In Those Parts of Norway Covered by Marine Silt and Clay of Late Weichse

STRATIGRAPHY AND DATING OF HOLOCENE GULLY SE DI MENTS IN OS, WE STERN NORWAY

EIVIND SØNSTEGAARD & JAN MANGERUD

SØnstegaard, E. & Mangerud, J.: Stratigraphy and dating of Holocene gully sediments in Os, western Norway. Norsk Geologisk Tidsskrift, Vol. 57, pp. 313-346. Oslo 1977. At Os, near Bergen, western Norway, there occur several gullies in glacio marine silt. The gullies, up to 450 m long and 7 m deep, lead into the lake Banktjørn, where the eroded sediments were deposited. The analyses of several cores from the lake sediments allowed the gully sediments to be iden tified and dated as having been formed in the period from 8,800 to 7,700 years B. P. Denudation rates for this period have been estimated at 1.45 mm/year in the silt covered area. The dating was based on pollen diagrams and 14C dates. The Holocene history of the vegetation, redeposition of pol len, and lateral variation of the pollen content in lacustrine sediments are also discussed.

E. Sønstegaard & l. Mangerud, Geologisk institutt, avd. B, Allegt. 41 , N-5014 Bergen Universitet, Norway.

In those parts of Norway covered by marine silt and clay of Late Weichse lian-Holocene age, gullies and landslide scars are the most common land forms, besides the river valleys. The main areas of marine silt and clay are the lowlands in eastem Norway (around Oslo, Fig. 1) and Trøndelag (around Trondheim, Fig. 1). In western Norway the fiords and valleys are so steep that only small areas were inundated by the sea, despite the marine limit over large areas being 50-100 m a.s.l. The land areas covered by silt and clay are, therefore, less than in eastem Norway and Trøndelag, and nearly always connected with ice-marginal deposits, as in the present area. The clay slides in Norway are often connected with catastrophic rui nation of farms and buildings, with the result that there are historical data from far back in time. Many geological and geotechnical studies have been performed to clarify the age, cause, and dynamics of the elay slides ( e. g. Holmsen 1953, Rosenqvist 1953, 1955, Bjerrum & Rosenqvist 1956, 1957, Jørstad 1968, Bjerrum 1971). The formation of gullies, on the other hand, is a continuous process that does not cause conspicuous destruction over short periods of time. There fore it has received less attention than landslides, but is perhaps as important an erosional agent when considered in the longer term. In Scandinavia the most extensive studies on gullies have been carried out along the river DaUi.lven in Sweden (von Post 1935, Høgbom 1905, De Geer 1914, Caldenius 1926, Frendin 1931, Wenner 1941, Lundqvist · & Hjelmqvist 1941, Lundqvist 1951). 314 E. SØNSTEGAARD & J. MANGERUD

Lundqvist & Hjelmqvist (1941) dated the formation of one of these, the Norrhyttan gully, by a method which in principle is the same as the one we have used. Sand layers descending from the Norrhyttan gully were identified in cores taken from a peat bog at the gully foot. These layers were dated with the aid of pollen analysis, showing that the gully had formed during three periods (ca. 3,700, 1,300, and 1,000 years B. C.) after the area had been raised above sea leve!. Lundqvist & Hjelmqvist assumed that the for mation of the Norrhyttan gully was connected with heavy precipitation, and that the climate had been moist during the three mentioned periods. Similar studies of fluvial sediments believed to have originated from gul lies, in Sæters dal, indicated that these, for the most part, had formed ca. 400 years A. D. (Lundqvist & Hjelmqvist 1941). We have investigated a system of gullies in Os, Hordaland (Fig. l and 2), putting main emphasis on the sediments originating from the gullies, and on age of gully formation. These gullies end in the lake Banktjøm (Fig. 2)� which receives virtually no minerogenic sediments from sources other than the gullies and a little brook. This situation offers a unique possibility to identify and date the gully sediments. The content of organic matter in these sediments was too low for radio carbon dating. We have therefore constructed pollen diagrams, and dated the sediments through correlation with radiocarbon-dated pollen boundaries from nearby localities. The pollen analysis also gave some results conceming distribution of pollen within the basin, and conceming the development of · the vegetation.

Fig. l. Map of the Bergen-Os area. Inserted a key map of southern Norway. HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 315

Methods Co ring A pistan corer with diameter 54 mm (GeonorAlS, Grini Mølle, Box 99 Røa, Oslo 7, Norway) was used for all sampling. We did, however, increase the length of the sampling tube from the standard 88 cm to 208 cm. The !ami nation of the sediments was virtually undisturbed (Figs. 7 and 8), indicating that the construction of the sampler is very good. The compression of the studied silty sediments very sel dom exceeded 5 %. About 70 m co res were sampled with this equipmentin and around the lake Banktjøm. Core l (Fig. 2) was taken through the lake ice during winter. The coring frame was placed on two 4 m long wooden beams 4" X 4", and fasten ed to the ice by means of chains. The seven coring localities are shown in Figs. 2 and 3. Cores 1-4 cover the entire investigated sequence and a pollen diagram was constructed for each (Figs. 15-18). From cores 5-7, samples were taken only at selected depths to locate important lithostratigraphical boundaries. For calculation of sediment volumes, some supplementary motor soundings were performed.

Grain-size analysis Grain-size analysis was performed only on sediments with loss on ignition below 10 %. Organic matter was removed with 10 % H202 solution. The samples were first washed through a 0.063 mm screen. The coarser material was sieved, while the fines were analysed by the pipette method (Krumbein & Pettijohn 1938). The fines were not dried before the pipette analysis, and the total weight was found by means of '0-sample' (Aarseth 1971).

Loss on ignition The loss on ignition (Figs. 15-18) was determined by ignition at 750°C for one hour. This has been the standard procedure in Geologisk institutt, Uni versitetet i Bergen (Mangerud 1970), although most other laboratories use 550°C (Birks 1970, Saamisto 1970, Berglund & Malmer 1971). To test the differing significance of these two ignition temperatures, 153 samples were ignited at 550°C and thereafter at 750° C. The mean difference in loss on ignition between these two temperatures for all samples was 0.90 % with a standard deviation of 0.44 %. To show how much of this difference is due to further combustion of organic matter, the samples were grouped in three categories, those with loss on ignition values (750°C) between 0-5 %, 5-50 %, and 50-100 % (Table 1). The results given in Table l show that the difference in loss on ignition is !east for the group containing most organic material. This indicates that the latter is almost completely combusted at 550°C. Consequently the in crease in loss on ignition at 750°C is essentially due to loss of water from clay minerals (Ekstrøm 1927) and possibly some released C02 from car- 316 E. SØNSTEGAARD & J. MANGERUD

Table l. Mean differences with one standard deviation between loss on ignition at 950°C, 750°C, and 550°C.

Mean difference of Mean difference of Loss on loss on ignition loss on ignition Number of Number of ignition at 750°C and at 950°C and samples samples 750° 550°C, with one 750°C, with one analysed analysed % standard deviation standard deviation % %

0-5 0.77 ± 0.36 73 0.67 ± 0.17 81 5-50 1.12 ± 0.40 50 0.90 ± 0.33 51 50-100 0.64 ± 0.31 30 0.78 ± 0.40 7 0-100 0.90 ± 0.44 153 0.76 ± 0.28 139

bonates and S from ferrosulfides (Andersen 1961, Digerfeldt 1975). It is concluded that an ignition temperature of 550°C is preferable. 139 samples were also ignited at 950°C. The mean difference between loss on ignition at 950°C and 750°C, 0.76 %, is probably due to chemical reactions other than combustion of organic matter. Standard deviation in the last case was 0.28 %.

