latter is contradictory to our results, which show a decline and disappearance of the temperate species N. iridea.Also, according to the factor analysis, three repeated increases of A. glomerata unit correspond well with three periods of reduced solar activity (Fig. 4): the Wolf (1300-1380 AD), the Maunder (1645-1715 AD) and the Dalton (1790-1820 AD) minima (Eddy, 1976;Mauquoy et al., 2002).Reduced solar activity has been long time considered as one of important players causing climate cooling (e.g.Haigh, 1996;Shindell et al., 1999;Bond et al., 2001;Mauquoy et al., 2002).Adercotryma glomerata is an agglutinated species, which is often associated with cold waters (< 4 • C) and a wide salinity range (28)(29)(30)(31)(32)(33)(34)(35) in the Labrador Sea, Canadian Arctic and Antarctic (Williamson et al., 1984).The Wolf sunspot minimum coincides with the onset of the Little Ice Age (Mauquoy et al., 2002), an abrupt ice-cap growth in the Arctic Canada between 1275 and 1300 AD (Miller et al., 2012) and an increase in drift ice off Iceland in the late 13th century, which caused the failure of Viking settlement on Greenland (Andrews et al., 2009).The climax of the Little Ice Age at ca. 1675/80-1710 AD, (Flohn, 1985;Lamb, 1983) is often associated with the Maunder minimum (Eddy, 1976).Therefore the appearance of A. glomerata in our Introduction

Conclusions References
Tables Figures

Back Close
Full record suggests that the intermediate Skagerrak water flowing into the fjord during the winter or early spring was colder than today, which supports the general idea of climate cooling after 1300 AD and then during ca.1675-1700 AD.It seems that in general the "non-stormy" period of the LIA is characterised by increased abundance of agglutinated species -A.glomerata, T. earlandi and E. scaber (Fig. 5).Wollenburg and Kuhnt (2000) reported A. glomerata prevailing in low-energy environments characterised by lower carbon fluxes in the Arctic Ocean.This is consistent with our results, showing proliferation of A. glomerata during ca.1600-1800 AD, which was characterised by a finer sediment interval, i.e. a lower amount of sandsized fraction (Figs. 4,5).Textularia earlandi, an omnivorous opportunist (Alve, 2010), is reported as a dominant faunal element in Canadian fjords at water depths ≥ 100 m (Schafer and Cole, 1988).Eggerelloides scaber is mentioned by Murray ( 2006) as a detritivorous shelf species, which demands salinity > 24 for most of the year and temperatures 1-20 • C. Therefore, proliferation of these agglutinated species might imply a change in quality of organic matter in Gullmar Fjord sediments during ca.1600-1800 AD; however, it is likely that another important process came into play during the cold LIA.This process is likely the dissolution of calcareous tests due to lowering of calcium carbonate saturation state at lower bottom water temperatures.A useful parameter indicating carbonate dissolution of foraminifera is the number of inner organic linings (Murray and Alve, 1999;Steinsund and Hald, 1994), which in our record shows higher values after 1500 AD.Another important measure is the calcareous/agglutinated ratio, which usually decreases with dissolution of calcium carbonate (Steinsund and Hald, 1994).In Gullmar Fjord this ratio displays clearly lower values during the time interval ∼ 1650-1850 AD.A number of corroded foraminiferal individuals indicated in water temperatures affects the benthic foraminiferal fauna and is a useful paleoenvironmental indicator.It is, however, plausible that degree of carbonate dissolution can be estimated as moderate, since together with agglutinated species we still find relatively high numbers of well preserved and shiny calcareous specimens and thin-shelled species like N. iridea and S. fusiformis.

Hyalinea balthica and amelioration of climate?
In general, based on the distribution of the main foraminiferal units, it appears that 1650 AD is a starting point for the establishment of new foraminiferal faunas in the fjord.This is a time when H. balthica, typical of the area during the 20th century, increases in Gullmar Fjord (Figs. 4,5).Hyalinea balthica together with B. marginata and A. glomerata are reported in the recent fauna at Fladen Ground, N North Sea, between Scotland and Norway (Klitgaard Kristensen and Sejrup, 1996).Hass (1997) suggested that assemblages of H. balthica in the Skagerrak are indicative of "stagnant conditions", by which he probably meant less intense bottom water currents on the southern flank of the basin.Indeed, in some of his cores (I KAL), there was an increase in H. balthica Full

