From pollen percentage to vegetation cover: evaluation of the Landscape Reconstruction Algorithm in western Norway

The Landscape Reconstruction Algorithm (LRA) with the two models REVEALS and LOVE is developed to transform pollen percentage data to vegetation cover. This paper presents the first study to evaluate LRA in a region with large topographic variations within a short distances. The REVEALS model estimates regional vegetation abundance based on pollen assemblages from large lakes (100–500 ha). Pollen surface samples from one large and 28 small lakes are used together with a combination of regionally derived pollen productivity estimates and available estimates from other regions of Europe. The results show a good relationship between REVEALS‐estimated forest cover and vegetation abundance based on the CORINE land‐cover data. The REVEALS results using various sets of pollen assemblages from small lakes were comparable to those using one large lake. Local vegetation abundance using the LOVE model was estimated around 26 lakes. For common taxa, such as Pinus and Poaceae, the LOVE‐based estimates of plant abundance match well with the distance‐weighted plant abundances based on vegetation maps. Our results indicate that the LRA approach is effective for reconstruction of long‐term vegetation changes in western Norway and other regions with high topographic relief when no major gradients exist in the pollen data.     

Relative pollen productivities of the major plant taxa of subtropical evergreen–deciduous mixed woodland in China

Quantifying the relationship between pollen and vegetation is an essential step in the pollen-based quantitative reconstruction of past vegetation cover. In this study, we use the Extended R-Value (ERV) model and a modern dataset of pollen (collected from moss polsters) and related vegetation from 50 sites in the Daba Mountains (subtropical China) to (i) estimate the relevant source area of pollen (RSAP) of the moss samples and the relative pollen productivities (RPPs) of nine major plant taxa-characteristic of the region, and (ii) evaluate the obtained RPPs. The RSAP estimates of moss polsters vary between 225 and 610 m depending on the ERV submodels and models of pollen dispersal and deposition used. The RPP estimates are different from values published in previous studies from temperate and subtropical China. This may be explained by differences in methodology, climate and vegetation (species composition and spatial distribution), of which vegetation is probably the most important factor. The ranking of the RPP estimates for the nine taxa is Pinus > Juglandaceae > D − Quercus (deciduous Quercus) > Poaceae > Rosaceae > Cyperaceae > Anacardiaceae > Castanea > Fabaceae. We use a ‘leave-one-out’ cross-validation strategy and the Landscape Reconstruction Algorithm (LRA) for pollen-based reconstruction of regional and local plant cover to evaluate the ERV model-based RPP estimates. Both the REVEALS (Regional Estimates of VEgetation Abundance from Large Sites)-based and the LOVE (LOcal Vegetation Estimates)-based plant cover using the RPP estimates are closer to the modern vegetation composition than pollen percentages, thus confirming the applicability of the ERV model and the LRA approach in subtropical China.     

The REVEALS model,a new tool to estimate past regional plant abundance from pollen data in large lakes: validation in southern Sweden

The REVEALS model was developed to reconstruct quantitatively regional vegetation abundance (in a 10⁴–10⁵ km² area) from pollen assemblages in large lakes (≥100–500 ha). This model corrects for biases in pollen percentages caused by inter‐taxonomic differences in pollen productivity and dispersal. This paper presents the first case study to validate REVEALS, using empirical data from southern Sweden. Percentage cover of modern regional vegetation in Skåne and Småland, two contrasting vegetation regions, was predicted with REVEALS for 26 key taxa, using pollen assemblages from surface sediments in 10 large lakes, and compared to the actual vegetation within 10⁴ km² compiled from satellite data, forestry inventories, crop statistics, aerial photographs, and vegetation inventories. REVEALS works well in predicting the percentage cover of large vegetation units such as total trees (wooded land), total herbs (open land), total conifers and total broad‐leaved trees, and it provides reasonable estimates for individual taxa, including Pinus, Picea, Betula, Corylus, Alnus, Tilia, Salix spp., Juniperus, Poaceae, Cyperaceae, Cerealia and Secale. The results show great potential for REVEALS applications, including (1) quantitative reconstructions of past regional land cover important for palaeoclimatology and nature conservation, and (2) local‐scale reconstruction of vegetation (<1 km² up to ～ 5 km² area) relevant for palaeoecology and archaeology. Copyright © 2007 John Wiley & Sons, Ltd.     