Pr.eparation of pollen All samples contained minerogenic particles which were removed according to the HF method (Fægri & Iversen 1966), the heavy liquid (bromoform) method (Frey 1955, Vorren 1972), or a combination of these two. The re maining material was treated with the acetolysis method (Fægri & Iversen 1966). Most of the samples treated with the HF method were stored several days in cold 35 % HF, and thereafter treated with 10 % HCl, etc. In the cl�y rich samples it was, however, often difficult to remove all the particles (fluorides?) which were precipitated during the treatment. During the last few years, the bromoform method has been used with success in Geologisk institutt, Universitetet i Bergen (Vorren 1972, Sønste gaard 1974, Skår 1975, Mangerud 1977). Vorren (1972) compared the re sults of the HF method and the bromoform method, and found no signif icant differences in pollen frequency. Hystrix was, however, under-repre sented in samples treated with the bromoform method. Skår (1975) found that Hystrix often lost the characteristic spines during the bromoform sep (j,tation. Other systematic differences may also result from the preparation tech niques, and the techniques used are therefore indicated in the last column in the pollen diagrams.

Pollen identifications The pollen analyses were performed in Botanisk Museum, Universitetet i Bergen, and the identifications refer to the reference collection in that de- HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 317

LEG END

"""\:? Gu11 y D Glaeiomarine silt

[,:)::::.:J parts of =,t;ilake Bonktjørn,:t�'fu.� �tiJ�!f --- Watershed of the 9anktiern cotchment o•eo

Coring locallty 3. • • ·- ... Brook in funnel

100 Contour interval 5m Fig. 2. Map of the investigated area. Localisation of the map is marked at Fig. l.

partment. Routine counting was done with use of phase contrast and objec tives 25/0.45 or 40/0.65. For difficult determinations, immersion oil ob jective 100/1.30 was used. The group 'unidentified pollen' consists mainly of grains unidentifiable 318 E. SØNSTEGAARD & J. MANGERUD due to corrosion, etc., with addition of a few unknowns. High percentages of unidentified pollen can result in considerable errors in the diagram, be cause the preservation of pollen varies between the species (Andersen 1966: 269, Cushing 1967, Birks 1970). In the diagrams from Banktjørn the cate gory 'unidentified pollen' constitutes 1-30 % of the total pollen, but in most of the spectra it is below 15 %. The group cf. Populus may contain some pollen of Taxus. In several samples it was very difficult to distinguish between the two species. Taxus occurs quite seldom in fossiliferous material in western Norway and then only in small amounts (Hafsten 1965: 46-47). On the basis of the rela tively high percentages, e. g. in Fig. 17, it is most likely that the pollen grains in question are Populus.

Description of the area Os is located on the outermost part of Fusafjorden, 20 km south of Bergen (Fig. 1). The investigations were carried out around the lake Banktjørn, 1-2 km inside the outermost deposits of the Herdla Moraines (Aarseth & Mangerud 1974). The marine limit, of Younger Dryas age, is ca. 60 m above the present sea level, while Banktjørn Iies 21 m a. s. l. The river Oselva flows in the same main valley as Banktjørn is found, but the river does not flow into the lake (Fig. 2). The lowest pass point between Oselva and Banktjørn is 31 m a. s. l., i. e. only 4 m above the river at the actual site. Nevertheless, no evidence has been found to suggest that the river has flowed over the pass and into Banktjørn after the pass emerged from the sea. Banktjørn occupies an area of 0.12 km2, and its catchment area is about l km2 (Fig. 2). Above the marine limit there is mainly bare bedrock with a thin discontinuous cover of soil, mainly weathering products and raw humus. In lower lying parts, there occur essentially glaciomarine silt and postglacial marine and lacustrine sediments (Fig. 2). The vegetation consists of pine forest on the bare hills surrounding the lake. On the silt and bog along the lake shore are copses of grey alder (Al nus incana), black alder (Alnus glutinosa), and birch (Betula pubescens). Most of the silt area is, however, cultivated.

Gully erosion processes In the glaciomarine silt there are several gullies. Some of them are connected and form systems of gullies, all of which empty into Banktjørn or into parts of the lake that are now filled in with lacustrine sediments (Fig. 2). The largest system of gullies, named the Flåten gully, ends in the southern part of the basin. It is 450 m long and up to 7 m deep (Figs. 2-5). Long profiles of the gullies are graded and the cross profiles are U-shaped (Fig. 5), indi cating a mature stage of development. Detailed investigations of the processes which have formed the gullies HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 319

Fig. 3. Air photo of the Flåten area seen towards SW (compare Fig. 2). Coring sites 1 to 7 are indicated. The Flåten gully is marked by a stippled black and white line. In the background lake Ulvenvatn to the right and the sea (Fusafjorden) to the left. May 1975. have not been undertaken, but some conclusions can be drawn from the forms and the redeposited sediments. In the gullies no springs have been observed, and ground water erosion is believed, therefore, to have been negligible in the formation of these gullies.

Fig. 4. Stereophoto of the Flåten gully. The gully ends at coring site 3. Photo Widerøe/Fjellanger 1966. 320 E. SØNSTEGAARD & J. MANGERUD

E'-tion Polilion øl tN mcu.l. . ,.."...,.Clllrben� ,..-,BP .

Fig. 5. Long and cross profiles of the Flåten gully. To the right is indicated the position of the shoreline at different times. The vertical scale is 5 times exaggerated.

This is also consistent with the glaciomarine silt being so fine grained (Fig. 6) and impermeable that the predominant amount of rain water will run off on the surface. Small landslides can, on the other hand, have contributed to the gully formation. The sensitivity of the glaciomarine silt is high, with an average value of 60 for five measurements. Two small, relatively fresh, landslide scars are observed on the east side of BanktjØm, and probably small slides took place during the entire postglacial period. There are, however, no per ceptible scars from landslides within the gullies. The silt derived from the gullies (Iacustrine silt, Fig. 6) is hetter sorted than the glaciomarine silt; it has an even lamination and a fairly constant loss on ignition (Figs. 16, 17), and structures indicating larger landslide activity were not observed. All these observations imply a continuous sedimentation from suspension during the entire period. Nordseth (1974 and pers.comm. 1975) emphasizes that in eastem Norway the main gully erosion takes place as lateral channel erosion while the main part of the sediment is delivered to the river by slow mass movement. On the basis of the given observations we conclude that this is also the case for the Flåten gully.

Lithostratigraphy The sediments in Banktjøm can be sub-divided into six (informal) lithostrati graphic units (Fig. 11): till, glaciomarine silt, marine silt, lacustrine silt, lacustrine gyttja, and peat.

Til! The lowest bed consists of a compact gravelly silt, of which we obtained only small samples with the boring equipment used. The thickness of this HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 321

CLAY SILT SAND GRAVEL 99

98 l l �i 95 :li i 1/ i V 90 l/ ilj / /i y� .' 80 lj l / !--•/ 7/ ·; 1/ .-·"'" 70 - I;;? i ."j.l .,...... , 60 1/ l j. l i 50 li '..1 40 /l l /j l 30 v

10 9 8 6 5 " 3 2 l o - 7 1 -2 -3 phi 1At2 Y32 1/16 Va V.. Yl l 2 " 8 mm 2 " 8 16 31 63 1000 """'"' micron Unbroken l ine: Lacustrine samples. Stippled line: Glaciomarine samples. Fig. 6. Grain-size distribution of sediments from Banktjørn. Curves from the most fine grained and coarse-grained samples analysed from each of the two beds are shown. bed is therefore unknown. The sediment contained shell fragments and we assume that it is a shell-bearing till. Similar shell-bearing tills have been observed in several excavations in the surrounding area (Undås 1963, H. Holtedahl 1964, Mangerud 1970, Aarseth & Mangerud 1974, Sønstegaard 1974).

Glaciomarine silt Above the till is a bed of laminated sediments (Fig. 7). The predominant grain size is silt, the microfossils are of marine origin, and we assume that it was deposited just beyond the front of a glacier. Therefore we have named it glaciomarine silt. The clay content in the glaciomarine silt is up to 26 % (Fig. 6). Coarse silt and sand occur in the lower part, thus causing a general fining-up within the unit. The lamination is due to alternating grain sizes (mainly clayjsilt) and same graded laminae, frequently ranging from sand to fine silt (Fig. 7). 322 E. SØNSTEGAARD & J. MANGERUD

Dept h (m)

- 10. 10

10.20

Fig. 7. Glaciomarine silt from core 4. Some of the laminae are graded.

