1 Introduction

Palynology refers to the study of pollen , spores and other microfossils that have an organic wall. More traditionally, this has focussed on microscopic description of the shapes, features and ornamentation that allows their description and taxonomic differentiation (Erdtman 1969; Moore et al. 1991). Pollen analysis uses the associations of different taxa, and variations in these associations over time, to reconstruct vegetation assemblages of past environments (Erdtman 1969; Moore et al. 1991).

Pollen analysis is a useful technique for palaeo-environmental reconstructions because it allows quantitative comparison of the relative presence of different vegetation species over time. Such studies have been most commonly applied to Holocene sedimentary deposits of lakes (Davis 1969; Kershaw 1976; Pennington 1979; Ellison 1994), bogs (Colhoun et al. 1991) and the off-shore ocean floor (Fredoux 1994), but studies to date have shown that application of pollen analysis to estuarine settings are also of value. Pollen analysis has been applied to past climate reconstructions in lacustrine and off-shore seabed deposits, and, while estuarine sediments have the potential to render such palaeo-climate records (Kraus et al. 2003; Malamud-Roam et al. 2006), they have been more widely used to reconstruct estuarine evolution (Woodroffe et al. 1989; Armour and Kennedy 2005).

Pollen analysis is associated with coring and radiometric dating methodologies as detailed by other chapters in this volume. These techniques are complementary with other retrospective methods and these include investigation of human records by interviews with local residents, repeated landscape photography, and, for older records, perusal of historical archives and interpretation of archaeological data (Dahdouh-Guebas and Keodam 2008). Spatial analysis of landscape change using aerial photography has been used to reconstruct vegetation changes from the 1940s to present. Since the 1970s reconstruction by remote sensing and geographical information systems analysis has been possible (Lucas et al. 2002). Pollen analysis of estuarine cores can contribute a longer term sedimentary and palaeo-ecological record to these other retrospective methods.

2 Pollen and Spores

A pollen grain is the male gametophyte of a flowering plant, while a spore is the gametophyte-producing cell of a lower plant. The quantity produced varies between species and between seasons. The amount of pollen and spores produced by different species is dependent on its evolved pollen /spore dispersal strategy. Wind-pollinated genera such as Pinus (pine), Elmus (elm), Betula (birch) and Quercus (oak), as well as weeds such as Plantago, produce large amounts of pollen and, therefore, have high representation in pollen assemblages found in estuarine wetland deposits. Insect-pollinated genera are usually those with bright flowers and echinate (spiny) pollen such as Hibiscus (a mangrove associate) (Fig. 18.1f). These produce far fewer pollen and their pollen are much less easily dispersed, so that insect pollinated taxa are largely absent or poorly represented in the pollen assemblages.

Fig. 18.1
figure 1

Spores and pollen of some species identified in the key (Table 18.1). (a) Davallia solida (b) Stenochlaena palustris (c) Asplenium nidus (d) Geniostona insulere (e) Wollastonia biflora (f) Hibiscus tiliaceus (g) Cocos nucifera (h) Excoecaria agallocha

Full size image

Both pollen and spores have an outer covering that is resistant to decay in anaerobic conditions, and by shape, size, aperture type and ornamentation (Fig. 1), enables the grain to be identified to its source species, genus or family depending on the degree of differentiation of features.

2.1 Description and Identification of Pollen and Spores

Palynology includes the description and identification of pollen and spores (Erdtman 1969; Moore et al. 1991), using microscopic examination of the grain’s size, shape, apertures such as slits or pores, and surface ornamentation. Taxonomic identification of pollen can be usually made to genus level, and sometimes can be to species level when features are sufficiently distinct. Pollen and spores differ from each other primarily by their relative size and shape; some, such as Pinus, have bisaccate wings which are most distinctive. A primary level of differentiation is the types of apertures in the grain, such as those with pores, which are called porate, those with slits—called colpate, and those with a combination of these (colporate) where the pore is usually in the centre of the slit. Combined with aperture identification is a reference to the numbers of each, such as: “mono-” is just one, “tri-” is three and “peri-” indicates many.

A secondary level of differentiation is the type of surface ornamentation on the grain, such as: psilate—minimal or no ornamentation; echinate—with spikes or spines; and reticulate—a net like pattern. There are other types as described by Erdtman (1943, 1952, 1969), Faegri and Iversen (1975) and Moore et al. (1991). These spines, bumps and spikes, for example, can all vary in length, size, and density. A further level of differentiation includes features such as the overall dimensions of the grain, the ornamentation if of a pronounced type, and the relative widths of the pollen or spore inner wall (exine) and outer wall (sexine) on the surface.

Pollen morphological descriptions from which to identify pollen and spore assemblages can either be obtained from the published literature (such as Moore et al. 1991), or, where no previous pollen analysis work has been undertaken, a key of pollen and spores can be developed from a reference collection.

Local or regional pollen identification reference literature is available for many parts of the world, but is not comprehensive. For a region of interest a database search can usually find references to some of the identified pollen , which is most useful when the information is organised into an identification key. A more general dichotomous identification key was published by Moore et al. (1991), but it is cumbersome to use and does not extend to regional species. Regional and local keys provide more ease of use and provide better identification of local species [e.g. Rull (2003) for Eastern Venezuela and Mao et al. (2012) for Southern China ]. For an area that has not previously been studied, such as the below case study on Samoa , reference slides of local pollen or spores can be prepared. A reference collection is developed by collecting anthers from the mature flower of species likely to appear in the pollen /spore assemblages such as wind pollinated flowers, or the fronds of ferns producing spores , and then treating these to a reduced version of the pollen concentration procedure using standard chemical treatments as described by Erdtman (1969) and Faegri and Iversen (1975; see below).

