Holocene pollen records from Caution Bay,southern mainland Papua New Guinea

Abstract

This study presents new palynological data from Caution Bay, south-central Papua New Guinea (PNG). It explores Holocene mangrove transitional events along a tide-dominated shoreline, and expands reconstructive detail for the wider southern PNG lowlands. Coinciding with recent regional archaeological research, this study also holds implications for interpretations of the Holocene settlement of Caution Bay and long-term people–landscape interactions. Data demonstrate a late-Holocene mangrove to mudflat transition, with vegetation patterns largely a function of low sediment loading, sedimentary redistributions and salinisation upon sea level decline c. 2000 cal. yr BP. These trends appear unique to Caution Bay, highlighting a variety of Holocene shoreline vegetation changes along the PNG south coast, each dependent on geomorphologic setting and hydrological fluxes. Further work is required to elucidate vegetation change inland of the coastal zone. Greater understanding of burning patterns and an archaeological collaboration are required to determine more concisely dryland plant spatial and temporal variability.

Keywords

Holocene Lapita archaeology mangrove palynology Papua New Guinea tropical coastal lowland

Introduction

The term ‘mangrove’ refers to plant species demonstrating physiological and morphological adaptation to the intertidal environment; to conditions of high-salinity, low oxygen, low nutrient availability and substrate mobility (Shearman, 2010). ‘Mangrove’ is used in equal reference to the plant community as it is to individual plant species. Global mangrove diversity is at its highest across the Indo-Pacific region, and the ecosystems of the Gulf of Papua (the Gulf) in southern Papua New Guinea (PNG) are some of the largest and most pristine (Ellison, 2005; Shearman, 2010). Across the Gulf, mangrove communities occupy a number of different environmental settings. These settings vary in their geomorphological context and in the physical processes responsible for sediment transport and deposition. Accordingly it is difficult to extrapolate from one setting to another to understand a particular locality’s historical trajectory.

The Gulf shoreline incorporates a number of rivers draining the mountains of central PNG. These rivers, with high sediment loads and/or seasonally variable outflow, form complex deltas. Here, mangroves distribute in response to microtopographic characteristics including distributary channels, point bars, river levees and interdistributary basins. Changes in mangrove patterning occur as a result of changes in geomorphological processes operating within the deltalic setting itself, mangroves being opportunistic in their habitat colonisation (Krauss et al., 2008; Woodroffe, 1990). The bulk of late-Quaternary mangrove research in PNG has been carried out within river-dominated settings, documenting rates of siltation (deltaic plain vertical accretion) and shoreline progradation (Chappell, 2005; Ellison, 2005) or the upriver extent of mangrove establishment (spatial positioning of upper and lower deltaic plains) (Chappell, 2005; Thom and Wright, 1983). Comparatively, little work has been conducted on tide-dominated settings, where the behavioural arrangement of mangroves and associated ecosystems is expected to differ. As the name suggests, tidal action (tidal range and directional current) is responsible for sediment distributions and environmental character. More extensive areas of low-gradient shoreline are available for mangroves, and plant successional patterns are more likely. The relationship between land and sea is also more open and direct (uninsulated). Complex fluvially formed mangrove habitats are absent, replaced by colluvial and alluvial deposits, saline-flats and chenier ridges (Krauss et al., 2008; Woodroffe, 1990). More generally, a lack of descriptive information from tide-dominated coastlines occurs on a background of little available palaeoecological data on the southern and coastal lowlands of PNG.Currently palaeoecological and palynological records on the whole are concentrated in the highlands of PNG (see reviews by Haberle, 2007 and Hope, 2007).

The present study area encompasses Caution Bay, a mangrove-fringed and inland savanna-dominated embayment on the southern mainland PNG coast. The purposes of this paper are (1) to provide palynological evidence describing mangrove Holocene history and littoral zone successional events along a tide-dominated shoreline east of the Gulf of Papua, and (2) to extend Caution Bay Holocene description inland thereby expanding reconstructive detail for the southern PNG lowlands more widely. Further, Caution Bay has been the focus of recent intensive archaeological research (e.g. David et al., 2011, 2012; McNiven et al., 2011, 2012). Therefore, the purpose of this paper is also (3) to comment where possible on the palaeoecology of Caution Bay for the timespan of the known archaeological record and its implications for people–environment relations.

Caution Bay

Caution Bay centres 20 km west of Port Moresby, positioned along the south coast, east of the Gulf of Papua within the Central Province of PNG (Figure 1). Climate encompassing Caution Bay is Tropical Savanna (a classification with limited range across PNG, see Gillison (1983) for mapping). Conditions strongly reflect a seasonal alternation in wind direction, temperature, humidity and, most significantly, rainfall which are broadly controlled by two systems. A northwest monsoonal system (known locally as lahara) occurs between December and April, associated with frequent-significant rainfall and warm, high humidity conditions. Southeasterly trade winds (laurabada) prevail between June and October during which drier-cooler conditions occur. Brief transitional phases occur between these two predominant wind and rainfall patterns. Annual rainfall averages 1000 mm and the seasonal temperature range is 28–32°C (mean maximum) (Fitzpatrick, 1965; McAlpine et al., 1983).

Figure 1.

Map of southern Papua New Guinea showing location of Caution Bay and places mentioned in text.

The south-central coast of PNG is an embayed reef-fringed coast. A number of environmental zones run parallel (in partly discontinuous form) to the shoreline. In succession inland a littoral plains complex, alluvial plain, coastal lowland and coastal hill-ridge formation comprise the Caution Bay catchment (and corresponding to the Papa, Boroko and Fairfax land-systems mapped by Mabbutt et al., 1965). Each environmental zone has a distinctive geology, pattern of topography, soils and vegetation. In this paper descriptions concentrate on the littoral plain.

Along Caution Bay the littoral plains zone is up to 1 km wide (Figure 2). Coastal features occur in the form of plains, spits and barrier-beaches and the main occurrence of tidal flats is behind these protective formations. The tidal flats (outer tidal, lower inner and higher inner flats) are differentiated on the basis of inundation depth and frequency. Sediments show a transition from the outer shore where wave and current action remove fine materials leaving sandy deposits, through a range of intermediate textures to peaty-clays in highest areas of shallow inundation and above. The area is drained by small creeks (many with shallow, shifting, transient, occasionally irregular channels). Here, estuarine patches combine inland watercourse and tidal influence (freshwater to brackish) (author observation; Löffler, 1985; Mabbutt, 1965). In Mabbutt et al.’s (1965) outline of the Caution Bay littoral zone, the outer tidal flats are described as occurring on an accreting coast, with the inner tidal flats (between current and inner beach ridges) slightly below to slightly above mean high water mark.