The lamination is distinct in the lower part of the unit but becomes gradually less obvious upwards. Small shell fragments of sand size occur in the layers of sand. The amount of other organic material in the silt is insignificant. Black zones can be found spread throughout the whole unit. These are most likely the result of finely distributed iron (Il) sulphide. The glaciomarine silt has a thickness of 2.5 to 8 m in Banktjøm, and the bed continues above the lake shore almost up to the marine limit (Fig. 2). This bed is the parent material into which the gullies are eroded. The pollen spectra from this silt (Figs. 15 and 17) show that it was deposited befare the expansion of the birch, and that the vegetation in the vicinity was a nearly tree-less tundra vegetation. This indicates that it was deposited during, or very soon after, the deglaciation of the area. The !ami nation and the obviously very high sedimentation rates indicate a deposition immediately in front of the glacier.

Marine silt The lamination in the glaciomarine silt disappears gradually upwards with an even transition to a non-laminated silt with a thickness of 2-3 m. The micro-fossils and shells indicate that this silt was deposited in a non-glacial marine environment, and it is therefore named marine silt. The sediment HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 323 is slightly coarser than the main part of the glaciomarine silt and the sorting is somewhat poorer in the coarsest fractions. The continuous transition from laminated to non-laminated silt is probably due to the gradual change in the environment which took place as the glacier retreated and the supply of the glacial detritus diminished.

Lacustrine silt The most distinct lithological boundary is between the marine silt and the next bed, called lacustrine silt. The microfossils show that this is the bound ary between marine and lacustrine sediments. All sediments below this boundary have greyish or greenish colours, while all sediments above have brownish colours due to content of organic matter. In the lacustrine silt the content of organic matter is, however, generally low, the loss on ignition being below 10 %. The bed termed lacustrine silt consists essentially of laminated clay, silt, and gyttja silt, including a few layers of sand and gyttja. The lamination is most distinct just above the lower boundary (Fig. 8). No laminae are clearly graded. There seems to be a coarsening-up within the unit (Fig. 17). Similar black zones to those in the glaciomarine silt occur in this bed.

Lacustrine gyttja The boundary between the lacustrine silt and the next bed, named lacustrine. gyttja, is not distinct. Within the lacustrine silt the content of organic matter, and thereby the brown colour, generally increases upwards. We have placed the boundary at a marked increase of the brown colour, which appears where the loss on ignition exceeds 10 %. For the later discussions we will stress that also in lacustrine gyttja, silt is a major constituent. In the central part of the lake (Fig. 15) the boundary is least clear and the loss on ignition i& only 10-15 % in the lacustrine gyttja. Outside the mouth of the Flåten gully the loss on ignition is usually 20-40 % (Figs. 16 and 18). Also in the lacustrine gyttja lamination occurs, but is less distinct than in the lacustrine silt below.

P eat A thin layer of peat covers most of the lacustrine silt and gyttja south of Banktjøm (Fig. 11), but is missing along the shore of the lake. The peat indicates that no sedimentation of silt occurs in the southem part of the area today.

The gully sediments The silt which has been eroded from the gullies must have been deposited in BanktjØm, and we refer to these sediments as the gully sediments. It is suggested that the identification and dating of these sediments also dates the gully formation. 324 E. SØNSTEGAARD & J. MANGERUD

Fig. 8. The sharp boundary between the non-laminated marine silt and the laminated lacustrine silt in core 3.

The gullies were eroded into the glaciomarine silt; therefore the gully sediments must lie stratigraphically above this unit, that is, in the marine silt andfor in the lacustrine silt and gyttja (Fig. 11). Four arguments indicate that the gully sediments occur almost exclusively in the lacustrine sediments:

The elevation of the gullies. - At the end of Younger Dryas Chronozone (10,000 years B. P.) the shoreline was 60 m higher than at present in Os (Aarseth & Mangerud 1974). The isolation of Banktjøm (21 m a. s. l.) from the sea is dated to 8,800 years before present by means of pollen analysis. In Fig. 5 the elevation of the shoreline at different times is indicated, as suming an even emergence between the two given points. The gullies lie between 21 and 45 m a. s. l. (Fig. 5), and therefore the formation cannot have begun until ca. 9,500 years B. P. However, the main part of the gully volume lies within the lowermost 10-12 m, and was therefore only exposed for some 300 years before Banktjøm was isolated from the sea. It is there fore reasonable to conclude that the erosion of the gullies started during the HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 325 marine sedimentation phase, but could scarcely have been completed during this period.

Grain-size distribution. - One of the cores (no. 3) was taken from imme diately outside the mouth of the Flåten gully. The lacustrine silt is coarser here than in the other cores (Fig. 9), indicating that the site was dose to the sediment source during the time of deposition. The marine silt, on the other hand, is finer outside the Flåten gully than further north and south, indicating that the gully was not the source of this sediment.

Sediment distribution. - The marine silt bed and the lacustrine silt bed have very different thickness distribution patterns. The thickness of the lacustrine silt is especially great at the mouths of gullies (Fig. 11), sug gesting that it was deposited from the gullies. The marine silt, on the other hand, has an even thickness over the entire basin (Fig. 11). At the gully outlets in the NW the thickness of the marine silt is even less than in the central parts of the basin.

Estimation of the volumes. - In the second and third paragraphs above, it

micron LACUST R INE SILT

'U "' � 30 - 10 o c ru z

10 (3) (4) (4) (2) o 1 2 3 4 5 Core number Fig. 9. Histograms showing the mean of the Md for samples from each core of lacustrine silt and postglacial marine silt. The figures in brackets give the number of samples analysed in each case. 326 E. SØNSTEGAARD & J. MANGERUD

o TOTA L AREA FLAT EN GULLY

5 CY) E 4 lt1 9 )( 3

Q) E ::J o >

� Volume of gullies ( eroded silt ) (:;-\:i:{{j Volume of minerogenic lacustrine sediments Fig. 10. Comparisons of the estimated volumes of gullies and of minerogenic lacustrine sediments.

is indicated that all of the gully sediments lie in the lacustrine sediments. During the lacustrine phase Banktjørn received minerogenic sediments al most exclusively from the gullies and a small creek in the northern part. Out of BanktjØrn there is practically no sediment transport. The volume of eroded silt in the creek valley and the gullies must therefore be equal to, or slightly larger than, the volume of silt in the lacustrine sediments in Banktjørn. The presumed silt surface before the erosion started has been recon structed through field work and the study of aerial photographs. The vol urne of eroded silt has then been calculated as the difference between the present surface and the primary surface by means of maps with a scale of 1:1,000 with equidistance being l m. This volume is calculated to be 310,000 m3 (Fig. 10). The volume of silt in the lacustrine sediments, 480,000 m3, was calculated on the basis of seven bore hoies (Fig. 2) and has been cor rected for loss on ignition and density in relation to the glaciomarine silt in which the gullies are formed. The calculations are, of course, only rough estimates, but indicate (Fig. 10) that the lacustrine silt and gyttja are able to contain all the gully sedi ments. Corresponding calculations are carried out separately for the Flåten gully. These calculations are more precise since the gully is more well defined and more coring was performed in the sediments assumed to be derived from this gully, i. e. the sediments south of the lake outlet (Fig. 2). The estimated volumes of silt eroded (130,000 m3) and accumulated (140,000 m3) from the Flåten gully are almost identical (Fig. 10). HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 327

The conclusion is that all the gully sediments Iie within the lacustrine silt and gyttja in BanktjØm. For the Flåten gully one can conclude even more precisely that the rninerogenic part of the lacustrine silt and gyttja is iden tical to the gully sediments.