Land use changes during the LIA
From the C/N record of core 9004 (Fig. 4) it is rather difficult to distinguish the sources of organic matter (marine vs. terrigenous) since ratio fluctuates within purely "marine" values.It has been suggested that C/N ratio reflects organic matter sources inaccurately, since nitrogen is affected by organic matter diagenesis and biological controls (Thornton and McManus, 1994).Therefore, using C/N ratio for degraded organic matter (which is usually the case for long sediment cores) the marine or terrigenous sources may be hard to distinguish.Filipsson and Nordberg (2010) studied δ 13 C on the tests from C. laevigata from the core 9004 and suggested that during the LIA changes in land use could contribute to the input of terrestrial carbon into the fjord.The terrestrial organic matter would change δ 13 C towards lighter values, which is the case indeed from ca. 1500 to 1850 AD.From terrestrial records it is known that during 1500-1700 AD there were increased losses of topsoil and finer particles in Sweden due to changes in land use and irrigation (Dearing et al., 1990) and periods of a widespread deforestation in Scandinavian countries (Bradshaw et al., 2005;Kaplan et al., 2009).Based on the pollen data from the Gullmar Fjord, Fries (1951) showed an increase in nonarboreal pollen (NAP), which started at the end of the Medieval Warm Period and lasted until ca.1650 AD (Fig. 4).This suggests an intensification of the land use including clearing of forests to increase croplands and pasture together with exploitation of timber for construction.Land use in Sweden and particularly Bohusl än most likely experienced a decline caused by outbreak of Black Death in ca.1350 AD, which wiped out between a half and 2/3 of the population (Harrison, 2000) and resulted in large scale abandonment of farms together with a series of wars, which followed thereafter (Fig. 4).The same was suggested for the Central Europe (B üntgen et al., 2011).In addition, there were several productive herring periods in the Bohusl än and Norwegian fisheries (Cushing, 1982;Corten, 1999) after ca.1550 AD.One of them, a so-called "The Great Herring Period" (Den Stora Sillperioden: 1748-1808 AD) resulted in the Introduction

Conclusions References
Tables Figures

Back Close
Full extensive production of fish oil, which required a lot of wood for boiling of herring guts and a substantial alteration of the landscape (Utterstr öm, 1959;Byron, 1994).

Storminess signals in the sediment lithology?
The Little Ice Age has been generally regarded as a stormy period (Lamb, 1983;De Jong et al., 2007, 2009).Based on granulometric data from Skagerrak, Hass (1996) divided the LIA into three parts: intervals 1350-1550 and 1750-1900 AD, characterised by a stormy mode of sedimentation and an interval 1550-1750 AD, characterised by a "calmer" sediment deposition.Using two raised bogs in near coastal part of SW Sweden, Bj örck and Clemmensen ( 2004) also reported an increased aeolian sediment influx (ASI) caused by winter storminess at 1810-1820, 1650 and 1450-1550 AD.The increased aeolian activity occurred at transitional phases between the Roman Warm Period and the Dark Ages (starting ca. 100 AD), at termination of the Medieval Warm Period (1050-1200 AD) and during the coldest phase of the LIA (1550-1650 AD) (Clemmensen et al., 2009).The climax of the LIA also coincided with sand movement and dune formation in Denmark (Clarke and Rendell, 2009;Clemmensen et al., 2009), which is in contrast with a hypothesis of a "calmer sedimentation period" at 1550-1750 AD (Hass, 1996).Due to a limited amount of the available sediment material, we could not perform a granulometric analysis to identify coarse silt (32-63 µm) and fine-sand (63-125 µm) fractions, used as proxies for storminess and strong bottom currents (Hass, 1996).Instead, we hypothesised that the sand-sized fraction (> 63 µm) may be used to some extent as a storminess signal.The occasionally increased amount of sand-sized fraction in the deepest basin may result from a turbulent flow causing erosion of the coarser material from the sill and slopes and its transport along the fjord during the water renewals.To investigate this further, we plotted the > 63-µm fraction against the frequency of geostrophic winds (Fig. 6), calculated from air pressure data (Alexandersson et al., 1998;Bj örk and Nordberg, 2003) and describing the regional wind direction during the storm events (Nilsson et al., 2004).As a result, the > 63-µm fraction and Introduction

Conclusions References
Tables Figures

Back Close
Full the number of storms show a relatively weak correlation with the exception of two sandy peaks at 1965 and 1915 AD as well as larger variability in the > 63-µm data set between 1870 and 1890 AD, which possibly correlates with periods of the higher storm and hurricane frequency (Fig. 6).The maximum sand-sized fraction reached at 1915 AD, given that a general uncertainty for 210 Pb CRS dating is 10-20 yr as suggested by Binford (1990), coincides with storms of 1921-1925 AD traced in Greenland ice by an elevated Na + content (Dawson et al., 2003).At the same time, the two strong hurricanes of January 2005 and 2007 (Iseborg, 1997;SMHI, 2009) are not visible in the sandy fraction records (Fig. 6).Generally the periods with increased variabililty of sandy fraction (1350-1600 and 1850-1900 AD), which may result from "stormy sedimentation" in Gullmar Fjord, coincide with the findings of Bj örck and Clemmensen ( 2004), Clarke andRendell (2009) andClemmensen et al. (2009), whereas the division into two stormy and one calm LIA phases generally agrees with the idea of Hass (1996).
An alternative or/and complementary explanation of increased sandy fraction in the record at the LIA onset and at ca. 1850 AD may be a sediment-laden sea ice.Omstedt and Chen (2001) and Omstedt et al. (2004) reported an increase of the maximum ice extent in the Skagerrak during the LIA.During the winter growing sea ice collects sediment from the near-shore areas of the fjord (Cossellu and Nordberg, 2010).These sediments are eventually transported to the deeper fjord areas by the offshore winds or high tides and may get deposited there during the spring thawing.