Figure 2.

Diagram of the littoral plains zone at Caution Bay showing landform, vegetation and core site position.

Landward the littoral plain is vegetated with mixed scrub and evergreen-deciduous thicket (frequently disturbed, author observation), grading to grassland. Shrubs and low trees (notably Terminalia, Clerodendrum, Scaevola, Hibiscus) combine with scattered herbs and lianes (Flagellaria and unidentified climbing ferns). Initially the grasslands are dense, composed of Imperata, Themeda and herbs (noting here greater sedge or broadleaf herb growth and shrub recruitment in localised moist Pandanus depressions). With distance inland the groundcover thins and becomes drier and lower in stature. In the landward to seaward transition bare saline mudflat occurs with low Tecticornia (chenopod salt marsh) and Sesuvium tussock-like vegetation, concentrated on occasional slight rises within or at the fringes of the mudflat (author observation; Heyligers, 1965; Paijmans, 1976). From field observation this patchy mudflat may be seasonally flooded by fresh water, while covered by seawater at extreme high tide.

At Caution Bay two mangrove communities predominate. A tall, dense canopied, one-layered forest of almost pure Rhizophora occurs seaward and at lower elevations subject to frequent tidal inundation. Secondary to Rhizophora is Bruguiera. On appearance, the Rhizophora forest is even-aged. Avicennia then dominates inland of Rhizophora, occupying a higher tidal position and forming an irregular to low-height open canopy woodland. Ceriops and Bruguiera were sparsely observed and tentative identifications of Aegiceras and Lumnitzera were made. The woodland incorporated no groundcover except for numerous Avicennia pneumatophores, marked also with the common occurrence of dead trees and fallen wood. The Avicennia woodland to open mudflat transition was largely abrupt (author observation).

Methods

Field coring and sample preparation

Three sediment cores were collected along a southwest–northeast transect spanning the outer to higher inner tidal flats and vegetation zones (Figure 2; with high and consistent pollen preservation in cores 1 and 3, as discussed in this paper). Having a series of cores reduces the influence of local land effects and maximises the recording of influential non-local factors such as sea level and climate (see Ellison, 2008). The transect orientation is complementary to a second collection sequence carried out by the first author, transgressing the coastal lowland. Core 3 is also c. 190 m from the archaeological site Bogi 1 (McNiven et al., 2011) located on the landward dune above the mudflat (Figure 2). Sampling was conducted by hand and at low tide. A sidewall D-Section corer was used, facilitating penetration through mangrove roots and sands while also minimising sediment compaction. Each core section was described in the field before being securely wrapped and prepared for transport. Core chronology is provided through an AMS ¹⁴C analysis of bulk organic samples secured at regular intervals from the sediment core centres prior to pollen sample collection. Dating samples were submitted to the Waikato Radiocarbon Dating Laboratory and resulting AMS determinations calibrated using the Fairbanks et al. (2005) and OxCal 4.1 (Ramsey, 2009) programs.

Laboratory technique

Pollen sample preparation followed standard technique as detailed in Bennett and Willis (2001) and as experimented with by Rowe (2007) for mangrove sediments elsewhere. Chemical preparations were selected to initially disperse then progressively remove humic acids, calcium carbonates, bulk organics and cellulose, silicates, and to ensure the pollen wall ornamentation was suitably visible. Particular attention was paid to the treatment of carbonates, given much of the coastal lowland to ridge systems are underlain by limestone. Silicate removal required equal diligence owing to the strong silicification observed in some subsurface lowland materials (i.e. quartz-filled pores or rock seams). Jorry et al. (2008) also describe the Gulf of Papua as a tropical mixed siliciclastic/carbonate system. Lycopodium spike additions served relative concentration calculations of both pollen and charcoal. Sediment subsamples were 2 cm³ in size.

Pollen identifications were assisted by the authors’ own floral field-specimen collection from the Caution Bay area, text reference material (e.g. Thanikaimoni, 1987), preserved reference collections held within Monash University’s School of Geography and Environmental Science, and online photographic sources such as the Australasian Pollen and Spore Atlas (AustralianPollen and Spore Atlas Members, 2007). Charcoal identification incorporated fragments of angular form, black and opaque. The charcoal count is defined as microscopic (10–125 µm), deemed representative of regional (catchment) fire activity (see principles outlined in Whitlock and Larson, 2001). For the diagrammatic presentation of data the Tilia suite of programs was used (Grimm, 1991). Pollen counts are expressed as a total pollen percentage, a sum value of 300 grains minimum.

Mangrove palynology: Interpretative considerations

Factors impacting pollen assemblage development and composition include plant pollen production, dispersal strategy, as well as the deposition site itself. Pollen analysis is classically applied to stratified waterlogged deposits. Grains fall on a water surface, sink and incorporate into a steadily building deposit. Waterbody study sites such as lakes or swamps are typically regarded as closed systems (Chappell and Grindrod, 1984; Pearsall, 2000). Different interpretative assumptions are apparent through the mangrove literature.

Palaeo-ecological intertidal studies classify such environments as open systems (occupants of dynamic settings). While distinct floristics, neat species segregation, and a precise ecological niche align mangroves to palynological study, as open systems the capability of sediment to retain pollen varies widely. Sediments may be lost or redistributed (Chappell and Grindrod, 1984; Grindrod, 1988). Mangrove soils are re-worked by tidal currents (Furukawa et al., 1997) and are subject to biogenic mixing. High tide muds are subaerially exposed for long periods and may suffer floodwater scouring and aerial erosion. Optimal conditions occur along muddy, low energy shores which support luxuriant plant communities containing high pollen producers, and where sediment accumulation is not excessive (Grindrod, 1985, 1988). Grindrod (1988) provides a good summary of differential pollen preservation and redistribution through north Australian case studies.

Many mangrove taxa are disproportionately represented via pollen. Considerable interpretative weight may be placed on the simple appearance of certain grains just as numerical guidelines are required for others in high proportion. Avicennia pollen is in general under-represented as distribution is closely related to source plants. Localised dispersal is a consequence of floral arrangement suited to bee pollination, small volume pollen production, and heavy ornate Avicennia grain structure (Duke, 1991; Grindrod, 1988). Similarly, Ceriops and Bruguiera are inefficient wind dispersers reliant on insect pollination. From all Bruguiera species and Ceriops tagal, pollen is ‘spring load’ ejected from the flower only when triggered by the body of a pollinator (Wightman, 2006). In contrast, Rhizophora species demonstrate abundant localised pollen distribution close to the parent source, with strong representation inshore to offshore (Grindrod, 1988). High pollen/ovule ratios, short pollen generation time, lack of nectar secretion and down-pointing flowers (unobstructive to dispersal) result in Rhizophora’s high production wind pollination (Tomlinson, 1986).