Dating of the erosion of the Flåten gully The content of organic matter in the sediments of BanktjØrn is too low to determine their ages with the 14C method. Dating was therefore carried out by means of pollen analysis. Well-defined pollen boundaries which are dated by the 14C method of Iacustrine gyttja from nearby areas are used. The dating and the pollen diagrams are discussed in further detail in the chapters on biostratigraphy. Four pollen diagrams have been constructed, three of which encompass the Flåten gully sediments (Figs. 16-18). Since all of the diagrams are from the same lake, the pollen boundaries are synchronous (Fig. 11) . The main results are given in Figs. 11 and 12. The deposition of the Flåten gully sediments began 8,800 years B. P. At the mouth of the gully (core 3, Fig. 11) the sediments were deposited very quickly (4.3-6.1 mmjyear) and the sediment surface reached the water leve! already by 7,700 years B. P. The content of organic matter (loss on ignition, Fig. 17) in the sediments at the mouth of the gully is very low. Thereafter the sedimentation took

Fig. 11. Stratigraphic correlations between cores 1-5 in BanktiØm. The lithostratigraph ical units are indicated with different signatures. Time synchronous pollen levels are indicated by lines with age. C0 = The Corylus rise, Cmax. = The Corylus maximum,

A0 = The Alnus rise, Q0 = The Quercus rise. 328 E. SØNSTEGAARD & J. MANGERUD

Fig. 12. Sedimentation rates of the minerogenic component of the lacustrine sediments in Banktiøm during different periods. The values are thus corrected for contents of water and organic matter. The lines connecting the boring sites are the synchronous levels given in Fig. 11. place towards the north ( core 2, Fig. 11) - where the content of organic matter soon after increased (Fig. 16) and the sediment is termed lacustrine gyttja - and towards the south (core 4), where the content of organic mat ter (Fig. 18) was high right from the start. Tims the deposition of the gully sediments can be described as a building out of a delta outside the gully mouth. The stratigraphical position of the Corylus maximum (Fig. 13) - and therefore the 8,400 isochrone line (Fig. 11) - is uncertain. Accordingly the sedimentation rates (Fig. 12) based on this age can be slightly changed. Nevertheless, the sedimentation rate in core 3 must have been lower during the first part of the lacustrine phase than during the following period (Fig. 12). The explanation is probably that the erosion increased gradually as the side branches of the gully were developed. In cores 2 and 4 deposition was naturally much slower during the first period than in core 3 (Fig. 12). It is, however, very steiking that the rate of deposition of minerogenic particles at these two sites did not increase after the sediments were built up to the surface at core 3 either, even though it is at these sites that the material from the Flåten gully was now deposited (Fig. 11). Therefore it is quite clear that the erosion in the gully diminished strongly after ca. 7,700 B. P.

Denudation rates Accepting the results above, the sediment transport and denudation rates for the drainage area to the Flåten gully can be calculated (Table 2). Prac tically all erosion has taken place in the silt area, and the erosion rates within this area are therefore calculated separately (Table 2). Measurements of sediment transport in several Norwegian rivers have HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 329

Table 2. Erosion rates calculated for the Flåten guily, compared with present-day erosion rates in Mønsterelva, eastern Norway (Nordseth 1974). The asterisks denote values calculated from information in Nordseth (1974).

Specific Drainage Denud- sediment Fluvial system area Period transport ation rate (km2) (mm/year) (t/km2 . year)

Flåten gully, 0.19 975 0.65 8,800-7,700 B. P. total area Flåten gully, 0.085 2175 1.45 8,800-7,700 B. P. silt and clay area Mønsterelva E + W, 27.2 420* 0.28* 1970 total area Mønsterelva E + W, 24.5 473* 0.31* 1970 silt and clay area

been carried out during the last few years (Ziegler 1974, Nordseth 1974). Of these, the two brooks Mønsterelva East (E) and Mønsterelva West (W) (Nordseth 1974) in a clay area east of Oslo (Fig. 1) are most comparable to the Flåten gully. In Nordseth (1974, Table 48) only specific suspension trans port is listed. To obtain the total sediment transport in Mønsterelva, we have added 10 % to the suspension transport values due to bedload (cf. Nord seth 1974: 156). Between 8,800-7,700 B. P. the specific sediment transport in the Flåten gully, 975 tfkm2 year, was about twice the value measured in Mønsterelva E and W (total areas, Table 2). However, according to Nordseth (pers. comm. 1975) the mean erosion rate in Mønsterelva during the postglacial time is also estimated to be twice the value measured in 1970, and therefore of the same order of magnitude as we have calculated for the Flåten gully. A more precise comparison of the given erosion rates has little value, be cause of the different sizes (Table 2) and topographic features of the catch ment areas, different periods of time to which the denudation rates refer (Table 2), and the different climate between eastem and western Norway.

Biostratigraphy Studies of pollen and other microfossils have been made with a view to solving stratigraphic problems. The results provided information too about the lateral distribution of fossil pollen in lacustrine sediments, and the development of the vegetation during the Holocene. The material is not ideal for solving these problems, the main weaknesses being the high minero genic content of the sediments, and the large distances between the pollen spectra in the diagrams. However, some results have been obtained. Radio carbon dates of the Corylus rise and the Alnus rise were available from two other pollen diagrams from the Os area (Fig. 14), and these diagrams are therefore included in the discussion. 330 E. SØNSTEGAARD & J. MANGERUD

The Holocene history of the vegetation on the coast of Hordaland is rel atively well known (Fægri 1944 a and b, 1950, 1954, Hafsten 1965, Hagebø 1967, Mamakowa 1968, Mangerud 1970, Bakka & Kaland 1971, Kaland 1974, Skår 1975). The pollen diagrams from Os (Figs. 14-18) display the same main features as described by the mentioned authors.

Pollen zones and chronozones The classification and terminology of pollen zones follows the proposals given in the International Subcommission on Stratigraphic Classification (1971) and Mangerud et al. (1974). We will stress that the names are strictly local and informal. The chronozones are according to the definitions in Mangerud et al. (1974: 119). The boundaries are identified on the basis of the following radiocarbon dated pollen levels:

The Corylus rise, 8,800 B. P., based on a date giving 8,960 -+- 220 (T - 1491) just below the Corylus rise in Haukelandsvatnet, Arna (Fig. 1) (Skår 1975). The Corylus maximum, 8,400 B. P., based on the date from lake Lepsøy vatn (Figs. 1 and 14). The Alnus rise, 7,700 B. P., based on the date from lake Lekvenvatn (Figs. 1 and 14). The Querqus rise, 5,700 B. P., based on several dates from Austrheim, 70 km north of Banktjøm (Kaland 1974 and pers. comm.).

Comments on the localities LepsØyvatn and Lekvenvatn Both diagrams were worked out several years ago in order to solve other problems. The lake Lepsøyvatn has a size of ca. 170 X 80 m, and the elevation is 20 m a. s. l. It is surrounded by peat, which has grown into former parts of the lake. Samples for pollen analysis were collected with a Hiller sampler in September 1965, on the south-eastern shore of the lake. At the coring site, brown lacustrine gyttja was found to a depth of 963 cm (Fig. 14). At this depth there was a sharp boundary, with bluish-grey marine silt. Pollen samples 9 and 10 (Fig. 14) are taken respectively just above and below the boundary, and they indicate that the lake was isolated from the sea contem porarily with the immigration of Corylus. Samples for 14C dating were collected in 1966 with a Hiller sampler. Care was taken to sample only the central part of the core, and it was therefore necessary to collect samples from seven corings. Bach sample covered the lowermost 5 cm of the gyttja, just above the sharp boundary to the silt. Even though the samples were collected with a Hiller sampler, the quality is as sumed to be good, due to careful sampling and favourable stratigraphy. Pollen-stratigraphically, the 14C date covers the lower part of the Corylus maximum (Fig. 14). The result was (T-580) 8,380 -+- 180 B. P. (Nydal et al. 1970: 211). HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 331

Thepollen of Ephedra distachya from Lepsøyvatn, discussed by Mangerud (1970: 129), was found in sample 8. The lake Lekvenvatn (Fig. l) is ca. 130 X 100 m, and the elevation is 37 m a.s.l. Parts of the former lake are filled in with gyttja and peat, espe cially along the southern shore. Samples for pollen analysis were collected with a Riller sampler in September 1965, at the southern shore. Below the peat at the coring site there was brown lacustrine gyttja to a depth of 722 cm (Fig. 14), and below the gyttja bluish grey marine silt. The pollen diagram indicates that a short hiatus exists in the bottom of the lacustrine gyttja, since Lekvenvatn emerged earlier from the sea than Lepsøyvatn, while the bottom of the lacustrine gyttja in Lekvenvatn seems to be younger than in Lepsøyvatn. Samples for 14C dating were collected with a 54 mm piston sampler (Geonor) in October 1971. Some simple pollen counts were performed in this core, and the very well-defined Alnus rise (Fig. 14) from ca. 3 % to 40 % was identified. The dated sample was taken from the beginning of this rise and 5 cm upwards, which pollen-analytically corresponds to the depth 680-685 cm in the pollen diagram (Fig. 14). The result was (T-1160) 7,710

-+- 100 B. P. (Gulliksen et al. 1975: 366).