An example of a spore and pollen key, developed for the estuaries of Samoa , is shown in Table 18.1. This key was developed from reference material and comparison with published descriptions of pollen and spores from elsewhere but known to be present locally (Erdtman 1952; Leopold 1969; Heusser 1971; Tissot 1973; Rao and Ong 1974; Faegri and Iversen 1975; Muller and Caratini 1977; Lloyd 1980; Johansson 1987; Caratini and Tissot 1988; Ward 1988; see Fig. 18.1). This key highlights some common difficulties in palynological research; it is difficult or impossible to distinguish between genera and species within the wetland families Poaceae and Cyperaceae (Moore et al. 1991; Deng et al. 2006). In pollen of the family Rhizophoraceae, Muller and Caratini (1977) described species within the Rhizophora genus that cannot be distinguished from each other with certainty on the basis of their pollen features. For example in tricolporate Rhizophora pollen , R. stylosa and R. mucronata, both have psilate-reticulate-perforate ornamentation, whereas R. apiculata, R. lamarkii and R. mangle all show reticulate-rugulate ornamentation. Similarly, the pollen of Bruguiera and Ceriops are indistinguishable from each other under the light microscope (Grindrod 1988). In the case of the Samoan example the problem is resolved by reduced species diversity, with genera such as Ceriops being absent so allowing the identification of these types to Bruguiera.

Table 18.1 Key to the pollen and spores of estuaries of Samoa, as well as mangrove areas of Tonga
Full size table

2.2 Contemporary Surface Pollen Assemblages

To resolve the relative representation of species in the contemporary vegetation community, relative to the pollen and spores in the sediment (Fig. 18.2), the taxonomic structure of the local vegetation community can be compared with proportional representation in sediment surface samples. Surface sampling within estuaries then allows reconstruction of local pollen assemblages, allowing interpretation of past vegetation from fossil pollen deposits, which are sampled by coring into the grey sediments of Fig. 18.2.

Fig. 18.2
figure 2

Relationships between intertidal mangrove species zonation, modern pollen and spores assemblage and fossil pollen and spores assemblage

Full size image

The distance a pollen grain is transported in an estuary depends on the depth of water and the degree of turbulence of the water (Brush and DeFries 1981), which varies with geomorphic setting and position in the estuary. Pollen in surface sediments in the Potomac estuary, U.S.A., were found to represent most of the trees growing in the adjacent region and to preserve their relative abundances (Brush and DeFries 1981). Some differential dispersal occurred owing to variations in the settling characteristics of different pollen types. Surface sample assemblages of pollen from the Whangapoua Estuary, N.Z., reflected a high degree of similarity between surface pollen and vegetation (Deng et al. 2006). They found that samples from open, saline tidal inundation areas had higher percentages of widely dispersed tree pollen , such as mangroves , whereas the pollen of the saltmarsh species Juncus kraussi was not represented in surface samples probably because of the fragility of its pollen grain.

Some wind-pollinated species have evolved wing-like bladders on the pollen to better allow long distance transport, such as the bisaccate shape of the conifer pollen of Pinus and Picea (Erdtman 1969: 253; Moore et al. 1991: Pl 5). This type of feature may produce differential sorting in the pollen population and distort the sedimentary record below the water surface with sinking of most deciduous pollen and flotation of most conifer pollen (Hopkins 1950). Special care should be taken therefore, in regions with abundant or common conifer species such as in North America . Floating pine pollen may raft and produce deposits with high concentrations in areas that have no nearby pine vegetation (Traverse and Ginsberg 1966). Eddies and water currents are known to cause differential settling of pollen of different grain sizes in lakes (Davis and Brubaker 1973), with genera such as pine affected but others, such as birch and oak, unaffected (Davis et al. 1971). While such effects must be taken into consideration when interpreting pollen assemblages that include species such as pine, other pollen , such as the mangroves Rhizophora and Bruguiera, which have pollen of similar sizes (Table 18.1), are more straightforward to interpret.

Studies of pollen in surface samples within mangrove ecosystems have shown that there is a high Rhizophora proportion in, and immediately adjacent to, the Rhizophora zone (Muller 1959; Cohen and Spackman 1977). Wijmstra (1969) used modern surface samples from mangroves to interpret fluctuations in sea-level from the Cretaceous to the Plio-Pleistocene in Surinam, interpreting a 90 % proportion of Rhizophora as indicative of a Rhizophora stand, and a 30 % proportion as representative of a site immediately adjacent to a Rhizophora stand, such as a seaward mud flat . Bartlett and Barghoorn (1973) interpreted the depositional environments of fossil pollen assemblages by reference to modern pollen distributions using the following conditions: (1) the percentage of pollen of Rhizophora plus Avicennia in a mangrove swamp where these species are growing can be between 45 and 95 %; (2) near shore sediments seaward of the mangrove swamps contain from 30 % to 50 % Rhizophora pollen ; and (3) sediments immediately landward of mangroves contain 45–10 % Rhizophora pollen , but also declining to less than 10 %.