Mangrove community pollen assemblages skew toward an over-representation of wind dispersed-pollinated taxa. In response to Rhizophora behaviour, interpretative guidelines have been generated from modern sediment studies. Rowe (2006, 2012) established Rhizophora values up to 95% as indicative of lower-tidal mangrove sites, between 30% and 60% as representative of upper tidal and <30% at the landward extremity of mangroves (extending into adjacent ecosystems such as mudflat). A 90% pollen proportion of Rhizophora has been outlined by Ellison (2008) as a Rhizophora stand, with a 30% proportion as immediately adjacent to Rhizophora growth. Bartlett and Barghoorn (1973) state between 45% and 95% Rhizophora plus Avicennia as representative of a mangrove swamp and 10–45% Rhizophora pollen (declining further to <10%) representing sediments immediately landward.

Major pollen types delivered long-distance into intertidal sediments are also wind dispersed. The open woodland structure dominating the Caution Bay coastal lowlands provide conditions ideally suited to wind dispersed-pollinated plant types. These are expected to demonstrate a ubiquitous pollen signal. Conversely, taxa of the mud flat and dune ridges display predominantly localised pollen distributions. Excessive close-range input may swamp regional representation (Poaceae, for example, may derive from inland-upland sources but its significance can be masked by pollen from the dune and littoral plain). In all cases the closed-canopied nature of mangrove growth can unfavourably filter or obstruct longer-distance pollen. Offshore winds may also limit landward airborne pollen deposition. Favourably, offshore winds may help to localise transport of mangrove pollen types such as Rhizophora (Chappell and Grindrod, 1984; Grindrod, 1985, 1988).

Results

Stratigraphy

Maximum hand-action core depths were recovered. Core 1 reached a depth of 180 cm and core 3 148 cm. Pale, inorganic, gritty (shell and stone) consolidated clays were observed at the base of each core and could not be penetrated further. In core 1 from 180 to 105 cm pale grey clay (10YR 7/1 to 6/1) is present incorporating significant shell fragment and sand proportions (a gradual decrease in shell continues to 92 cm). No fibrous organics or root materials occur through this lower unit. Above 105 cm is organic (patchy fibrous and mud-like) clay, very dark brown in colour (10YR 2/1). Fine root material occurs throughout and complete molluscan shells are occasional. In core 3 dark coarse organic clay (7.5YR 3/1 to 2.5/1) is present from 148 to 14 cm. Unidentifiable organic remains and fine root material occur (the former contributing to some clay staining) but the proportions of clay, shell grit and sand decrease up-core, particularly above 80 cm. The upper 14 cm is mottled fine grey clay (2.5YR 4/1) with localised red-brown colour (7.5YR 3/4). This clay includes neither sand nor shell fragment and visible plant material is almost absent. A narrow interface blends the two core 3 units, spanning 9 to 14 cm.

Chronology

Twenty samples were submitted for AMS ¹⁴C analysis (Table 1). Each core demonstrates a number of dating inconsistencies. To counterbalance dating problems, numerous dates were obtained in a scheme aimed toward (at minimum) highlighting ‘populations’ of dates with depth, revealing broad time periods. Where inversions do occur, these discrepancies may still be considered within a reasonably similar age range or ‘group’ (see Björck and Wohlfarth, 2001, discussions). Therefore, while Table 1 reveals some mixture of AMS results, an overall chronological–stratigraphical integrity is demonstrated. That is, broad vertical core integrity exists with youngest dates found toward the surface and oldest with depth. Core 3 incorporates an upper 12 cm spanning the past c. 950 years. Below this age dates consistent with depth span c. 950–1740 cal. yr BP, followed by a lowest zone of mixing with an age up to c. 1860 cal. yr BP prior to a basal date approaching 2000 cal. yr BP (between 80 and 90 cm depth two dates are near-identical at 1732±38 and 1738±36 cal. yr BP). The upper 20 cm of Core 1 appears to be modern, below which a mixed zone centres across c. 450–650 cal. yr BP before a single date of 1714±34 cal. yr BP at 140 cm depth. This lower age is anomalous, although noted to correspond to a majority of results from Core 3, simultaneously recorded within a horizon of comparable sedimentary character.

Table 1.

Caution Bay core sample ¹⁴C AMS results (Ramsey (2009) and Fairbanks et al. (2005) for calibration).

Core	Depth (cm)	Sample ID	δ¹³C‰ (±0.2)	¹⁴C age, yr BP	Calibrated age, cal. yr BP (95.4% probability)	Calibrated age, cal. yr BP
1	20	Wk-31177	−26.7	101.4±0.3%	Modern	Modern
1	60	Wk-31178	−26.9	681±24	655–559	653±23
1	100	Wk-31179	−26.5	379±23	486–322	454±48
1	130	Wk-32616	−25.4	572±25	645–530	568±35
1	140	Wk-31180	−21.3	1791±25	1714–1558	1714±34
1	150	Wk-32625	−24.2	675±75	680–560	648±27
1	160	Wk-32617	−26.4	432±25	525–460	498±13
1	170	Wk-32618	−26.1	520±25	625–505	530±13
1	180	Wk-31181	−27.1	296±25	443–156	349±50
3	12	Wk-31187	−25.9	1021±24	933–897	936±13
3	50	Wk-31188	−26.3	1363±24	1292–1179	1288±10
3	70	Wk-32619	−26.3	1526±25	1520–1345	1399±30
3	80	Wk-31189	−26.0	1806±27	1773–1557	1732±38
3	90	Wk-32620	−25.4	1811±25	1825–1635	1738±36
3	100	Wk-32621	−25.9	1437±25	1375–1295	1325±19
3	110	Wk-31190	−25.7	1714±27	1692–1420	1616±44
3	120	Wk-32622	−24.8	1912±25	1930–1815	1857±25
3	130	Wk-32623	−26.8	1817±25	1825–1635	1746±37
3	140	Wk-31191	−26.2	1681±25	1600–1414	1572±32
3	150	Wk-32624	−24.9	2005±25	2005–1885	1949±29

Problems associated with accurately dating periods of coastal sedimentation and vegetation change are numerous. These are discussed by several authors (Chappell and Grindrod, 1984; Ellison, 2008; Grindrod, 1988; Grindrod and Rhodes, 1984) and are broadly related to the open nature of the environment as discussed above. Mixing of dates from Caution Bay is seen largely as a result of:

Sediment mixing and intrusive carbon. Grindrod (1988) describes mangrove sediments as ‘living soil’. Burrowing fauna result in bioturbation; crab-built mounds for example were observed along the study transect near the high water limit, and it is probable that any disused burrows would subsequently fill with tidally distributed material. Intrusive carbon may also arise through young plant root penetration into older sedimentary layers (Furukawa et al., 1997; Grindrod, 1988; Kristensen et al., 2008).