Redeposited pollen Redeposited pollen is usually an important problem in sediments as rich in minerogenic material as the lacustrine sediments in Banktjørn. The pollen content in the laminated glaciomarine silt, which is the source for the mi nerogenic component is, however, extremely low. Therefore the content of redeposited pollen is also low. The marine Hystrix, being resting cysts of dinoflagellates, are the most frequent microfossils in the glaciomarine silt. Hystrix also occur sporadically in the lacustrine silt, where most of it is believed to have been redeposited together with the glaciomarine silt from the gullies. Some of the Hystrix oc curring just above the isolation contact may, however, be connected with inwashed sea water during high tides and storms, or lowering of the threshold which consists of silt and sand. The samples from the glaciomarine silt in cores 1 and 3 contain on aver age 130 % Hystrix, calculated on � (AP + NAP), while those in the lacu strine silt between the Corylus and the Alnus rise, from all cores, contain 1.8 % Hystrix. If we suppose that all these Hystrix are redeposited, it fol lows that secondary pollen from the glaciomarine silt constitutes less than 1.4 % of the pollen deposited during the first 1,100 years of the lacustrine phase, which was the most active period of gully erosion.

Lateral variation of the pollen content In pollen analysis usually one diagram only is worked out from a lake or bog. Thus the question has often arisen to what degree a pollen diagram is representative of the entire basin. Some studies have been conducted on 332 E. SØNSTEGAARD & J. MANGERUD this particular problem (e. g. M. B. Davis 1967, 1968, M. B. Davis et al. 1971, R. B. Davis et al. 1969, Horowitz 1969, Berglund 1973, Pennington 1973), and the conclusion appears to be that within a pond or lake there are small lateral variations in the composition of pollen, especially in the case of small and medium-sized lakes. However, as a rule, there is a differ ence between shallow, onshore, and deeper areas. This may be due to a differential supply of pollen, different floating characteristics of the pollen grains, wind, flow of water, etc. From lake Tveitavatn, Stord, Hafsten (1965)

�-----�-----�2�-r--- �.---,---�,--� Vo o A • 20l0.050 10 20 JO 20JO•oso 10 .O 50�1'010 10 20l0.05060 · - AlnU5 assemb1oge zone

lC�rylus ossembloge l zone

Alnus assemblage zone

as�J'A��age zone

Corylus assemblage zone

Belula-NAP oss. zone

Fig. 13. The variation of the content of Corylus, Pinus, and NAP pollen between the cores l, 2, 3, and 4 in lake Banktjøm. The comparison is only carried out for the Corylus zone. The vertical scale is changed, assuming a constant rate of sedimentation within this zone in each core. The numbers of the pollen samples (see Figs. 15-18) are given. HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 333 has two pollen diagrams, also discussed in Bakka & Kaland (1971: 20). Raf sten explains the small differences between the diagrams in terms of greater influence of local vegetation in the diagram nearest to the shore. The pollen diagrams from Banktjøm, which from a pollen-analyticalview is a medium-sized lake, make it possibleto comparethe composition of ancient pollen in different parts of the basin. Since the boundaries of the Corylus zone are most distinct, it is easiest to make the comparison within this zone. In Fig. 13 the vertical scale is changed assuming a constant rate of sedi mentation between the Corylus rise and the Alnus rise for each single core. Far too large distances between pollen spectra, especially in core 4, reduce the value of the comparisons. The NAP values are in most levels somewhat lower in the central part of the basin (core 1) than in the narrow bay in the south (cores 2, 3, and 4). A similar distribution was found by M. B. Davis et al. (1971). In the case of Banktjøm the reason may be that the narrow bay received more local pollen from the field vegetation and the shore, whereas in the central part the tr((e pollen from more extensive upland dominated. The Corylus curves are similar in the main shapes, except that the pro nounced peak fails to appear in core 4, certainly due to Jack of spectra. The bare rock hills surrounding Banktjøm were probably at that time, as they are at present, dominated by pine forests. Hazel was probably the most important species on the silt area. This may explain the high Corylus values in core 3 where local hazel pollen could be supplied both directly through air and by the brook in the Flåten gully. Pinus pollen is sometimes found to accumulate more in the shallows and along the shores (R. B. Davis et al. 1969), sometimes in the central part of the lake (Pennington 1947). The latter is the case for Banktjøm, where the explanation may be that the pollen content in the central core reflects the regional vegetation, whereas the pollen composition in the bay sediments (cores 2, 3, and 4) is more influenced by both the pollen supply from the local shore vegetation (birch, alder, herbs, etc.), and, as mentioned, from hazel forests on the silt area. In their broad features the pollen curves in the four diagrams are similar in shape. The pollen composition seems, however, to have been slightly in fluenced by vegetation nearest to the individual boring localities. This is in accordance with the conclusions of Berglund (1973) concerning recent pollen sedimentation in a lake which both in shape and size (500 da) can be com pared with BanktjØm (120 da).

The NAP assemblage zone The definition of the zone is that NAP constitutes more than 60 % of the total pollen sum. The main taxa of the herbs are Gramineae, RumexfOxyria, and Artemisia (Figs. 15 and 17). The vegetation must have been very apen, with a few stands of Salix, Juniperus, and Betula (possibly only B. nana). The Pinus pollen is certainly 334 E. SØNSTEGAARD & J. MANGERUD

long-distance transported, and the low percentage of this, and other exotic pollen, indicates that the pollen composition reflects the local vegetation. The deficiency of trees could be due either to cold climate, delayed im migration, or edaphic conditions. During the last few years it has generally been assumed in Scandinavia that the open vegetation in the time interval between the deglaciation and the establishment of the birch forests was not climatically determined. Probably the NAP zone reflects a belt of heath/ tundra vegetation along the retreating ice front, and thus the zone is time transgressive in a profile along the direction of deglaciation. Tree birches (Betula pubescens agg.) grew in Hordaland already during the Allerød Chronozone (Mangerud 1970, 1977, Skår 1975). However, we do not know whether it survived along the coast of Hordaland, or any place in Norway at all, during the following cold Younger Dryas Chronozone (Fægri 1936, 1944 a, Hafsten 1963, Mangerud 1970, 1977). According to Berglund (1966:98) it possibly did not survive in southem Sweden either. At present it is unknown how far the tree birch had to spread during the Preboreal Chronozone before it reached the Os area. The time interval be tween the deglaciation and the Betula rise (establishment of birch forests) is probably only 100-200 years, which may indicate that the birch did not have to spread far and that some birch trees indeed survived the Y ounger Dryas in Hordaland.