High proportions (>50 %) of mangrove pollen , particularly when combined with high pollen concentration in the source sediment, therefore indicates a mangrove environment, declining to 50 % or less in immediately adjacent environments (Grindrod 1985, 1988; Chappell and Grindrod 1985; Woodroffe et al. 1985, 1989; Ellison 1989, 2005; Behling et al. 2001; Ellison and Strickland 2015). These patterns are shown in Fig. 18.2, with high proportions of Rhizophora pollen immediately under the Rhizophora zone, and high proportions of Bruguiera pollen immediately under the Bruguiera zone.

Relationships between saltmarsh vegetation and pollen /spore records have been less extensively studied and have been thought to be problematic, with taxonomically lower herbaceous vegetation being more prone to long-distance transport of pollen (Beecher and Chmura 2004). Graminoid pollen are less reliable indicators of source vegetation, while Plantago maritima and Cyperaceae pollen are better indicators of their source zones (Beecher and Chmura 2004). Saltmarsh pollen are better indicators of salinity conditions (Byrne et al. 2001), than elevation (Beecher and Chmura 2004), with the Cyperaceae being indicative of freshwater conditions for example.

To exemplify how a modern analogue can be established from fieldwork and laboratory analysis, the relative pollen representation from three surface samples was collected from different mangrove zones in American Samoa : Rhizophora , mixed Rhizophora/Bruguiera, and Bruguiera (Fig. 18.3). They were concentrated for pollen analysis in a similar manner to the stratigraphic samples, and identified using the pollen key shown in Table 18.1. Results show high representation in the surface pollen assemblage by the mangrove zone present at each site, as described in Fig. 18.2. Study of the distribution of pollen in surface sediments provides a modern analogue to allow interpretation of fossil pollen and spore assemblages (Fig. 18.2). Assemblages can also be compared with surface distributions found in different mangrove settings elsewhere in the world (Muller 1959; Grindrod 1985). The Samoan mangroves occupy a very narrow margins around relatively small estuaries compared with locations elsewhere, and, consequently, the mangrove pollen appear diluted by pollen from adjacent environments, compared with reference data compiled from larger mangrove settings.

Fig. 18.3
figure 3

Pollen and spore assemblages in surface samples in different mangrove forest types, Samoa

Full size image

2.3 Fossil Pollen and Spores Assemblages

Most pollen analysis research has examined sequential fossil pollen deposits for reconstruction of past vegetation patterns adjacent to lakes (Davis 1969; Kershaw 1976; Pennington 1979; Ellison 1994) or bogs (Colhoun et al. 1991), with a focus on climate or environmental change interpretations. As estuaries also provide wet, anaerobic depositional conditions, estuarine sediment allows the long-term preservation of these records. Techniques are the same as for other depositional settings and include sampling of stratigraphic cores, radiometric dating, fossil pollen identification and counting, and pollen diagram interpretation.

3 Coring Strategy and Fossil Pollen Analysis

Optimal coring site selection for reconstructing a pollen record depends upon the research question. Traditional pollen analysis work from lacustrine deposits (Davis 1969; Pennington 1979; Ellison 1994) suggests that the deeper central parts of an estuary (“central basin ” of Dalrymple et al. 1992; Skilbeck et al. 2017) usually offer the greatest thickness of sediment and most anaerobic conditions. However, the centre parts of an adjacent wetland can also provide a location representative of the larger area, because it is less influenced by local edge factors such as human developments, and will be more likely record more regional factors.

3.1 Coring

Coring techniques are covered in a separate chapter of this book (Skilbeck et al. 2017), so comments here only relate to pollen analysis where the samples are collected from cores. For pollen analysis research it is best to core at low tide . Sample contamination is further prevented by dismantling and washing the corer each time used if it is of a type requiring reuse of components, and cleaned tools such as spatulas used for sub-sampling .

The corer diameter controls the amount of sample that can be obtained from each stratigraphic depth, which for pollen analysis can only require 1 cm3, not including other analyses that may be of interest to the research question. Sidewall opening corers (e.g. Hiller corer, Russian Peat sampler; see Skilbeck et al. 2017) have the advantage of not causing compaction so are better used for research questions where elevation is critical, such as for sea-level reconstruction . As both types of corers allow the sampler chamber to be closed before the core section is retrieved, contamination by layers above is avoided. Additionally, the core sampler should be washed before opening the cylinder, to avoid contamination by material stuck to the outside of the corer during retrieval.

3.2 Sample Concentration

Preparation of pollen for slide-based analysis involves removal of as much extraterraneous material from the sediment as possible in order to concentrate the pollen . Chemical concentration removes unwanted mineral, and other macrofossil material, just leaving the pollen and spores , and the pollen concentration procedure developed by Erdtman (1943, 1969), Faegri and Iversen (1975), and Moore and Webb (1978) is summarised below. It should be carried out in a fume cupboard with due care because of the chemicals required.

The number of samples concentrated at each session is, in the author’s experience, controlled by the capacity of the centrifuge used. Samples, and any ancillary equipment, should be kept separate, and a written record of sample reaction during each step kept, as this gives insight into its makeup. An individual sample volume of about 1 cm3 is placed in a 15 ml polypropylene test tube. Sample spacing is determined by the aims of the study and the sedimentation rate . If pollen concentration is to be determined a commercially available exotic pollen tablet (called a spike) is added, with the number of tablets added subject to the pollen concentration present. It is best to have the exotic pollen concentration in the slide presentation about equivalent in concentration to the fossil pollen .