Allochthonus carbon: Although most carbon is known to be autochthonus (i.e. formed in situ), material (especially finer particulates) is likely also introduced from elsewhere in mangrove environments or from inwash terrestrial and estuarine sources. Reworking of sediment through aggressive wave and tide action is also a factor (Furukawa et al., 1997; Grindrod and Rhodes, 1984; Kristensen et al., 2008).

Pollen and charcoal records

Core 1 (Figure 3; ≥650 (1715–650) ca lyr BP–present; 45 samples, seaward littoral 9.375°S, 147.003°E)

Mangrove pollen comprises 90% (minimum) of the total pollen sum throughout core 1. Diversity in mangrove taxa is not equally significant; Rhizophora values (dominated by genus type A) overwhelm the diagram and few other mangrove taxa are characteristic. Bruguiera/Ceriops followed by Avicennia and Sonneratia are minor, though are relatively constant in representation.

Figure 3.

Caution Bay core 1 pollen (percentage) and microcharcoal (concentration per cm³) assemblage plotted against depth and calibrated age. Percentages derived from total pollen sum inclusion. Exaggeration values ×10.

The core 1 pollen diagram has been divided into two zones. Divisionis based mainly on palynological richness. Zone 1 (120–180 cm; >1715–650 cal. yr BP) incorporates highest non-mangrove values, dominated by Myrtaceae, Fabaceae and Arecaceae as well as Poaceae and Cyperaceae. Rubiaceae, Euphorbiaceae and Chenpodiaceae are more sporadic. Terminalia is recorded solely in zone 1. Non-mangrove diversity is progressively lost through zone 2 toward the surface; Pteridophyta disappear from the record mid-core and the subzone 2b is defined on the near-absence of herbaceous pollen and diminished secondary mangrove taxa. Charcoal concentrations decline with depth. Greatest, albeit also the most variable, charcoal values occur across the subzone 2a–2b boundary (28–76 cm, c. 550 cal. yr BP to present). A prominent charcoal spike is also recorded at the zone 1–2 boundary.

Core 3 (Figure 4; c 2000 cal. yr BP–present; 38 samples, landward littoral 9.372°S, 147.006°E)

Figure 4.

Caution Bay core 3 pollen (percentage) and microcharcoal (concentration per cm³) assemblage plotted against depth and calibrated age. Percentages derived from total pollen sum inclusion. Exaggeration values are ×10.

Core 3 incorporates a significantly more diverse pollen assemblage relative to core 1, and is the only record demonstrating the replacement of one Caution Bay vegetation community by another. The core 3 pollen record is divided into four zones, as follows:

Zone 1 (148–90 cm; c. 2000–1740 cal. yr BP). Throughout the zone mangrove representation is at or above 80% of the pollen sum with clear Rhizophora-A dominance. Other mangrove taxa (Rhizophora-B, Bruguiera/Ceriops and Avicennia) are secondary and record ongoing minor percentages (<10%). Lumnitzera and Excoecaria are occasional. Notably, Rhizophora-A values fluctuate to a greater extent than the secondary mangrove taxa (a trend apparent for the entire core). A diverse suite of non-mangrove pollen is represented (Poaceae aside, terrestrial taxa are intermittent). Herbaceous taxa cluster toward the lower half of zone 1 with Boraginaceae, Chenopodiaceae, Zornia, Flagellaria, Menispermaceae and Acanthaceae recorded at an earlier stage to Desmondium, Asteraceae and Apiaceae. Tree types, Myrtaceae, Pandanus, Ficus and the Ulmaceae genera occur somewhat more frequently than other taxa. Proteaceae, Stemonurus, Araliaceae, Kleinhovia and Bombax, Apiaceae and Crotalaria only occur in zone 1. Charcoal concentrations are variable but tend to decrease toward the top of the zone.

Zone 2 (90–50 cm; c. 1740–1300 cal. yr BP). Collectively, high mangrove values are maintained. Individually, increases in the representation of secondary mangrove taxa occur, notably toward the top of the zone. Rhizophora-A numbers continue to predominate. The non-mangrove component is again dominated by Myrtaceae, Pandanus, Ulmaceae, Poaceae and Desmodium, but with a shift toward increased Fabaceae and Nothofagus. Cyperaceae and Pteridophyta are periodically missing within zone 2 and Boraginaceae noticeably declines. Charcoal concentrations increase sharply from zone 1 and remain elevated (peaking at 80 cm depth).

Zone 3 (50–12 cm; c. 1300–1000 cal. yr BP). Mangroves continue to represent over 90% of the pollen sum. Rhizophora-A alone contributes to 80%, but begins to demonstrate fluctuating decline. Avicennia (and Rhizophora-B to a lesser extent) increase further on values seen in zones 2 and 1. Bruguiera/Ceriops remain constant and all other mangrove taxa are sporadic. Terrestrial tree and shrub taxa continue to be dominated by Myrtaceae and Fabaceae. Trema, Terminalia, Euphorbiaceae and Arecaceae are present in the lower half of the zone, prior to Araucariaceae and expansion in Pandanus and Casuarina. Barringtonia increases up core from zone 2. Poaceae and Desmodium percentages remain dominant, although the number of minor herbaceous taxa observed has declined. Cyperaceae values are constant. Flagellaria and Boragninaceae increase toward the top of the zone. Charcoal concentrations decrease from zone 2 but show a single sample rise at 16 cm.

Zone 4 (12–0 cm; c. 1000 cal. yr BP to present).The prominent feature of this zone is the marked decline in Rhizophora-A values and corresponding strongest representation of Avicennia. Rhizophora-A declines to 24% of the pollen sum as Avicennia increases to 32%. Excoecaria and Rhizophora-B also increase and Lumnitzera is introduced to the record. Bruguiera/Ceriops values are maintained; Sonneratia is lost. Total non-mangrove pollen sum contributions rise significantly to 38%. Myrtaceae, Pandanus, Arecaceae and Casuarina all show sustained increases and combine with new taxa Hibiscus, Aracariaceae, Podocarpus and Phyllocladus. Celtis and Barringtonia show an initial presence only to decline in the top samples. The record of Poaceae, Cyperaceae and Pteridophyta increases sharply and Chenopodiaceae is well represented. Interestingly, in the transition into zone 4 the higher total non-mangrove presence has not included a more diverse suite of taxa. Charcoal concentrations decrease to near-lowest.