Advance in Pollen Source Area

The conception of pollen source area has attracted broadly attention since it was proposed in the 1960s. In this paper,it can be subdivided into Normal Source Area of Pollen (NSAP) and Relevant Source Area of Pollen (RSAP) based on reviewing former studies and our recent work of vegetation survey and pollen analysis in the Northern China. The NSAP means a source area of pollen input a sedimentary basin,indicating a distance area beyond where the pollen contents of major pollen types do not show evident increase with increased distance. The RSAP means a source area of pollen assemblages in a basin has the closest relationship with surrounding plants,indicating a distance area beyond where the relationship between pollen assemblage and surrounding vegetation do not show obvious increase with increased distance. The changes of sedimentary basins radii and pollen grain size influence the NSAP and RSAP evidently. Wind speed also changes the NSAP, but does not significantly affect RSAP. The pollen of a sedimentary basin is composed of local pollen,regional pollen and extra regional pollen,which usually account for 30%~45%,25%~60% and 10%~30%,respectively. Local pollen which refers to the pollen coming within the RSAP range,is a reflection of the surrounding vegetation,and has important implications to the reconstruction and succession of local vegetation. Regional pollen which refers to the pollen coming from within the range of NSAP and excluding the range of local pollen,is the important carrier of studies for vegetation restoration and climate change. The pollen that comes outside the range of NSAP is called Extra regional pollen,which is mainly from greater distance by upper air flow,and has no indicator significance to the regional vegetation.     

Quantitative estimates of temperature changes over the last 2700 years in Michigan based on pollen data

Temperatures for the past 2700 yr are estimated using well-dated pollen data from northwestern lower Michigan. The pollen data were sampled from sediment cores of four lakes along a 75-km transect, with fine-grained morainic soils around the two western lakes and sandy outwash soils around the lakes to the east. Climatic reconstructions based on the pollen data from the sandy sites show less temperature change than the reconstructions from the other sites, because variations in the composition of the vegetation at the sandy sites are edaphically restricted. One of the cores studied was dated by counting visible annual laminations (varves). The cores from the other lakes were dated based on three radiocarbon dates per core as well as the historically determined age of the settlement horizons. All the time scales were cross-checked using pollen-stratigraphic correlation between the four sites. A calibration function was developed using a network of modern pollen and climate data covering all of lower Michigan. Based on this calibration function, the 2700-yr reconstruction for Marion Lake indicates an estimated growing-season temperature range of 1.3°C between extreme 30-yr means. Mild conditions persisted prior to ca. A.D. 400, but a cold interval occurred between ca. A.D. 500 and 800. The well-marked warm period evident from ca. A.D. 1000 to 1200 was the last time when temperatures were about equal to the 1931–1960 mean. A prolonged longed cooling occurred after A.D. 1200 and reached 1°C below the 1931–1960 mean by the 1700s. A warming of 0.5°C is indicated from ca. A.D. 1750 to 1850. The estimated temperatures for the 1830s at Marion Lake agree with the instrumental data for that period and this provides some validation of the calibration-function results.     

Late quaternary vegetational change in the Kotzebue sound area, northwestern Alaska

Two sediment cores from Kaiyak and Squirrel lakes in northwestern Alaska yielded pollen records that date to ca. 39,000 and 27,000 yr B.P., respectively. Between 39,000 and 14,000 yr B.P., the vegetation around these lakes was dominated by Gramineae and Cyperaceae with some Salix and possibly Betula nana/glandulosa forming a local, shrub component of the vegetation. Betula pollen percentages increased about 14,000 yr B.P., indicating the presence of a birchdominated shrub tundra. Alnus pollen appeared at both sites between 9000 and 8000yr B.P., and Picea pollen (mostly P. mariana) arrived at Squirrel Lake about 5000 yr B.P. The current foresttundra mosaic around Squirrel Lake was established at this time, whereas shrub tundra existed near Kaiyak Lake throughout the Holocene. When compared to other pollen records from north-western North America, these cores (1) represent a meadow component of lowland. Beringian tundra between 39,000 and 14,000 yr B.P., (2) demonstrate an early Holocene arrival of Alnus in northwestern Alaska that predates most other Alnus horizons in northern Alaska or northwestern Canada, and (3) show an east-to-west migration of Picea across northern Alaska from 9000 to 5000 yr B.P.     