Calcium carbonate is then removed from the sediment by adding 10 % HCl, and gently stirring with a wooden applicator stick while the test tube is placed in a warm water bath at about 90 °C. This stage may not be necessary for samples from estuaries with little carbonate sediment, but from marine-influenced settings reactive effervescence can be strong. This stage also disperses exotic pollen if added to determine pollen concentration. Acid digestion should be repeated until effervescence ceases, and then washed with distilled water, with the supernatant removed following centrifuging after each treatment.

Humic compounds are removed by adding 10 % NaOH or 10 % KOH, with the test tubes placed in a 100 °C water bath for 15 min, gently stirring and adding distilled water to prevent increased concentration by evaporation. The sample is repeatedly washed with distilled water until the supernatant runs clean, which can require as many as eight repeats for peaty sediment. Difficulty in cleaning at this stage can be an indication that the sample has been exposed to the air, in which case pollen analysis may not be possible as the pollen will also have been broken down.

The addition of 10 % NaOH or 10 % KOH has the additional effect of further dispersing the remaining sediment. If material larger than 150 μm is still present the sample can be passed through a 150 μm sieve into a beaker to remove the large organic fragments. The sample is then reconcentrated using a centrifuge.

For removal of silicates , hydrofluoric acid (concentration of 1.15 g/ml for 48 % soln.) treatment will be needed. This is necessary for most estuarine sediments sourced from the erosion of bedrock . Polypropylene test tubes are recommended for this reason. Sufficient HF is added to each test tube so it is at least half full. The tubes are then placed in a rack in a hot water bath for an hour, or left overnight at room temperature. After decantation of the HF from the sample, it is then washed in distilled water, and examined using a smear on a slide to see if treatment was effective. If glassy mineral material is still present, then a further treatment with HF is necessary to eliminate all silicates . In some situations, where silicate sediment sources are low or where it may be desirable to keep sedimentary records of ash falls or siliceous sponge spicules, HF treatment is not used (Ellison 1989, 1996).

To render the pollen wall ornamentation more visible, acetylosis can be employed. The sample is first washed in glacial acetic acid to eliminate water with which acetic anhydride violently reacts, and a 9:1 mixture of acetic anhydride and concentrated sulphuric acid added so the test tube is about 2/3 full. The test tubes are then placed in a hot water bath for about 5 min, followed by washes with glacial acetic acid then distilled water.

At this stage most of the unwanted sediment should have been eliminated and this will be evident by the remaining sample occupying a reduced volume in the test tube. If resistant organic matter still remains, this can be removed by oxidation using 3 % bleach, adding a small amount then diluting this with distilled water according to the speed of the reaction. The success of this procedure can be assessed by visual examination of a smear on a microscope slide. A final wash with distilled water will have the sample ready for examination or storage.

Dehydration of the concentrated sample is sequential, first adding 50 % ethanol along with two drops of a stain such as safranin (this stain is optional) to render the pollen or spore features more visible under the microscope. The ethanol is decanted and the sample further dehydrated using 75 % then 95 % ethanol, each time adding about 10 ml of solution. Finally, Tertiary Butyl Alcohol is added, the sample centrifuged and the supernatant decanted. Samples, while in this alcohol, are transferred to labelled storage vials using pipettes. Silicon oil is added to the vials to preserve the pollen concentration for later examination. This is a good medium for microscope work as it allows the pollen to be moved and turned over for easier identification. The vials should be left uncapped at 47 °C to evaporate the alcohol to leave the pollen preserved in silicon oil, after which they can be capped and stored at room temperature.

3.3 Microscopic Identification and Counting

Slides of the pollen sample are prepared from the concentration stored in silicon oil. It is best to keep the density of concentrated sample low on the slide by spreading it out using a coverslip before affixing this. The cover slip can be sealed by use of nail varnish around the edges.

For pollen identification and counting, a multiple tally counter is useful. Otherwise, a simple table can be used. Counting is undertaken by moving the slide progressively along exclusive transects at low magnification (×100), increasing to high magnification (×600) when pollen is found.

During microscope inspection, deteriorated pollen may be encountered. This is an issue because the record may not be usable or representative of the source flora owing to differential preservation . Types of deterioration of pollen have been reported by Delcourt and Delcourt (1980) where pollen grains are damaged (Berglund and Ralska-Jasiewiczowa 1986), appearing etched, pitted, degraded or crumpled perhaps by oxidation or physical damage. Pollen with relatively thick exines can be resilient to corrosion compared with thinner and more fragile palynomorphs such as the Poaceae and some fern spores . Types and degrees of corrosion can however be used to interpret periods of disturbance in the sedimentary record , such as reworking of sediment (Wilmshurst and McGlone 2005). If the depositional environment has remained saturated and not exposed to wet-dry cycles then pollen preservation should be good, particularly for more saline sediment (Campbell and Campbell 1994).