The Betula-N AP assemblage zone The lower boundary of this zone is defined by the Betula rise and the cor responding decrease of NAP. The upper boundary is the Corylus rise. Obviously real forests of birch establish themselves during this zone. The beginning of this is contemporary with the lithostratigraphical transition from glaciomarine to postglacial marine silt, probably only 100-200 years after the deglaciation. The con tent of Pinus is still below 10 %, but slightly higher than in the NAP zone. As the production of local pollen must have increased, this indi cates that Pinus grew closer, but certainly not in the Os area. The Juniperus curve has a distinct maximum, before it disappears, close to the upper boundary of the zone. Chronostratigraphically this corresponds to the beginning of the Boreal Chronozone (9,000 B. P.). Skår (1975) found a similar maximum in a diagram from Haukelandsvatnet, Arna (Fig. 1). Also in a diagram (Mamakowa 1968: pl. Il) from Lerøy (Fig. 1), a Juni perus maximum appears in the same stratigraphical level. Mamakowa (1968: 20) assumes, however, an older age for this maximum, referring to the Juniperus expansion as characteristic for the zone transition III/IV (Younger DryasfPreboreal). In Denmark (e. g. Iversen 1960) and southem Sweden (e. g. Berglund 1966: 134) a similar Juniperus maximum is indeed typical for the Younger DryasfPreboreal boundary, as referred to by Mamakowa. However, in Hor daland this maximum appears 1,000 years later. The reason is probably that HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 335

Fig. 14. Pollen diagrams from the lakes Lekvenvatn and LepsØyvatn (Fig. 1).

the light-demanding juniper was shaded out by the birch forests in Denmark and southem Sweden already early in the Preboreal Chronozone. In Horda land the vegetation was more mosaic, and probably Juniperus stands re mained important, especially on areas with thin soil cover, until the pine immigrated and shaded it out. Hazel immigrated nearly contemporaneously with pine, and may also have contributed to the superseding of the juniper, especially on more fertile ground.

The Corylus assemblage zone The lower boundary of this zone is defined by the Corylus rise, and the upper by the Alnus rise. Pinus increases at the same level as Corylus, while Betula strongly decreases. The age of this local zone is from 8,800 years B. P. to 7,700B. P. The shapes of the Corylus- and Pinus curves during the Boreal Chrono zone are often discussed. A double Corylus maximum has been regarded as typical for western Norway (Fægri 1944: 28, Hafsten 1965: 39-40). This, however, is seen neither in our diagrams, except possibly Lekvenvatn (Fig. 336 E. SØNSTEGAARD & J. MANGERUD

Banktjørn,core 1

LEGEND FOR SEDIMENT CLASSIFICATI ON Fig. 15. Stratigraphy and pollen diagram for the coring 1 in Banktjørn. eat P Gytt ja (Loss on ignition > 10%) Gyttja silt (3"/o< toss on ignition

Lam mated silt (Loss on ignition< 3°/o)

Nonlaminated silt ( - �� -J Sand

Til!

Locking samples

14), nor in other diagrams from northern Hordaland (Fægri 1954, HagebØ 1967, Mamakowa 1968, Aa 1974). In Denmark the Pinus maximum appears before the Corylus maximum, in eastern Norway the two maxima are simultaneous, while along the coast of western Norway the Corylus maximum is usually before the Pinus max imum. The latter is also the case in the diagram from Lepsøyvatn (Fig. 14), while the diagram from Lekvenvatn is indistinct in this respect. The four diagrams from lake BanktjØrn (Figs. 15-18) exhibit a few small differences concerning the Corylus and Pinus maxima. The main picture is nevertheless clearly as shown in the diagram from the centre of the lake (Fig. 15); there HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 337

S T R AT l G RAPHY

�wstdne lonns ��dne �nns S�•es (excl lwete� �-r.--- � ..�r-r.r-TN �I,A� "-rr-rrTT�

' ' � i ' IJP f i L� i � i. � • i bl lill! J j ' lllllllll i I.l Lllll ! Il l � l � lill l ) li lli ! l i j l l l l l lli i lll l l I l l : llilllil 10 10 JO s 10 ID s 10 10 5 s s 5 5 1n s s s s s s s 5 10 ID l 5 10 10 30 l 5 10 20 l s i s 10 20 JO s l s l s l s 10 20 l j , . lj . 751 l l Il l

�� \ / l"' l'" � �\r\ l l \ --

-- 1 Il 1\ -- 1\ l

- Calculated ------>-- on I(AP+NAP) I (AP+NAP+spores) -----,... I(AP+NAP)

appears ane Pinus maximum on each side of the Corylus peak. This means that the Pinus and Corylus maxima are more or less simultaneous, as in eastern Norway. The explanation of the double Pinus maximum may be that the pine im migrated before the hazel; it colonized first the fertile silt area, but there the more shadow-tolerating hazel soon after ousted the pine. Later, when the pine forests were well established on the hills with toa poor edaphic con ditions for the hazel, the percentages of pine pollen again reached the same values as before the hazel peak. von Post (1924) was the first to point out that the size of the Corylus max imum increased from continental to oceanic regions, and this conclusion has later been confirmed by several investigations. This appears also clearly in a cross section of the Hordaland county. The maximum percentages of hazel, calculated on sum AP, are in the two eastern sites Voss (Fægri 1950) and Kinsarvik (Anundsen & Simonsen 1967) below 15 %; at Eidslandet (Aa 1974) further west 31 %; and at Stord (Hafsten 1965) and around Ber gen (Hagebø 1967), elose to the coast, ca. 50 %. Os has a localization similar to Stord and Bergen, and the hazel maximum has also a similar 338 E. SØNSTEGAARD & J. MANGERUD

Banktj ørn, core 2

Fig. 16. Stratigraphy and pollen diagram for the coring 2 in Banktjøm. size, being indeed only 28 % in Lekvenvatn, but 42 % in Lepsøyvatn, 60 o/o in the centre of Banktjørn, and more than 80 % in the diagram (Fig. 17) from the bay where we have postulated a strong local influence from the hazel forest.

The Alnus assemblage zone The lower boundary of this zone is defined by the very distinct Alnus rise, which is unambiguous in all diagrams. Due to the few samples analysed higher up, we have so far not defined any upper boundary, and in the present diagrams the zone consequently continues to the surface. HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 339

�RAPHY l;o"..om locustdne fossns r>'>dM,.,,;,,

10 10 t>O 30 10 10 10

1/ l l "'-..

E. Z(AP+NAP) ------Z(AP+NAP on ...pores) ----,.. Z (AP+NAP)

Near the lower boundary first Corylus and thereafter Pinus decrease markedly, and later never reach the peak values in the zone below. The Alnus is usually above 30 % throughout the zone in the present diagrams. In other diagrams from western Norway, the Alnus curve has a tail far below the increase which we used for the definition of the boundary. Similar tails are found in the diagrams from Lekvenvatn and Lepsøyvatn, but are lacking in all the diagrams from Banktjøm. The drastic increase of Alnus is, no doubt, due to Alnus glutinosa (black alder) (cf. Vorren 1973), and the exceptionally high values are the result of a belt of Alnus glutinosa growing along the lake shores. W e believe that the discussed increase also indicates 340 E. SØNSTEGAARD & J. MANGERUD

Banktjørn, core 3

BIOSTRATIG

------Calcul Fig. 17. Stratigraphy and polJen diagram for the coring 3 in BanktjØm. the immigration of Alnus glutinosa. Hagebø (1967) assumes, however, that the tail indicates that Alnus glutinosa immigrated simultaneously with the Corylus maximum, and that the increase is due to increased humidity. We HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 341

'HY

•olved total dlogrom ;;�ne Spore• N .P Lo::,: :::�· l ol!� e " c æ.� l � " � ... �1 11 � , l� c8. "' 1i 1, � � �� i X a � i l � ' ·� l E I � �' ! i E l� t � � .. l � 'Jil � i5 IP i 1H IJ 1 J: l��sl��s .}20 l� 10 l � 10 l: 10 .L n 20 I!ILJ l s l s l s l s lI11s l 5 l 5 ! l U 5 ! il �1J Ill �� l r 1 10 30 !l � 5 1 5 i UJ 10 30 U. Il E � 1/"'... � li;;: 1/ / l/ y V \ ...... -.....__ ! li �-- E / �F 1\ l/ � 1\ 1\ \ " ..\ l �� [\ i\ \l\ l( l/ l l \ \ 1/ l\ \ 1\1\ i\ r; -- l\ ��· '"' ,\ "' l \ \ 1/ ---,; l '\ r/ l\ r; 1\ � \ --- l� ·r; ['-, V J l\ 1/ \ 1/ � � ..__