Absolute pollen productivity and pollen–vegetation relationships in northern Finland

Pollen productivity is one of the most critical parameters for pollen–vegetation relationships, and thus for vegetation reconstruction, in either pollen percentages or pollen accumulation rates. We obtain absolute pollen productivity of three major tree types in northern Finland: pine (Pinus sylvestris), spruce (Picea abies) and birch (Betula pubescens ssp. pubescens and B. pubescens ssp. czerepanovii treated as one taxon). Long‐term monitoring records of pollen traps from 15 sites (duration: 5–23 years) and tree volume estimates within a 14 km radius of each trap were compared to estimate pollen productivity (grains m^?3 a^?1) of these trees using a regression method. The slope of the linear relationship between pollen loading and distance‐weighted plant abundance represents pollen productivity. Estimated productivities of pollen (×10⁸ grains m^?3 a^?1) for pine, spruce and birch are 128.7 (SE 31.5), 341.9 (SE 81.3) and 411.4 (SE 307.7), respectively. The birch estimate (P > 0.05) is not as good as the others and should be used with caution. Pollen productivities of pine, spruce and birch in northern Finland are, in general, comparable to those of congeneric species in other regions of Europe and Japan. Although the year‐to‐year variations are significant, our volume‐based estimates of pollen productivity for pine and spruce will be essential for quantitative reconstruction of vegetation in the region. Copyright © 2009 John Wiley & Sons, Ltd.     

Pollen‐inferred palaeoclimate reconstruction in the Alps during the Lateglacial and the early Holocene: how to estimate the effect of elevation and local parameters

The modern analogue technique (MAT) was applied to six pollen sequences from the Belledonne Massif (northwestern French Alps) to estimate the effect of altitude and local parameters on pollen‐based climate reconstruction during the Lateglacial and the early Holocene. The six sites (Le Vivier (345 m a.s.l.), Les Etelles (700 m a.s.l.), La Coche (980 m a.s.l.), Montendry (1330 m a.s.l.), Le Grand Leyat (1660 m a.s.l.), La Gouille (1800 m a.s.l.)) are located in different vegetation belts (mixed deciduous woods, conifer woods, alpine pastures with maple). The main vegetation changes in the past are recorded at each site. The evolution of four climate parameters (coldest month temperature, warmest month temperature, mean annual temperature, annual precipitation) was quantitatively inferred from pollen data using MAT. The curves obtained were compared to the Les Etelles site, which was the least affected by non‐local pollen transport. The results show consistent trends for the climate parameters reconstructed at the different sites. However, the reconstruction does not indicate a decrease in temperature values related to the increasing elevation. Difficulties in reconstructing the altitudinal variations of climate parameters from pollen data during the periods studied are discussed and perspectives for improvement are considered. Copyright © 2009 John Wiley & Sons, Ltd.     

Late quaternary pollen records from the Kobuk and Noatak river drainages, northwestern Alaska

Pollen diagrams from Joe and Niliq Lakes date to ca. 28,000 and 14,000 yr B.P., respectively. Mesic shurb tundra grew near Joe Lake ca. 28,000 to 26,000 yr B.P. with local Populus populations prior to ca. 27,000 yr B.P. Shrub communities decreased as climate changed with the onset of Itkillik II glaciation (25,000 to 11,500 yr B.P.), and graminoid-dominated tundra characterized vegetation ca. 18,500 to 13,500 yr B.P. Herb tundra was replaced by shrub Betula tundra near both sites ca. 13,500 yr B.P. with local expansion of Populus ca. 11,000 to 10,000 yr B.P. and Alnus ca. 9000 yr B.P. Mixed Picea glauca/P. mariana woodland was established near Joe Lake ca. 6000 yr B.P. These pollen records when combined with others from northern Alaska and northwestern Canada indicate (1) mesic tundra was more common in northwestern Alaska than in northeastern Alaska or northwestern Canada during the Duvanny Yar glacial interval (25,000 to 14,000 yr B.P.); (2) with deglaciation, shrub Betula expanded rapidly in northwestern Alaska but slowly in areas farther east; (3) an early postglacial thermal maximum occurred in northwestern Alaska but had only limited effect on vegetation; and (4) pollen patterns in northern Alaska and northwestern Canada suggest regional differences in late Quaternary climates.     