3.4 Pollen Density in Sediment

Pollen concentration data can provide additional information on the frequency of the occurrence of fossil pollen taxa, and provide a quantitative comparison of pollen influx, if the sedimentation rate is known. Absolute concentration of the pollen is described in grains cm−3 of sediment. This can be determined by adding a known number of exotic pollen to a measured volume of the sediment, prior to commencing pollen concentration (Benninghoff 1962). While it is standard procedure to add the exotic pollen at the start of the pollen concentration procedure, on the assumption that there is no selective loss of fossil and exotic pollen during preparation, Mertens et al. (2009) recommended adding Lycopodium tablets at the end of processing, finding that more Lycopodium spores relative to dinoflagellate cysts were lost during concentration. This was because concentrations were considered to be affected during the warm HF treatment, and there may also be losses if a sieving stage is included (Mertens et al. 2012).

Absolute pollen concentrations can be determined by purely volumetric sampling methods (Grindrod 1985) but problems can be encountered with samples containing coarse sediment with losses from the sample during laboratory preparation. Use of exotic pollen grains compensates for these problems as the exotic pollen are presumed to be lost at an equivalent rate to the fossil pollen , so remain proportionally representative (Benninghoff 1962). Concentration is determined by the following equation:

$$ \mathrm{Concentration}=\frac{\mathrm{Total}\ \mathrm{number}\ \mathrm{of}\ \mathrm{exotics}\ \mathrm{added}}{\mathrm{Number}\ \mathrm{of}\ \mathrm{exotics}\ \mathrm{counted}}\times \frac{\mathrm{Number}\ \mathrm{of}\ \mathrm{fossil}\ \mathrm{pollen}\ \mathrm{counted}}{\mathrm{Volume}\ \mathrm{of}\ \mathrm{the}\ \mathrm{subsample}} $$

It can be expected in estuarine sediment that:

  1. 1.

    Pollen densities in the central basin of an estuary should be low and assemblages indicative of the catchment vegetation;

  2. 2.

    At intertidal locations, such as salt marshes and mangroves , pollen densities should be high and dominated by the local vegetation with some catchment pollen also present.

4 Applications of Pollen Analysis to Estuarine Palaeo-environmental Reconstruction

Pollen analysis of cores reveals evidence of vegetation changes over time which, in the cases of estuarine settings, gives insight into geomorphic evolution and relative sea level change. In the South Alligator River system of the Northern Territory, Australia , the pollen records show that extensive mangroves were present between 7000 and 5500 years BP at the time of sea-level stabilisation, and before fluvial sedimentation at these sites replaced the mangroves (Woodroffe et al. 1985; Grindrod 1988; Woodroffe et al. 1989). Sea-level conditions were slowly transgressive earlier in the record, with an increasing influence of marine processes, and corresponding changes in mangrove species present, from upper mangrove to sea-fringing species. The pattern then reversed, with marine tolerant species being replaced by more terrestrial species as the wetlands were infilled.

A similar evolutionary sequence is present in the nearby Mary River (Woodroffe and Mulrennan 1993) where a large mangrove area developed with sea-level stabilisation around 6500 years BP, and remained in place until 4000 years BP. During this time the rate of sedimentation was up to 10 mm a−1, declining after 6000 years BP. Sediment infill caused the mangrove forest to be replaced by freshwater wetlands , under which the rate of sedimentation was no more than 0.5 mm a−1 (Woodroffe and Mulrennan 1993).

By contrast, gradual coastal retreat of mangrove zones, with slowly rising sea-level, was demonstrated from the extensive, anastomosed estuaries of southern New Guinea (Ellison 2005, 2008). Pollen diagrams of stratigraphy up to 6 m depth showed landward Bruguiera zones at lower levels for most of the Holocene, replaced at around 3000 years ago by seaward Rhizophora zones. This indicated landward retreat of the mangroves , during relative sea-level rise.

The following estuarine case studies further demonstrate how pollen analysis of cores can be used to interpret Holocene estuarine evolution and sea level changes.

4.1 Case Study: American Samoa

Estuarine mangroves occur on islands of sufficient elevation to develop a river system and so deliver significant quantities of sediment to the coastal zone (Gehrke et al. 2011). These areas are the most extensive mangroves of the Pacific islands, for example, the Rewa delta , Viti Levu, the Dumbéa delta , New Caledonia, and the estuaries of Palau and Ponape (Ellison 2009a). The estuarine areas receive terrigenous sediment from the island and they also accumulate vegetative debris to form a mud.

These environments are demonstrated by the mangroves of Tutuila, American Samoa (Fig. 18.4). Tutuila is 32 km in length, 2–9 km wide, with steep and rugged terrain descending from a largely continuous central ridge of over 300 m, with a maximum elevation of 653 m. Samoa marks the eastern boundary of mangroves in the Pacific. Two common species exist, Rhizophora samoensis, which forms the seaward zone just above mean sea-level, and Bruguiera gymnorrhiza, which forms the landward zone that approaches the high tide mark (Fig. 18.2). The total area of such habitats in American Samoa is relatively small, but they consistently occur at the mouth of each river in small estuaries. Tides are microtidal, with a mean range of 0.77 m and spring range of 0.83 m (NOAA 2012).

Fig. 18.4
figure 4

Coastal stratigraphy of Tutuila, American Samoa , showing locations of the Masefau and Alao cores on which pollen analysis was carried out

Full size image

The estuaries all have a stratigraphy of shallow mangrove mud, a mixture of peat and basaltic or calcareous sand . With only 10–30 % organic matter they are not defined as peat . The inorganic content results from the steep gradients of this relatively young volcanic island, with resultant high rates of slope-inwash. Pollen analysis was carried out on the cores retrieved from the estuary of Masefau and an inland wetland at Alao that resembles an enclosed estuary (Fig. 18.4).