1/ li l ; �F•E lE 197< Z'(AP+NAP) ------n -Z'(AP+NAP+sporesJ-.., Z'(AP+NAP)

find it more probable that the tails are due to early immigration of Alnus incana (grey alder) (Tallentire 1973, 1974), but this problem must be solved by means of macrofossils, as the two species cannot be distinguished by 342 E. SØNSTEGAARD & J. MANGERUD

Banktjørn, core 4

LIT HOSTRATIGRAPHY BIOSTR

Fig. 18. Stratigraphy and pollen diagram for the coring 4 in BanktjØm. means of pollen morphology. It is worth noting that Mangerud (1977) found up to 11 % of Alnus in Late Weichselian sediments in Hordaland. The sum of the constituents of the mixed oak forests (QM) increased soon after the Alnus rise. However, the Ulmus and Quercus curves have long tails, which go through the entire Corylus zone. Characteristic is a small peak in the lower part of the tail (Figs. 14 and 15). Both Hagebø (1967) and Mamakowa (1968) assumed that a few trees of Quercus and Ulmus grew in Hordaland during the Boreal Chronozone. Such an assump tion can explain the tails, but the delayed main expansion is in that case surprising. The further course of the QM curves is similar to previous results from western Norway: Ulmus decreased, while Tilia increased and reached a peak value approximately simultaneous with Quercus.

Conclusions The sediments from the Flåten gully were deposited in lake BanktjØm di rectly outside the gully, more or less like a delta. This is evident from the lateral thinning of the bed and the fining of the lacustrine silt away from the gully mouth (Fig. 11). The erosion of the gullies in the Os area started about 8,800 years B. P., shortly after the sediments had been raised above sea leve!, and continued HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 343

TIGRAPHY

i\ lb

l l l 1\ 1\ r; l � E. ON> "AA" • 1 .ulated .r P) :-. ------on (AP+NA '*l+ ,r(AP+NAP-Kpores) .r (AP+NAP)

until the gullies reached bed rock in their back ends, about 7,700 years B. P. During the first 400 years the denudation rate was moderate, possibly due to the fact that side branches had not developed yet. The erosion was at a maximum in the period 8,400-7,700 years B. P., and we assume that the entire gully system was active at this time. The denudation rates during the time of gully erosion have been calcu lated to 1.45 mm/year within the silt-covered area and 0.65 mm/year for the whole catchment area (Table 2). The formation of the gullies do not appear to have been dependent on climate, in contrast to the assumption made by Lundqvist & Hjelmqvist (1941) for the gullies in Dalarne in Sweden. This difference is reasonable, as BanktjØrn Iies in an area with oceanic climate and heavy precipitation, while Dalarne has continental climate with relatively little precipitation. Thus precipitation would never be a limiting factor for the gully formation at Banktjørn, as it was during relatively warmfdry periods in Dalarne. The lateral variation of the pollen composition within the lake sediments in Banktjørn is found to be small. The composition is, however, slightly in fluenced by the vegetation nearest to the individual boring localities. The pollen diagrams display a vegetational development mainly similar to that concluded from previous studies in western Norway. A maximum of Juniperus just before the immigration of Pinus and Corylus is discovered, and compared to the 1,000 years older Juniperus maximum in Sweden and 344 E. SØNSTEGAARD & J. MANGERUD

Denmark. The Boreal Pinus and Corylus maxima are found to be nearly simultaneous, as opposed to the general view that the Corylus maximum should be first along the coast of Norway.

Acknowledgements. - Mr. Kjell SØgnen, Mr. Harry lsachsen, Mr. Terje SæbØ anJ cand. real. Odd Pedersen provided assistance with coring. Mr. Johan Lund has per formed some of the pollen preparations. Identifications of pollen were discussed with cand. real. Peter Emil Kaland, cand. mag. Jan Berge, and cand. real. Dagfinn Moe. Peter Emil Kaland kindly permitted us to use unpublished t4C dates from Austrheim. The figures were drawn by Miss Ellen Irgens and Mr. Masaoki Adachi. Dr. Ron J. Steel, cand. real. Tore O. Vorren, cand. real. Kjell Nordseth, and cand. real. Kjell R. Bjørklund kindly read through the manuscript critically. Dr. Steel also corrected the English language. To all these persons we proffer our sincere thanks. We also thank the Norwegian Research Council for Science and Humanities (NAVF) for financial support. These investigations were started by Mangerud. The field work and laboratory analysis were carried out by Sønstegaard in his cand. real. thesis under Mangerud's guidance. We have written the present article together. April 1976

REFERENCES Andersen, S. T. 1961: Vegetation and its environment in Denmark in the Early Weich selian Glacial (Last Glacial). Dan. Geol. Unders. Il Række, Nr. 75, 175 pp. Andersen, S. T. 1966: Tree-pollen rain in a mixed deciduous forest in South Jutland (Denmark). Rev. Palaeobot. Palynol. 3, 263-275. Anundsen, K. & Simonsen, A. 1967: Et Pre-Borealt brefremstøt på Hardangervidda og i området mellom Bergensbanen og Jotunheimen. Univ. i Bergen, Arbok, Mat.-Na turvit. ser. 1967, 1-42. Bakka, E. & Kaland, P. E. 1971: Early farming in Hordaland, western Norway. Nor. Archaeol. Rev. 4, 1-35. Berglund, B. E. 1966: Late-Quaternary vegetation in eastern Blekinge, southeastern Sweden. A pollen-analytical study. I. Late-Glacial time. Opera Botanica 12, 180 pp. Berglund, B. E. 1973: Pollen dispersal and deposition in an area of southeastern Swe den - some preliminary results, pp. 117-129. In H. J. B. Birks & R. G. West (eds.): Quaternary Plant Ecology, Blackwell Sei. Publ. Oxford. Berglund, B. E. & Malmer, N. 1971: Soil conditions and Late-glacial Stratigraphy. Geol. Føren. Stockh. Førh. 93, 575-586. Birks, H. J. B. 1970: Inwashed pollen spectra at Loch Fada, Isle of Skye. New Phytol. 69, 907-820. Bjerrum, L. 1971: Kvikkleireskred. Et studium av årsaksforhold og forbygningsmulig heter. Nor. Geotekn. Inst. 89, 1-14. Bjerrum, L. & Rosenqvist, I. Th. 1956: Norske leirskred og deres geoteknikk. Nor. Geotekn. Inst. 15, 1-11. Bjerrum, L. & Rosenqvist, I. Th. 1957: Some experiments with artificially sedimented clays. Nor. Geotekn. Inst. 25, 1-13. Caldenius, C. 1926: Ravinbildningen i Gustavs. Sver. Geol. Unders. Ser. C, no. 339. Cushing, E. J. 1967: Evidence for differential pollen preservation in Late Quaternary sediments in Minnesota. Rev. Palaeobotan. Palynol. 4, 87-101. Davis, M. B. 1967: Pollen deposition in !akes as measured by sediment traps. Geol. Soc. Am. Bull. 78, 849-858. Davis, M. B. 1968: Pollen grains in lake sediments: Redeposition caused by seasonal water circulation. Science 162, 796-799. Davis, M. B., Brubaker, L. B. & Beiswenger, J. M. 1971: Pollen grains in lake sedi ments: Pollen percentages in surface sediments from southern Michigan. Quaternary Res. l, 450-467. Davis, R. B., Brewster, L. A. & Sutherland, J. 1969: Variation in pollen spectra within !akes. Pollen et Spores XI, 557-571. HOLOCENE GULLY SEDIMENTS IN OS, W. NORWAY 345