Charcoal signatures defined by multivariate analysis of charcoal records from 10 lakes in northwest Wisconsin (USA)

We show how sedimentary charcoal records from multiple sites within a single landscape can be used to compare fire histories and reveal small scale patterns in fire regimes. Our objective is to develop strategies for classifying and comparing late-Holocene charcoal records in Midwestern oak- and pine-dominated sand plain ecosystems where fire regimes include a mix of surface and crown fires. Using standard techniques for the analysis of charcoal from lake sediments, we compiled 1000- to 4000-yr-long records of charcoal accumulation and charcoal peak frequencies from 10 small lakes across a sand plain in northwestern Wisconsin. We used cluster analysis to identify six types of charcoal signatures that differ in their charcoal influx rates, amount of grass charcoal, and frequency and magnitude of charcoal peaks. The charcoal records demonstrate that while fire histories vary among sites, there are regional patterns in the occurrence of charcoal signature types that are consistent with expected differences in fire regimes based on regional climate and vegetation reconstructions. The fire histories also show periods of regional change in charcoal signatures occurring during times of regional climate changes at ~ 700, 1000, and 3500 cal yr BP.     

Paleohydrological implications of Holocene peatland development in northern Michigan

Sediment, pollen, and plant macrofossil stratigraphies from two small oligotrophic Chamaedaphne-Sphagnum peatlands provide data about local hydrologic changes in northern Michigan during the Holocene. Gleason Bog started about 8000 yr B.P. as a shallow pond that supported rich fen vegetation. After it was partly filled with peat and sand (about 4000 yr B.P.), the vegetation changed to oligotrophic bog. At Gates Bog paludification starting about 3800 yr B.P. caused peat accumulation over sand without an initial pond phase. The onset of peat accumulation at both sites is attributed to a rise in the water table resulting from the onset of cool and moist late Holocene climates. The water table of Gleason Bog is linked to the water level of adjacent Douglas Lake, which may have undergone a simultaneous rise. The results emphasize the individuality of hydrological conditions and hydroseral development in northern Michigan peatlands.     

Changes in the parkland-boreal forest boundary in northwestern Ontario over the Holocene

Changes in vegetation were tracked from a well-dated sediment core from a boreal lake, Lake 239, at ～200-year resolution over the Holocene. This presently oligotrophic lake is located ～100-km east from the present-day parkland-forest ecotone in northwestern Ontario. Near-shore sediment core transects from Lake 239 have previously shown this lake was at least 8-m lower than present in the mid-Holocene, or ～58% less lake volume in comparison to today. Large shifts were expected in the terrestrial vegetation if the low lake levels were related to climate. The core from Lake 239 shows increases in the relative abundance and concentration of pollen such as Cupressaceae and Ambrosia, indicating a more open boreal forest between ～4500–8000 cal yr BP. Pollen-based inferences of average, summer and winter temperatures suggest that temperatures were on average up to 1–2 °C warmer than today, with winter temperatures up to 4 °C warmer. The pollen inferences also suggest enhanced precipitation, likely in the summer, but with an overall increase in evaporation and evapotranspiration resulting in reduced effective moisture. To assess regional climate changes, pollen-based reconstructions of temperature and precipitation were developed and synthesized from sediment cores from eight previously published lakes, from which pollen sites were available to both the west and east of Lake 239, spanning present-day prairie lakes to forested lakes up to 300 km east of the prairie-boreal ecotone. All sites show shifts in pollen assemblages that indicate a warm mid-Holocene period; prairie sites west of the Experimental Lakes Area (ELA) show mid-Holocene decreases in precipitation relative to today, whereas sites near or east of ELA show consistent increases in precipitation, but with increased temperatures and enhanced evaporation during the mid-Holocene.     

Importance of climate,forest fires and human population size in the Holocene boreal forest composition change in northern Europe

The relative importance of climate, forest fires and human population size on long‐term boreal forest composition were statistically investigated at regional and local scales in Fennoscandia. We employ pollen data from lakes, reflecting regional vegetation, and small forest hollows, reflecting local vegetation, from Russia, Finland and Sweden to reconstruct the long‐term forest composition. As potential drivers of the Holocene forest dynamics we consider climate, generated from a climate model and oxygen isotope data, past forest fires generated from sedimentary charcoal data and human population size derived from radiocarbon dated archaeological findings. We apply the statistical method of variation partitioning to assess the relative importance of these environmental variables on long‐term boreal forest composition. The results show that climate is the main driver of the changes in Holocene boreal forest composition at the regional scale. However, at the local scale the role of climate is relatively small. In general, the importance of forest fires is low both at regional and local scales. The fact that both climate and forest fires explain relatively small proportions of variation in long‐term boreal vegetation in small forest hollow records demonstrates the complexity of factors affecting stand‐scale forest dynamics. The relative importance of human population size was low in both the prehistorical and the historical time periods. However, this is the first time that this type of data has been used to statistically assess the importance of human population size on boreal vegetation and the spatial representativeness of the data may cause bias to the analysis.     