The Masefau mangrove estuary has an area of about 6.38 ha (Gilman et al. 2007), and is the second largest mangrove area of American Samoa , with an approximate centre at 14° 14′ 36″ S, 170° 43′48″ W. The area of the catchment contributing to the estuary is approximately 362 ha. The estuary has sand bars constricting the outlet to the ocean, and the estuary basin is, therefore, river dominated. The pollen diagram from the 150 cm core at Masefau is shown in Fig. 18.5. The forest at this swamp presently is dominated by Bruguiera with an understorey of Acrostichum ferns. At the sediment surface was a shallow inorganic layer that contained no pollen and spores , due to soil erosion infill following a cyclone. From 10 to 150 cm depth, deposited since 410 ± 60 years BP, assemblages are dominated by 40–70 % Bruguiera pollen , with 20–40 % Acrostichum spores .

Fig. 18.5
figure 5

Pollen diagram from the Masefau estuarine area, American Samoa

Full size image

This record shows long term stability with consistent domination by Bruguiera, and an ecosystem becoming increasingly organic-rich over time as it became established and more productive. The consistent presence of Acrostichum indicates disturbance by storms and/or human usage of the mangrove wood resources. American Samoa has a mean spring tidal range of 0.83 m (NOAA 2012), and the occurrence of mangrove-dominated facies down to 1.5 m, therefore, indicates relative sea-level rise and valley drowning.

At Alao (Fig. 18.4) a wet Hibiscus forest occurs at 1.8 m above present sea-level, cut off from the ocean by a 3.4 m high sand ridge. The surface of the core site was surveyed revealing it to be 1.7 m above sea-level. The wetland was cored in two locations, each showing a buried mangrove peat . In the core shown in Fig. 18.6 the base was hard rock, and dense mangrove peat occurred between 2.8 and 1.6 m depth, which was dated between 3600 ± 110 and 2330 ± 60 years BP. Above this exists 1.5 m of clay. The base of the peat is at 2.7–2.6 m depth or 1.0–0.9 m below present sea-level, and is dated at 3600 ± 110 (Table 18.2). Pollen analysis showed domination in lower levels by Cyperaceae pollen which, at 240 cm depth, declined to 60 %, after which mangroves appeared, revealed by 10 % Bruguiera pollen and 5 % Rhizophora pollen (Fig. 18.6). Bruguiera pollen occurs at 20–30 % from 2.2 to 1.9 m depth, then it declined to less than 10 % from 1.7 m to the top of the peat at 1.5 m, 0.2 m above present sea-level. Rhizophora increased from 40 %, at 2.2–1.9 m depth, to 50–80 % from 1.7 to 1.5 m depth. Herbs are represented as minor occurrences, with occasional evidence of Plantago lanceolata and Wollastonia biflora. Pollen of inland trees were mainly of Cocos nucifera but this was represented at low and occasional levels, with isolated occurrences of Ficus sp. and Pandanus sp. The organic content in the mangrove unit of the Alao core (Fig. 18.6) was far higher than that found at Masefau (Fig. 18.5), and the sedimentation rate far lower (Table 18.2).

Fig. 18.6
figure 6

Pollen diagram from Alao, American Samoa

Full size image
Table 18.2 Radiocarbon dates from mangrove stratigraphy, American Samoa
Full size table

The Alao pollen diagram (Fig. 18.6) spans 1200 years of vegetation history , over a time when the Pacific Islands were being colonised by people who likely arrived in Samoa 3000–2800 years BP (White and Allen 1980; Kirch 1997). At lower levels is evidence for an early sedge/grass wetland which infilled and became colonised by mangroves, first by the landward species Bruguiera and then the seaward species Rhizophora . This sequence is in contrast to that typically observed under coastal infill , which typically progresses from Rhizophora (lower elevation) through Bruguiera (slightly higher) to a freshwater swamp (highest) as the coastline progrades when sea-level is stable. Therefore, this sequence is indicative of a rising relative sea-level similar to what was shown at Masefau (Fig. 18.5). There are few or no weeds typically associated with people. Plantago and Acrostichum, which are associated with human disturbance , only increase towards the top of the profile. At this point, about 2400 years ago, the wetland is suddenly replaced by clay infill , probably related to catchment disturbance . This is very similar to the sudden infill of coastal freshwater lakes and wetlands of Mangaia in the Cook Islands with the arrival of people and commencement of catchment disturbance (Ellison 1994; Kirch and Ellison 1994).

4.2 Case Study: Cameroon Estuary, Central Africa

The Cameroon estuarine area is one of three significant mangrove areas in Cameroon, located south of the horst of Mount Cameroon (4095 m) in the central section of the coast where the Bimbia, Mungo, Wouri, Dibamba and Sanga Rivers all discharge (Fig. 18.7). The Sanaga River is the largest in Cameroon originating in the Adamawa plateau, with a length of 918 km and catchment area of 133,000 km2 (Van Campo and Bengo 2004). Freshwater discharge averages 2072 m3 s−1, with a mean of 473 m3 s−1 in the drier season in March, and reaching over 5000 m3 s−1 during the wet season from August to November, when flow is very powerful (Morin 1980; Giresse et al. 2005).