De Geer, S. 1914: Niplandskap vid Dalaiilven. Sver. Geo[. Unders. Ser. C, No. 252. Digerfeldt, G. 1975: A standard profile for Littorina transgressions in western Skåne, South Sweden. Boreas 4, 125-142. Ekstrom, G. 1927: Klassifikasjon av svenska åkerjordarter. Sver. Geo/. Unders. Ser. C. 345, 161 pp. Frendin, H. 1931: Bidrag til kiinnedomen om ravinbildningarnas dynamik. Geol. Foren. Stockh. Forh. 53, 81-97. Frey, D. G. 1955: A differential flotation technique for recovering microfossils from inorganic sediments. New Phytol. 54, 257-258. Fægri, K. 1936: Quartiirgeologische Untersuchungen im westlichen Norwegen. l. Ober zwei Priiboreale Klimaschwankungen im siidwestlichsten Teil. Bergens Mus. Arbok 1935, Naturvit. rekke, Nr. 8, 40 pp. Fægri, K. 1944 a: Studies on the Pleistocene of Western Norway. Ill. Bømlo. Bergens Mus. Arbok 1943, Naturvit. rekke, Nr. 8, 100 pp. Fægri, K. 1944 b: On the introduction of agriculture in western Norway. Geo/. Foren. Stockh. Forh. 66, 449-462. Fægri, K. 1950: Studies on the Pleistocene of western Norway. IV. Univ. i Bergen, Ar bok 1949, 1-52. Fægri, K. 1954: On age and origin of the beech forest (Fagus silvatica L.) at Lygre fjorden, near Bergen (Norway). Dan. Geo/. Unders. 2. række, no. 80, 230-249. Fægri, K. & Iversen, H. 1966: Textbook of Pollen Analysis. Munksgaard, København. 237 pp. Gulliksen, S., Nydal, R. & Løvseth, K. 1975: Trondheim Natura! Radiocarbon Measure ments VII. Radiocarbon 17, 364-395. Hafsten, U. 1963: A Late-Glacial pollen profile from Lista, South Norway, Grana Palyn. 4:2, p. 326. Hafsten, U. 1965: The Norwegian Cladium Mariscus communities and their Post-Gla cia! history. Univ. i Bergen, Arbok 1965, Naturvit. Ser. nr. 4, 1-55. HagebØ, S. 1967: Pollenanalytiske undersøkelser av den postglaciale utvikling i Ber gensdalen. Unpublished cand. real. thesis, Universitetet i Bergen. 62 pp. Holmsen, P. 1953: Landslips in Norwegian quick-clays. Nor. Geotekn. Inst. 2, 187-194. Holtedahl, H. 1964: An Allerød fauna at Os, near Bergen, Norway. Nor. Geol. Tidsskr. 44, 315-322. Horowitz, A. 1969: Recent pollen sedimentation in lake Kinneret, Israel. Pollen & Spores XI, 353-384. Hogbom, A. G. 1905: Om s. k. jiislera och om vilkåren fOr dess bildning. Geol. Foren. Stockh. Forh. 27. International Subcommission on Stratigraphic Classification 1971: Preliminary Report on Biostratigraphic Units. 50 pp. 24th Int. Geo!. Congress, Montreal. Iversen, J. 1960: Problems of the early Post-Giacial forest development in Denmark. Dan. Geol. Unders. IV rekke, 4, 1-32. Jørstad, F. A. 1968: Leirskred i Norge. Nor. Geogr. Tidsskr. 22, 214-219. Kaland, P. E. 1974: Ble lyngheiene skapt av fimbulvinter eller menneskeverk? Forskningsnytt, no. 4, 7-14. Krumbein, W. C. & Pettijohn, F. J. 1938: Manual of Sedimentary Petrography. Appleton-Century-Crofts, Inc., N. Y. 549 pp. Lundqvist, G. 1951: Beskrivning till jordartskarta over Kopparbergs liin. Sver. Geol. Unders. Ser. Ca, No. 21, 213 pp. Lundqvist, G. & Hjelmqvist, S. 1941: Beskrivning till kartbladet Hedemora. Sver. Geol. Unders. Ser. Aa. No. 184. Mamakowa, K. 1968: Lille Bukken and LerØy - Two pollen diagrams from Western Norway. Univ. i Bergen Arb., Mat.-Naturvit. Ser. Nr. 4, 42 pp. Mangerud, J. 1970: Late Weichselian vegetation and ice-front oscillations in the Bergen district, western Norway. Nor. Geogr. Tidsskr. 24, 121-148. Mangerud, J. 1977: Late Weichselian marine sediments containing shells, foraminifera and pollen, at Ågotnes, western Norway. Nor. Geo/. Tidsskrift 57, 23-54. Mangerud, J., Andersen, S. T., Berglund, B. E. & Donner, J. J. 1974: Quaternary strati graphy of Norden, a proposal for terminology and classification. Boreas 3, 109-128. Nordseth, K. 1974: Sedimenttransport i norske vassdrag. Geografisk institutt, Univ. i Oslo, 177 pp. 346 E. SØNSTEGAARD & J. MANGERUD

Nydal, R., Løvseth, K. & Syrstad, O. 1970: Trondheim Natura] Radiocarbon Measure· ments V. Radiocarbon 12, 205-237. Pennington, W. 1947: Lake sediments: pollen diagram from the bottom deposits of the north basin of Windermere in northem England. Phil. Trans. Roy. Soc. B 233, 137-175. Pennington, W. 1973: Absolute pollen frequencies in the sediments of ]akes of different morphometry. In H. J. Birks & R. G. West (eds.): Quaternary Plant Ecology, 79-104. Blackwell Sei. Publ. Oxford. von Post, L. 1924: Ur de sydsvenska skogarnas regionala historia under post-arktisk tid. Geol. Foren. Stockh. Forh. 46. von Post, L. 1935: Saters dal. Svenska turistforeningens små hiiften, Stockholm. Rosenqvist, l. Th. 1953: Considerations on the sensitivity of Norwegian quick-clays. Nor. Geotekn. Inst. 2, 195-200. Rosenqvist, l. Th. 1955: Investigations in the clay-electrolyte-water system. Nor. Geotekn. Inst. 9, 1-125. Saamisto, M. 1970: The Late Weichselian and Flandrian history of the Saimaa Lake complex. Soc. Sei. Fennica, Comm. Physico-Math. 37, 107 pp. Skår, K. 1975: lsavsmeltning og postglaciale flaumskredvifter i Arna, Hordaland. Unpublished cand. real. thesis, Universitetet i Bergen. 156 pp. Sønstegaard, E. 1974: Kvartærstratigrafiske undersøkelser i Os, Hordaland. Un published cand. real. thesis, Universitetet i Bergen. 158 pp. Tallentire, P. A. 1973: Some data on the history of alder in Trøndelag, Norway. Grana 13, 18-24. Tallentire, P. A. 1974: The palaeohistory of the grey alder (Alnus incana (L.) Moench.) and black alder (A. glutinosa (L.) Gaertn.) in Fennoscandia. New Phytol. 73, 529-546. Undås, l. 1963: Ra-morenen i Vest-Norge. J. W. Eide, Bergen. 78 pp. Vorren, T. O. 1972: lnterstadial sediments with rebedded interglacial pollen from inner Sogn, West Norway. Nor. Geol. Tidsskr. 52, 229-240. Vorren, T. O. 1973: Glacial Geology of the area between Jostedalsbreen and Jotun heimen, South Norway. Nor. Geol. Unders. 291, 1-46. Wenner, C.-G. 1941: Ravinundersokningar i Gustafs. Geogr. Ann. H. 1-2, 53-101. Ziegler, T. 1974: Materialtransportundersøkelser i norske breelver 1972. Rapp. no. 2-74, Hydrologisk avd. Norges Vassdrags- og elektrisitetsvesen, Oslo. 95 pp. Aa, A. R. 1974: lsavsmeltinga i Eksingedalen og områda omkring. Unpublished cand. real. thesis, Universitetet i Bergen, 144 pp. Aarseth, l. 1971: Deglaciasjonsforløpet i Bjømafjordsområdet, Hordaland. Unpublished cand. real. thesis, Universitetet i Bergen. 106 pp. Aarseth, l. & Mangerud, J. 1974: Younger Dryas and moraines between Hardanger fjorden and Sognefjorden, western Norway. Boreas 3, 3-22.