Conclusions
The Little Ice Age period is clearly seen in the benthic foraminiferal records from the deep basin of Gullmar Fjord.The onset of the LIA is indicated by an increase in coldwater species Adercotryma glomerata.The first phase of the LIA has been characterised by a stormy but milder climate, which is indicated by the presence of Nonionella iridea.Maximum abundances of this species are likely to mirror a short and abrupt warming event at ∼ 1600 AD.It is likely that due to deforestation, extensive fishing and Introduction

Conclusions References
Tables Figures

Back Close
Full land use changes in the second part of the LIA, there was an increased input of terrestrial organic matter to the Gullmar Fjord, which is indicated by lighter δ 13 C values and an increase of detritivorous and omnivorous foraminiferal species, such as Textularia earlandi and Eggerelloides scaber.At the same time, the general climate deterioration during the climax of the LIA (1675-1704 AD), as suggested by the increase of agglutinated species, may have caused some carbonate dissolution as well as variations in marine primary productivity, which therefore led to a decline of N. iridea dependant on fresh phytodetritus.We also suggest that an increase of Hyalinea balthica could be indicative of general climate warming at 1600-1743 and 1813-1940 AD.Introduction

Conclusions References
Tables Figures

Back Close
Full  Full  Full  Full     (Mauquoy et al., 2002).Gray and black rectangles indicate productive herring periods from the Bohusl än fishery (gray and gray with dotted frames) and from the Norwegian fishery (black) (Ljungman, 1883;Cushing, 1982).The thick dashed line with circles shows the total non-arboreal pollen (NAP; %) in the Gullmar Fjord, which is indicative of the land use intensification (Fries, 1951) J J J J J J J J J J J J J J J J J JJ J J J J JJ J J J J J J J J J J J J J J J J J J J J J J J J JJ JJ J J J JJ J J J J J J J J J J J J J J J J J J J J J J J J JJ J J J JJ J J J JJ J J J J JJ JJ J J J J J JJ J J J J J J J J JJ J J J J J J J J J J J JJ JJ J J J J J J J J J J   Forams /g J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J 0 15 30 45 0 15 30 45 0 15 30 45 (% ) 0 20 T .e a rl a n d i (% (% ) Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Fig. 2 .
Fig. 2. Age model based on 210Pb (core GA113-2Aa) and 14 C AMS datings (core 9004).For AMS dating a reservoir correction of 500 yr has been applied.Dashed lines at 357-371 cm indicate the primary position of a layer referred as landslide and turbidite and removed from the age model(Polovodova et al., 2011).

Fig. 5 .
Fig. 5. Absolute and relative abundances of main foraminiferal species, together with some sediment proxies shown as enlargement for the Little Ice Age (LIA) period.Dashed lines indicate the main faunal changes at ca. 1650 and 1850 AD discussed in the text.

Fig. 6 .
Fig. 6.Sand-sized fraction (> 63 µm) in cores GA113-2Aa and 9004 and G113-091 versus a number of geostrophic winds for triangle Gothenburg-Lund-Oksøy, calculated from air pressure data (Alexandersson et al., 1998; Bj örk and Nordberg, 2003).The thin grey columns indicate number of days with winds ≥ 25 m s −1 regarded as storms, whereas thick black columns show days with hurricanes (≥ 34 m s −1 ).The black line indicates a 10-yr running mean for winds of speed ≥ 25 m s −1 .The asterisks show some of the historically documented hurricanes of ca.40 m s −1 , which struck the west coast of Sweden: The Autumn Storm of 1969, and The January storms of 2005 and 2007 (Iseborg, 1997; SMHI, 2009).

Table 1 .
Radiocarbon dating results based on 12 double mollusc shells recovered intact from the cores 9004 and G113-091.

Table 2 .
The factors, which resulted from a CABFAC factor analysis with varimax rotation performed on raw foraminiferal data for species contributing > 5% to the assemblages.

Table 3 .
The varimax scores for factors 1-4.The bold numbers indicate foraminiferal species with high (> 1) absolute values of factor scores.