Fig. 18.7
figure 7

Map of the Cameroon estuary mangrove area showing vegetation zone distributions and location of the coring site

Full size image

Cameroon has a total mangrove area of 1957 km2, and six true native mangrove species exist. The dominant species is Rhizophora racemosa, accounting for over 90 % of mangrove forest including those in seaward zones. The other species present are R. mangle, R. harrisonii, Avicennia germinans and, towards land, Laguncularia racemosa and Conocarpus erectus are found (Corcoran et al. 2007; Ellison and Zouh 2012). All of these species are of the Western mangrove species group, found in America and West Africa (Tomlinson 1994), in contrast to the Eastern group represented in Samoa (Table 18.1). Undergrowth in upper zones can, however, also include the pan tropical Acrostichum aureum (Pteridaceae) where the canopy is disturbed. Nypa fruticans was introduced in 1910 from southeast Asia, and is today occasional. Mangrove associate species occurring on landward margins include Annona glaba, Cocos nucifera, Guiborutia demensei, Alchornea cordifolia, Dalbergia ecastaphyllum and Drepanocarpus lunatus, Athocleista vogeli, Pandanus candelabrum, Hibiscus tilaceus, and the grasses Bambusa vulgaus, Paspalum vaginatum and Sesuvium portulacastrum (Ajonina 2008).

A core was collected from the centre of the southern Cameroon estuary mangrove area (Fig. 18.7) and recovered from the centre of the mangrove area at Nkamba (03° 44′ 46.7″ S; 009° 40′ 41.42 E), 20 m inside the mangrove forest from a side creek off the river bank. The forest height of Rhizophora racemosa was 30–35 m, and the canopy was open. The stratigraphy comprises organic, very dark grey, silty clay (5Y 3/1) from the surface to 0.70 m, with a less organic lens at 0.55 m (Fig. 18.8). Below this, from 0.70 to 2.20 m, the sediment was a dark grey silty clay (5Y 4/1), grading to firm, very dark grey silty clay (5Y 3/1) until 3.80 m depth. At 3.80 m the first appearance of fine sand occurred, which was of grey colour (5Y 3/1). The core ended at 4.60 m in fine sand , with calcareous fragments, that was too stiff to penetrate further. There was no macro-fossil evidence in either core of mangrove presence below the organic silty clay upper layer, with the stratigraphy below this shown to be very inorganic . Percent organic matter was determined by ignition at 550 °C for 2 h (Bengtsson and Enell 1986), and results showed a consistent decline with depth, from 20 to 25 % near the surface to c. 18 % below 1.5 m, though with some outliers (Fig. 18.8).

Fig. 18.8
figure 8

Nkamba Core stratigraphy and pollen diagram, showing pollen percentage of most frequent taxa, sums of ecological groups, pollen concentration, and percent organic matter trend with depth

Full size image

Accelerator Mass Spectrometry radiocarbon dating results are shown in Table 18.3. The pollen diagram (Fig. 18.8) shows the relative representation of each mangrove taxon recorded in samples as a percentage of the total pollen sum, which includes mangrove taxa, non-mangrove trees, shrubs, herbs, aquatics and ferns, as well as mangrove percentage of this pollen sum.

Table 18.3 AMS and calibrated radiocarbon dating ages from a central core from the Douala estuary intertidal area
Full size table

Results show that mangroves were dominant, at about 60 % of the total pollen sum, between 0 and 60 cm, before falling below this level to 20–40 % of the total pollen sum (Fig. 18.8). Rhizophora was present, at 25–35 % of the pollen sum, from 0 to 30 cm in depth, below this falling to 15–30 % of the pollen sum. The landward species Laguncularia, along with Avicennia, were also present at 5–15 % of the pollen sum above 1.0 m, below this falling to lower levels. Conocarpus erectus was mostly present at all depths, but in low proportions. Nypa fruticans is present at very low levels in upper samples. This introduced species normally occurs in landward zone mangroves (Tomlinson 1994). Cocos nucifera was mostly present at all depths, at 10–15 % of the pollen sum, while the landward forest genus, Pandanus, was present in significant proportions through deeper sections of the core below 80 cm depth. Pollen concentrations fluctuated between 750 and 1500 grains cm−3 (Fig. 18.8).

Results demonstrate a stratigraphy , in the Cameroon estuarine mangrove area, of lower levels of inorganic sand at 3.5 m below the mangrove surface (Fig. 18.8), fining upwards to inorganic silty clay to 0.8 m below the mangrove surface, with shallow levels of more organic silty clay above this from 0.7 m below the surface. Organic matter levels are slightly higher in this surface unit, but are overall very low, indicative of significant allochthonous inorganic sediment input. Sand occurs at 3.5 m (Fig. 18.8), indicating earlier, higher energy conditions being reduced as the embayment infilled and the western sand ridge became more protective following its growth by longshore drift. The sand -to-silt upwards fining sequence demonstrates an environment of decreasing wave and/or current energy, followed more recently by colonisation by mangroves as indicated by the upper organic content to the silty clay.

Pollen concentrations are also very low at c. 1000 grains cm−3 compared with other study sites also having the Atlantic mangrove species group (Behling et al. 2001, 2004), which is further indicative of inorganic sediment dilution of pollen deposition . There are inorganic flood lenses apparent, particularly at 55 cm.