Discussion

Late-Holocene mangrove succession at Caution Bay

The sediments collected at Caution Bay were deposited over a late-Holocene period spanning the past 2000 years. The consolidated nature of sediments below cores 1 and 3 is considered to pre-date site vegetation development. The change in site sedimentation, to fibrous dark muds, and the earliest timeframe c. 2000–1700 cal. yr BP mark the onset of mangrove growth. Mangroves are a group of highly reactive opportunists. They rapidly colonise newly deposited but stable intertidal sediments and in doing so help reinforce these sediments and promote further sedimentation (Blasco et al., 1996). Across the Greater Australian monsoonal tropics Rhizophora(ceae) forest is shown to have established directly upon postglacial saltwater inundation (e.g. Grindrod, 1988; Rowe, 2007; Woodroffe et al., 1985). This is also the case at Caution Bay; approaching 2000 cal. yr BP sediment accumulation initiating a lower tide permitted a well-formed Rhizophora mangrove forest to colonise and become entrenched across the area.This was to an extent greater than that seen today (with ‘big swamp’ spatial and character equivalency, see Rowe, 2007; Woodroffe et al., 1985). The genus Rhizophora requires frequently inundated habitat and demonstrates optimal and continuous development on well-drained perennially wet muddy substrate (Duke and Bunt, 1979). Rhizophora pollen sum content of at least 85% indicates that suitable sedimentation was present and provided conditions for dense Caution Bay plant growth at c. 2000 until approximately 1300–1000 cal. yr BP. Rhizophora forest appears to have established right up the tidal profile to border the terrestrial fringe during this period.

Enough minor pollen evidence exists to demonstrate that the Rhizophora forest was not a monospecific ecosystem, but subject to local structural variations. Openings are a common occurrence in young to middle-aged Rhizophora stands for example(Paijmans, 1976; Wightman, 2006). Canopy gaps would have facilitated the non-mangrove pollen inclusions recorded through core 1 and the lower zones of core 3, also admitting sufficient light for patches of undergrowth development including scattered ferns (e.g. Acrostichum), Acanthaceae, Brownlowia and occasional palm types. As the Rhizophora forest matures greater timber volume (Paijmans, 1976) largely closes the canopy and the penetrative ability of light and long-distance pollen declines (as portrayed in core 1 zone 3). Bruguiera is a common Rhizophora associate (Duke and Bunt, 1979; Wightman, 2006) and the Bruiguiera/Ceriops pollen type likely represents this genus when recorded with high Rhizophora . Bruguiera can form concentrated patches (Wightman, 2006) so the Caution Bay distribution may have been not of uniformly scattered individuals amongst the Rhizophora, but isolated groves.

Between c. 2000 and 1300–1000 cal. yr BP Avicennia is only a minor mangrove coinhabitor. Modern-day distribution is restricted by a degree of shade intolerance which would have affected an ability to initially coexist with Rhizophora (Duke, 1991). However, Avicennia is a genus demonstrating wide physiological tolerance to higher salinity, intertidal position and temperature and will outcompete Rhizophora when conditions shift inland beyond the lower-tidal window (Duke, 1991; Wightman 2006). At Caution Bay this shift approaches 1000 cal. yr BP. It is at this time that Rhizophora retreats seaward to establish its present-day distribution. The decline in Rhizophora values to 25% of the pollen sum in core 3 zone 1 confirms its onsite absence. Simultaneously Avicennia expands to establish and dominate a landward upper tidal zone. Given the biology of Avicennia pollen production and dispersal, values up to 32% are significant endorsement of dominant presence. Ceriops is also a competitive mangrove taxon toward the landward margins of tidal water (Paijmans, 1976; Wightman, 2006). Ceriops does not become an important canopy tree at Caution Bay, but may form a consistent fringing shrub to Avicennia woodland. Site succession away from Rhizophora includes development of an open mud-flat expanse, which after 1000 cal. yr BP is the zone bordering the terrestrial limit. Chenopodiaceae pollen suggests salt marsh communities occupy patches on an otherwise unvegetated mudflat. Local Poaceae is likely represented by the salt-tolerant Sesuvium and Sporobolus.

Pain and Swadling (1980) use landform features at Caution Bay’s Vaihua Inlet (c. 1.8 km southeast of the study site) to demonstrate fluctuations in local sea level through time. Exposed sandspit systems and the mudflats behind them are read as evidence of a sea level at least 3 m above present, approximated to sometime ‘during the last few thousand years’. Supportive dated regional compliations of sea level data indicate sustained mid-Holocene highstand followed by a pronounced fall after 2000 cal. yr BP (Lewis et al., 2012; Perry and Smithers (2011) interpret a northeast Australian +1–1.5 m highstand at c. 7000 cal. yr BP through to c. 2000 cal. yr BP. Sloss et al. (2007) propose an east Australian stable +1.5 m highstand between 7400 and 2000 cal. yr BP. Data from northern Australia also place sea level as 1–2 m above present until 2000 cal. yr BP (Nott, 1996; Woodroffe, 2009)). Perry and Smithers (2011) argue that an Australian smoothly falling sea level model remains the more plausible. Lewis et al. (2012) state the nature of mid- to late-Holocene sea level fall remains contentious and likely location specific. From sites in southern West Papua, Ellison (2005) indicates that high Holocene relative sea level was maintained at an estimated 0.67 mm/yr over the last 4000 year period. For a majority of the mid to late Holocene, a large proportion of the Caution Bay littoral zone sat well below increased sea levels and associated high-tide reach (cf. Ellison, 2005; Grindrod, 1985, 1988). Sediment accumulation was such that the sediment accommodation space remained unfilled at this time, the situation changing 2000–1700 cal. yr BP and thereafter. Comparatively it is difficult to calculate Caution Bay sediment accumulation rates given date intermixing and subsequent use of age ‘population’ trends. The sea level calculation by Ellison (2008), being from a large deltaic environment, precludes direct application of the 0.67 mm/yr figure. Ellison’s (2008) research reveals a transgressive Holocene mangrove sequence where discussions incorporate not only relative sea level rise as a factor in mangrove change but a degree of subsidence across the deltaic setting. Caution Bay is not interpreted as equally geomorphologically active.