Middle Holocene dry period in the northern Midwestern United States: Lake levels and pollen stratigraphy

Four cores from Lake Mendota in the Yahara River valley of south-central Wisconsin provide pollen, charcoal, and sediment-stratigraphic evidence of a middle Holocene dry period in the northern Midwest. The lake level was lower between about 6500 and 3500 yr B.P. indicating that runoff from the upper Yahara River basin may have ceased during the drier seasons of the year. Derived estimates of precipitation between 6500 and 3500 yr B.P. indicate a decrease of about 10% (down to about 700 mm) from the present precipitation value of 800 mm. A warming between 6500 and 3500 yr B.P. is also interpreted based on a change of vegetation from rich mesophytic forest before 6500 yr B.P. to Quercus savanna, with increased charcoal abundances after that time. After 3500 yr B.P., a closed Quercus forest and decreased charcoal suggest a cooler and wetter climate. The qualitative paleoecological interpretations, quantitative precipitation reconstructions based on pollen from the region (including the Lake Mendota data), and estimates of hydrologic budget based on lake-level changes, all show indications of a dry and warm middle Holocene. Evidence from a large regional array of sites also supports these interpretations.     

Modern pollen assemblages from surface lake sediments and their environmental implications on the southwestern Tibetan Plateau

The relationships amongst modern pollen assemblages, vegetation, climate and human activity are the basis for reconstructing palaeoenvironmental changes using pollen records. It is important to determine these relationships at regional scales due to the development of vegetation under different climatic conditions and human activities. In this paper, we report on an analysis of modern pollen assemblages of 31 surface lake samples from 31 lakes (one sample per lake) on the southwestern Tibetan Plateau where the knowledge of modern pollen and their relationships with vegetation, climate and human activities is insufficient. The region includes five vegetation zones: sub‐alpine shrub steppe, alpine steppe, alpine meadow and steppe ecotone, mountain desert and alpine desert. The lakes span a wide range of mean annual precipitation (50–500 mm) and mean annual temperature (?8 to 6 °C). Modern pollen assemblages from our samples mainly consist of herb taxa (Artemisia, Cyperaceae, Poaceae, Chenopodiaceae, etc.) and some tree taxa (Pinus, Fagaceae, Alnus, etc.). The results indicate that modern pollen assemblages are able to reflect the main vegetation distribution. Redundancy analysis for the main pollen types and environmental variables shows that precipitation is the leading factor that influences the pollen distribution in the study area with the first axis capturing 13.7% of the variance in the pollen data set. The Artemisia/Chenopodiaceae ratio is valid for separating the desert component (<2) from the steppe and other vegetation zones (>2) but is unable to distinguish moisture variations. The Artemisia/Cyperaceae ratio is able to identify meadows (<1) and steppes (>1) and can be used as a moisture index on the southwestern Tibetan Plateau. Our results show that an appropriate range is needed for a modern pollen data set in order to perform pollen‐based quantitative climate reconstructions in one region. It is essential to perform modern studies before using pollen ratios to reconstruct palaeovegetation and palaeoclimate at a regional scale.     

Stick-Nest Rat Middens as Sources of Paleoecological Data in Australian Deserts

The existing paleoenvironmental data from the Australian arid zone lack sensitivity and come from only a few sites. Macrofossils and pollen from four dated middens of the stick-nest rat (Leporillus spp.) were analyzed from two sites in Western Australia. Animal and plant macrofossil remains were well preserved and provided evidence of change in species distribution within the last 1150 yr. Brush-tail possum and golden bandicoot have contracted their ranges in the recent past, possibly since the introduction of cats into Australia. An undescribed lacewing was also a significant find. Pollen preserved in parts of the same midden and in middens from different sites indicates that records are sensitive to the composition of the local vegetation when the midden was built. Pollen spectra are quite different from playa lakes, which record largely regional vegetation. Pollen preserved in the fecal pellets, desiccated urine, and grass mat nesting material provided similar information but some differences were apparent, suggesting dietary preferences were reflected in the fecal component. The pollen record suggested a trend to less-wooded vegetation cover in central Australia between 900 and 300 yr B.P.     