The pollen diagram (Fig. 18.8) shows higher proportions of mangrove pollen near the surface, but never reaching much more than 60 %. The modern analogue is shown by the surface sample, which although under well-established tall Rhizophora stands, only had 60 % mangrove pollen , which indicate significant riverine pollen in-wash. Proportions of 70–80 % mangrove pollen have been found to indicate a mangrove environment, declining to 50 % or less adjacent to the mangrove ecosystem such as in mudbanks off-shore (Muller 1959; Bartlett and Barghoorn 1973; Grindrod 1985, 1988; Ellison 1989; Behling et al. 2001; Ellison 2005, 2008; Ellison and Strickland 2015). The upper 60 cm of the core has mangrove pollen proportions of 60 % and represents a mangrove environment, declining below this to around 40 %, which represents mud flats off-shore from the mangroves, which become more coarse grained , inorganic and calcareous with depth.

In the upper mangrove section, the long-term sedimentation rate calculated above the 720 ± 40 years BP date, at 1950 mm depth, is 2.7 mm a−1, and the lower radiocarbon dates indicate rapid sedimentation and mixing such as in off-shore bay conditions (Table 18.3). In estuaries of southern Indonesia, Ellison (2005) showed net mangrove sedimentation rates of 0.6–1.5 mm a−1, hence rates in the Cameroon estuary are significantly higher. Pollen concentrations in both Cameroon cores were low, at 500–1500 grains cm−3, but tending to be higher in the landward core relative to the seaward core. This is to be expected from the greater local dominance of pollen sources. Pollen concentrations in southern Indonesian mangrove cores showed similar trends, though with much higher concentrations (Ellison 2005).

There is no evidence of mangrove stratigraphy lower than present mangrove elevations (Fig. 18.8) such as found by Bartlett and Barghoorn (1973), Grindrod (1985, 1988) and Ellison (2005), a result replicated at the seaward edge of estuarine mangroves (Ellison and Zouh 2012). Subsiding coasts, such as in southern New Guinea, show mangrove pollen (>80 %) are consistently dominant up to 6 m depth in a location with a <3 m tidal range (Ellison 2005). This comparison indicates that the coastline of the Cameroon estuary is prograding seawards with stable sea level before the recent increases in global sea levels of the last few centuries. In conditions of sedimentation surplus, mangroves colonise seaward either into bays, especially off-shore of river mouths, or over reef flats, but only where these accrete to reach the upper intertidal range. In conditions where relative sea-level rise exceeds sediment accretion rates, mangroves die back from the seaward edge and retreat landward, again following the upper intertidal range (Ellison 2008). This is demonstrated from the extensive estuarine swamps of southwest Papua (Ellison 2005, 2009b).

Results of this core therefore, suggest that relative sea-level has been stable on the central Cameroon coastline for the last several centuries, and both fairly rapid inorganic , river-dominated sedimentation rates and seaward progradation is occurring. This is shown by significant levels of mangrove pollen in the stratigraphy , only at shallow depths where mangroves currently occur, and is supported by low pollen concentrations and the low organic fraction throughout. Recent sea level rise has, however, caused some retreat of the seaward edge of mangroves (Ellison and Zouh 2012).

Relative sea-level trends at coastal sites are an important component of a vulnerability assessment with respect to climate change, but can be difficult to quantify at remote sites (Ellison 2015). Where the coastline lacks a long term tide gauge, relative sea-level trends can be reconstructed from pollen analysis of mangrove stratigraphy . Stratigraphy and pollen analysis results from the Cameroon estuary mangrove area indicate a prograding sedimentary sequence with stable sea-levels over the last few hundred years. While sea-level has been stable, sand and silt has accumulated in the embayment to reach tidal levels, allowing colonization by mangroves.

5 Conclusions

Pollen analysis of estuarine sediment cores, with selected radiometric dates and pollen laboratory work to reconstruct past vegetation assemblages, with comparison to modern analogues of surface pollen distributions, is a relatively inexpensive technique for estuarine reconstruction . It does, however, require specialist laboratory skills in the use of pollen concentration and procedures of pollen and spore identification. Its limitations relate to the successful discovery in estuaries of pollen rich stratigraphy , which, most commonly have been found to feature intertidal vegetation deposits that tend to give evidence of relative sea level change.

Pollen records are best where there is an even time/depth sequence of stratigraphy with limited reworking, and where the pollen concentration is relatively high. Central areas of wave-dominated, and tide -dominated, estuaries are both characterised by lower energy zones infilled by organic muds that can be occupied by extensive intertidal areas of saltmarsh or mangroves. These conditions have provided good estuarine sites that are suitable for palaeo-environmental reconstruction as shown by the Northern Territory Alligator River system (Woodroffe et al. 1985; Grindrod 1988) and southern New Guinea (Ellison 2005).

In terms of geomorphic setting, some estuarine types are better suited to pollen analysis techniques than others. Lower levels of the Cameroon estuary core (Fig. 18.8) are an example of a poor pollen depositional environment , with high inorganic sand content and low pollen concentrations. As indicated by the lower radiocarbon dates from 5 to 2 m depth (Table 18.3) there is a high level of reworking. This is similar to early stages of fluvial infill of a bay as described by Cooper (1993, 1994) from South African estuaries. River-dominated estuaries are less likely to have a fine resolution sea-level record in pollen sequences owing to reduced marine influences relative to wave-dominated, and tide -dominated, estuaries.