A regressive Holocene succession of mangrove occurs at Caution Bay. For tidal-dominated systems, and during conditions of Holocene sea level stillstand, replacement of outer mangroves by more landward zones occurs not as a result of shoreline regression, but as a result of vertical sediment infilling, with implication in waterlogging and salinity. In the absence of large sediment supplies mangroves behave within a simple zonation (Woodroffe, 1990).

A small, slow sediment supply would explain the thin core depths, and a late c. 2000 cal. yr BP extensive mangrove (Rhizophora) presence coincidental with a fall (and stabilization) in sea level conditions. Small initial sediment supplies would have affected Caution Bay’s ability to override the c. 7000–2000 cal. yr BP highstand and infill the sediment accommodation space to facilitate plant growth through this period. It also suggests the catchment hinter-to-lowland surfaces and watercourses draining into the Bay were not suppliers of major sediment volume at this time. Seaward, and post 2000 cal. yr BP, south-coast PNG sediment transport may have incorporated embayment deposition, but in a pattern initially preserving widespread Rhizophora growth in equilibrium with water height. Subsequent shifts in sea level and/or tidal redistribution of sediment reworked deposits such that the embayment was raised to incorporate upper (inner) tidal zones favoring Avicennia. Any significant movement of sediment from offshore may also have been enough to form a natural levee or ridge. With any tidal barricading effect waterlogging and salinisation would ensue (cf. Woodroffe, 1995). This concept draws upon Pain and Swadling (1980), who observe not only sediment accumulation on the seaward side of the current Vaihua Inlet mangroves, but suggest that the sandspits at this site developed as sandbars associated with mangroves during high sea level. These were then abandoned as sea level fell and mangroves advanced seaward. Inland-sourced sedimentation must also be acknowledged, pending further investigation. Preliminary stratigraphic assessments from excavations across Caution Bay are signaling changes in sedimentation on the lowland plains coincident with the vegetation changes seen in this study (McNiven et al. (2012) also outline a number of hypotheses regarding upslope erosion and concomitant downslope sedimentation). Either way, the surface at core 3 reached and then exceeded the level of mean high-tide water. With less frequent tidal inundation (flushing), periodic desiccation (evaporative) and increases in soil salinity occurred landward, influencing mangrove composition and, more extremely, prompting tree dieback and salt marsh–unvegetated mudflat establishment. For mangroves, marked interspecific differences occur in both the range of salinity tolerances and the salinity in which growth is maximal (Blasco et al., 1996;Wightman, 2006). Paijmans and Rollet (1977) confirm the case of salinity as a major factor affecting differential distribution of mangrove species at Galley Reach, 30 km west of Caution Bay. It remains to examine inland vegetation and landscape change for further information.

A mosaic of coastal plain vegetation through the late Holocene

Combined with extra-local mangrove values, Figures 3 and 4 incorporate local, regional and long-distance terrestrial pollen. Pollen compositions suggest the existence of a late-Holocene terrestrial mosaic, including dune herbaceous, dune woodland and mixed coastal scrub or thicket communities in closest proximity to the mangrove-mudflat zone. In turn, sclerophyll woodland comprise the immediate background. Of the terrestrial taxa recorded, many occur within more than one vegetation community across the catchment area of today. Taxa are notably shared between the coastal thicket and (semi)deciduous hill-forest, including Terminalia, Ficus, Celtis, Fabaceae, Trema, Kelinhovia and Bombax (noting also a greater propensity toward shared emergent or canopy tree taxa). Based on pollen transport behaviors (see for example the Grindrod (1985) model for components in the transfer of pollen across a Queensland chenier plain), the pollen delivered to cores 3–1 appears to be from the immediate dune and low-lying plain rather than derived from hinterland sources in any great (interpretable) quantity (isolated occurrences of Ilex, Nothofagus, Araucariaceae, Podocarpus and Phyllocladus are not included in discussion as these are montane forest taxa (Paijmans, 1976), their pollen having originated outside the Caution Bay catchment). Because floristically the littoral thicket and lowland hill-forests are related, trends in the latter may be inferred from Figures 3 and 4. It follows that where thickets and forests trend toward expansion and taxa extend beyond these typical habitats, surrounding vegetation may also shift toward a mixed woodland rather than strict eucalypt canopy composition.

Through the c. 2000 year record the mangrove then mudflat vegetation was backed by the above-mentioned littoral dune and plains. From the base of core 3 a coastal thicket is present and at Caution Bay mixed evergreen and deciduous taxa cover the coastal dune ridge. In definition, ‘scrub’ comprises vegetation dominated by shrubs and/or low stunted trees, with a concentrated presence of herbaceous climbers. A ‘thicket’ differs through the presence of an open tree canopy above the scrub layer (Heyligers, 1965). Initially, trees such as Celtis and Bombax are dominant over Proteaceae, Boraginaceae, Melastoma and Crotalaria. The period c. 2000–1740 cal. yr BP also demonstrates greater diversity, and the presence of taxa such as Ficus, Euodia and Kelinhovia indicate a more moist ecosystem than that after 1740 cal. yr BP. The dune system through this earlier time period would also have been quite stable.

From the landward dune boundary thicket would grade gradually with available environmental moisture, interchanging with poorly drained depressions (characterizing the alluvial plains zone). Lower-lying swale-like depressions are occupied by Pandanus, swamp grasses and sedges, ferns and the aquatic herb Nymphaea/Nymphoides. Here, the Pandanus and Nymphaea/Nymphoides signal intermittent waterlogging (fresh) on the plains, their more frequent early occurrence in core 3 (zone 1) corresponding with indicators of moist beach thicket c. 2000–1740 cal. yr BP. Only Pandanus increases in distribution after 1000 yr BP, potentially expanding to occupy more cross-mosaic communities. Pandanus is also an indicator of disturbance (see Prebble et al., 2005).

Beginning 1740 cal. yr BP the coastal thicket experienced a decline in tree cover and canopy diversity, signaling a community transition toward coastal scrub. Canopy tree species such as Celtis, Kleinhovia and potential Fabaceae species decrease and the increased occurrence of Fabaceae and Desmodium (with spot presence of Scaevola) represent first-stage initiation of scrub expansion. In the past 1000 years the coastal ridge hosts a more prominent lower layer of shrubs such as Hibiscus and Solanaceae as well as ferns and the climber Flagellaria. Coastal woodland also appears to become more prominent beginning 1300 cal. yr BP. This woodland is initially dominated by Barringtonia followed by Casuarina. Palm types and Terminalia are incorporated; Arecaceae is increasingly common after 1000 cal. yr BP. Paijmans (1976) describes ‘Barringtonia formations’ as featuring wide-crowned tree habit with a low closed cover of Acanthus, ferns, grasses and herbs. Casuarinas similarly form an open canopy and are also a secondary forest indicator (Prebble et al., 2010). Myrtaceae pollen values are small, and form a less consistent component of the sclerophyll woodland than Poaceae throughout the entire record period.