Calibrating pollen data in climatic terms: Improving the methods

When properly calibrated, Holocene pollen data provide an important source of quantitative information about Holocene climates. Multiple linear regression of modern climate and pollen data allows the development of statistical calibration functions that transform percentages of certain pollen types into quantitative estimates of climatic variables, and these functions, when applied to Holocene pollen data, yield estimates of climatic variables for past times. Confidence intervals for the climatic variables provide estimates of the statistical errors. These interval estimates are based upon the following statistical assumptions: (1) the regression model is appropriate; (2) the errors in measuring the climate variables are independent, normally distributed and have constant variance; and (3) no outliers are present. We outline the steps to be followed in calculating calibration functions, including (1) selecting the calibration region; (2) selecting a pollen sum; (3) analyzing scatter diagrams of a given climate variable against each pollen type; (4) deleting outliers and transforming pollen data; (5) performing the regression; and (6) testing the appropriateness of the statistical assumptions. We used available computer programs for most of this study. In addition, we developed new software to compute the Moran statistic to test for spatial autocorrelation among the regression residuals, using the dual of the Voronoi diagram to describe the spatial relationships among the sites. In order to illustrate the sequence of procedures, we used data from the lower peninsula of Michigan to develop a calibration function for July mean temperature and then used Holocene pollen data from central lower Michigan to estimate past temperatures.     

Late Wisconsin vegetational and climatic history at Kylen Lake, northeastern Minnesota

Kylen Lake, located within the Toimi drumlin field, is critically positioned in relation to Late Wisconsin glacial advances, for it lies between the areas covered by the Superior and St. Louis glacial lobes between 12,000 and 16,000 yr B.P. The pollen and plant-macrofossil record suggests the presence of open species-rich “tundra barrens” from 13,600 to 15,850 yr B.P. Small changes in percentages of Artemisia pollen between 14,300 and 13,600 yr B.P. appear to be artifacts of pollen-percentage data. Shrub-tundra with dwarf birch, willow, and Rhododendron lapponicum developed between 13,600 and 12,000 yr B.P. Black and white spruce and tamarack then expanded to form a vegetation not dissimilar to that of the modern forest-tundra ecotone of northern Canada. At 10,700 B.P. spruce and jack pine increased to form a mosaic dominated by jack pine and white spruce on dry sites and black spruce, tamarack, and deciduous trees such as elm and ash on moist fertile sites. At 9250 yr B.P. red pine and paper birch became dominant to form a vegetation that may have resembled the dry northern forests of Wisconsin today. The diagram terminates at 8410 ± 85 yr B.P. Climatic interpretation of this vegetational succession suggests a progressive increase in temperature since 14,300 yr B.P. This unidirectional trend in climate contrasts with the glacial history of the area. Hypotheses are presented to explain this lack of correspondence between pollen stratigraphy and glacial history. The preferred hypothesis is that the ice-margin fluctuations were controlled primarily by changes in winter snow accumulation in the source area of the glacier, whereas the vegetation and hence the pollen stratigraphy were controlled by climatic changes in front of the ice margin.     

Reconstructing pathways of aeolian pollen transport to the marine sediments along the coastline of SW Africa

The distribution of pollen in marine sediments is used to reconstruct pathways of terrigenous input to the oceans and provides a record of vegetation change on adjacent continents. The wind transport routes of aeolian pollen is comprehensively illustrated by clusters of trajectories. Isobaric, 4-day backward trajectories are calculated using the modelled wind-field of ECHAM3, and are clustered on a seasonal basis to estimate the main pathways of aeolian particles to sites of marine cores in the south-eastern Atlantic. Trajectories and clusters based on the modelled wind-field of the Last Glacial Maximum hardly differ from those of the present-day. Trajectory clusters show three regional, and two seasonal patterns, determining the pathways of aeolian pollen transport into the south-eastern Atlantic ocean. Mainly, transport out of the continent occurs during austral fall and winter, when easterly and south-easterly winds prevail. South of 25°S, winds blow mostly from the west and southwest, and aeolian terrestrial input is very low. Generally, a good latitudinal correspondence exists between the distribution patterns of pollen in marine surface sediments and the occurrence of the source plants on the adjacent continent. The northern Angola Basin receives pollen and spores from the Congolian and Zambezian forests mainly through river discharge. The Zambezian vegetation zone is the main source area for wind-blown pollen in sediments of the Angola Basin, while the semi-desert and desert areas are the main sources for pollen in sediments of the Walvis Basin and on the Walvis Ridge. A transect of six marine pollen records along the south-western African coast indicates considerable changes in the vegetation of southern Africa between glacial and interglacial periods. Important changes in the vegetation are the decline of forests in equatorial Africa and the north of southern Africa and a northward shift of winter rain vegetation along the western escarpment.