Caution Bay burning

Microcharcoal concentrations are included in Figures 3 and 4. Mangrove vegetation does not typically burn. In the Galley Reach mangrove survey, Paijmans and Rollet (1977) comment upon lightning strike damage and the minimal local tree scorching with no associated fire spread. Usefully, locally derived charcoal may be discounted in interpretation. However, the closed nature of mangrove canopies questions the extent to which charcoal particles from regional catchment fires have been able to penetrate into the trunk space and accumulate into fully representative records. Additional evidence was therefore sought. Microcharcoal remains were examined from the archaeological site Bogi 1 (Figure 5), considered to potentially provide more accurate reflections of regional burning owing to its position above the mudflat and on the coastal foredune. In exploration of this potential, the upper 24 archaeological excavation units (XU) were examined.

Figure 5.

Caution Bay site Bogi 1 microcharcoal concentrations presented per excavation unit (XU) and with a magnified insert demonstrating values between XU6-24 (Square C). Age results originally published in McNiven et al. (2011).

All charcoal records fluctuate. High, increasing charcoal concentrations were recorded at Bogi 1 between XU6 and 1, corresponding to the past c. 1500 years. Secondary peaks occur across XU9-8, again corresponding to 1500 cal. yr BP, and at XU14 (c. 2050 cal. yr BP). The highest (and least variable) charcoal concentrations through pollen core 3 occur within zone 2 (c. 1740–1300 cal. yr BP) culminating in a peak closer to c. 1740 cal. yr BP. Through pollen core 1 an isolated lower spike in charcoal has been recorded at c. 1500 cal. yr BP. Charcoal concentrations then decline before a series of higher values occur between 750 and 300 cal. yr BP. Interestingly, all three charcoal records correlate, reinforcing increased catchment burning through approximately 2000–1400 cal. yr BP.

‘Variable’ is a term commonly used to describe the late-Holocene climate of Greater Australia and adjacent Pacific rim. Regional climate demonstrates an overall mid- to late-Holocene drying trend, upon which variable episodes are superimposed. Lees (1992; north mainland Australia) details this drying trend as being interrupted by periods of increased precipitation, occurring from 3500 and 2800 yr BP, again between 2100 and 1600 yr BP and at several stages in the last 1000 years. Palaeoclimate trends from Port Moresby and the Central Province also incorporate periodically wetter conditions (e.g. c. 2500 yr BP, increasing further through 1700–1200 yr BP and predominantly from 1000 to 700 yr BP at Waigani Lake; Osborne et al. (1993) suggesting increases in available moisture equally likely through decreased evaporation and temperature as increased precipitation). Lake Wanum (northern lowlands Markham Valley, Morobe Province) reveals increases in late-Holocene lake levels (Prebble et al., 2010). Rowe (2006, 2007) demonstrates island swamp water initiation in Torres Strait c. 2500 and again after 1000 yr BP. Drier climatic episodes are potentially associated with persistent large-scale sinking, dry southeasterly trade flows over the Greater Australian tropics inhibiting the onset of the Asian-Australian monsoon. These are similar to atmospheric flows associated with modern-day El Niño-Southern Oscillation (ENSO). Prominent ENSO activity after c. 5000 cal. yr BP has been suggested from the western Pacific (Haberle et al., 2001) as well as northern mainland Australia (Haberle, 2005). Mid- to late-Holocene climatic implications following the flooding of the Sahul and Sunda shelves, and opening of the Indonesian ‘throughflow’ (Pacific to Indian Oceans, and consequential ‘high-energy-window’), is also to be considered (Schulmeister and Lees, 1995).

With numerous authors indicating late-Holocene wet–dry phasing, variability in landscape fire and some soil movement would be expected. Higher precipitation and coinciding highest charcoal records at Caution Bay may represent a local fire regime decoupled from climatic dictation and maintained through anthropogenic action irrespective of the moist conditions. It may represent a perceived anthropogenic need for greater control of increased plant biomass facilitated by moister climate phases. A Caution Bay shift c. 1740–1300 cal. yr BP between coastal thicket and scrub and a greater representation of coastal woodland-grassland is consistent with recurrent drier (and/or variable) conditions as well as fire impact. Increased secondary forest and disturbance indicators are conceivably also driven by greater rainfall variability, burning and/or eroded sediments.

Palaeoecological-archaeological dialogue

Recent archaeological results from Caution Bay demonstrate pottery-using peoples (Lapita) settled the region 2900 cal. yr BP with pre-ceramic occupation extending further, to at least 5000 cal. yr BP along the coast and 6000 cal. yr BP on adjacent lowlands. Implications of the first known occurrence of Lapita peoples on the New Guinea mainland are discussed in David et al. (2011) and McNiven et al. (2011). The region’s palaeoecology, as discussed in this paper, has implications for further understanding the human occupation of Caution Bay. Particular attention is drawn to the parallel timing of mangrove development, final major midden concentrations at site Bogi 1 (post-Lapita dated to 2000–2150 cal. yr BP and with minimal foredune development thereafter), and evidence of Caution Bay as having hosted a dynamic, non-transient Lapita community occupying (interacting) with shoreline to inland locations (David et al., 2012; McNiven et al., 2011). Increased burning and vegetation shifts at Caution Bay after c. 2000 cal. yr BP further parallel evidence from the West New Britain and Western Highlands Provinces, correlated with archaeological indicators for changes toward systematic plant management and reduced settlement mobility that suggest widespread and increasing dependence on plant food production(Denham and Haberle, 2008; Lentfer et al., 2010).

It is worth considering that a widespread, established Rhizophora forest dating to c. 2000 cal. yr BP would effectively barricade shoreline access. Rhizophora’s multibranched tree with arched-root habit would hinder easy human and notably watercraft movement and/or usher land access by way of penetrative watercourse channels. Late-Holocene changes in sea level, potentially tidal direction and/or strength, may also harbour consequences not only for intertidal ecologies, but near-coast human sea-travel patterns and predictability. As a near-monoculture environment, the Rhizophora forest would host less human-use resources than found in communities of more diverse composition and structure, including the post 1300–1000 cal. yr BP tidal zonation. Further, the manner in which mangrove communities secure sediments excludes the intertidal zone as an offshore source of sediments capable of contributing meaningfully to foredune development post 2000 cal. yr BP. A significant cultural trend for Caution Bay generally is the disappearance of most settlements in the study area around 2000 cal. yr BP (after some 900 years of Lapita followed by post-Lapita villages). After 2000 cal. yr BP, occasional settlements come and go in the study area (especially around 1700 cal. yr BP) and by 1200 years ago there is evidence of occupation of the Boera district immediately to the southeast. Notably, Boera is not fringed by mangroves and it is suggested that the shifting village locations were influenced by the development of dense mangroves fronting the Bogi 1 shoreline and much of Caution Bay.

Retrieved ceramics of excavated horizons dated from c. 2900–2600 cal. yr BP in Caution Bay comprise the Lapita material cultural ‘signature’ (they are as Galipaud (2006) describes the ‘best witnesses’ to the Lapita tradition). From a palaeoenvironmental perspective however, much discussion and debate surrounds so-called associated signature trends. Taking the western Pacific in example, patterns in island transformation are evident. Relatively little initial impact is recorded. Changes in sediment distribution and vegetation composition then develop as Lapita population densities increase. Landscape transformations include loss of arboreal taxa and noticeably shifted sediments from the hills to coast (cf. Anderson (2002) citing numerous cases. See also Sand (1997) and Stevenson (1999) for New Caledonia, Anderson and Clarke (1999) and Enright and Gosden (1992) for Fiji). Extensive plant loss and an erosional influx of sediment onto the littoral plains zone is not evident at Caution Bay based on the data generated in this study. In explanatory difference, the Lapita peoples were not the first occupiers of Caution Bay, and likely entered into a landscape already modified by humans. Any subsequent ‘associated signature’ changes imposed by Lapita simply may have had less of a shock landscape value, or were more effectively absorbed by the greater New Guinean (non-isolate island) land-mass. Indeed, McNiven et al. (2012) highlight that because of required interaction with existing social and environmental landscapes, regional south PNG Lapita settlement was a more complexly negotiated process comparative to other parts of the western Pacific, questioning overtly simple transplant methodology and value. In looking forward, therefore, further inland evidence is required where additional direct palaeoenvironmental description can combine with archaeological deduction. The focus will need to be on environmental change local to Caution Bay per se, on the movement and likely source of accumulating sediments, and to further pursue palynological disturbance indicators and decreases in vegetation cover. Such focus will then extend the understanding of climatic fluctuations, and provide a more specific framework for comparative study against Quaternary highland PNG change. Focus toward long-term distinction and distribution between coastal lowland and hill-forest types (dry evergreen forest, deciduous forest), the floristics and structure of mixed woodland (grassed to wooded), fire occurrence as well as a better understanding of the formation and influential role of littoral zone sandy ridges and foredunes is also suggested.

Conclusion

The Caution Bay cores 1 and 3 sequence illustrates a late-Holocene mangrove–mudflat transition leading from the lower intertidal environment through Rhizophora, Rhizophora-Bruguiera, Avicennia, Avicennia-Ceriops, and Chenopodiaceae-Cyperaceae dominated communities including the present bare supratidal mudflat. The dating noise potentially produced by tidal or floodwater sediment redistribution and by bioturbation appears to be minimal relative to the palynological signal. There is no evidence that the littoral succession suffered interruption and it is proposed that the timing and mangrove pattern is largely a function of low sediment loading, sedimentary redistribution and salinisation, all upon sea level decline c. 2000 cal. yr BP. Caution Bay demonstrates that for the south coast (including Gulf of Papua) site-alternate environmental settings possess individual palaeorelationships, both in terms of substrate characteristics and patterns of sedimentation as control measures on mangrove vegetation, and in the manner in which these alter as the land–sea relationship has adjusted during the Holocene.

Simultaneously for PNG, this paper provides new and expansive inland, southern lowland palaeoecological investigation (see also McNiven et al., 2010; Osborne et al., 1993). Late-Holocene coastal alluvial and dune thickets were replaced by scrub formations and woodlands as regional climate incorporated a number of variable wet–dry phases. Beyond the immediate coastal zone, however, inland results are tentative. Changes across the Caution Bay catchment have been inferred from those at the coast, and from this stage in the data the entire landscape may be continuously responding and shifting. Further work is required, where patterns (or strategies) in burning and an archaeological collaboration will play a big part in determining spatial and temporal variability. For the archaeology a key to this future analysis will be an ability to separate and extend in understanding local and Lapita-colonist environmental interaction and influence.

Footnotes

Acknowledgements

This study was carried out in 2010. Warm thanks to the people of Porebada, Boera, Papa and Lealea for their field assistance and local insights into the Caution Bay environment. Thanks toSimon Coxe and Laura Bates for assistance with fieldwork; Tanja Harding for general support, and Fiona Petchy and Alan Hogg (Waikato Radiocarbon Laboratories) for radiocarbon determinations. The assistance received from the PNG National Museum is gratefully acknowledged. Within the School of Geography and Environmental Science at Monash University (Clayton Campus), thanks also to Ursula Pietrzak for laboratory advice and Beverley Brigham for drafting the figures. Advice from two anonymous referees was also gratefully received and is acknowledged.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. BD thanks the Australian Research Council for grant and QEII Fellowship DP0877782.

References

Anderson

(2002) Faunal collapse, landscape change and settlement history in remote Oceania. World Archaeology 33: 375–390.

Anderson

Clarke

(1999) The age of Lapita settlement in Fiji. Archaeology in Oceania 34: 31–39.

Australian Pollen and Spore Atlas Members (2007) The Australasian Pollen and Spore Atlas v1.0. Canberra: Australian National University, available at: http://apsa.anu.edu.au/

Bartlett

Barghoorn

(1973) Phytogeographic history of the isthmus of Panama during the last 12,000 years (a history of vegetation, climate and sea level change). In: Graham

(ed.) Vegetation and Vegetational History of Northern Latin America. Amsterdam: Elsevier, pp. 203–299.

Bennett

Willis

(2001) Pollen. In: Smol

Birks

HJB

Last

(eds) Tracking Environmental Change Using Lake Sediments. Volume 3: Terrestrial, Algal and Siliceous Indicators. Dordrecht: Kluwer, pp. 1–32.

Björck

Wohlfarth

(2001) 14C chronostratigraphic techniques in palaeolimnology. In: Smol

Birks

HJB

Last

(eds) Tracking Environmental Change Using Lake Sediments. Volume 3: Terrestrial, Algal and Siliceous Indicators. Dordrecht: Kluwer, pp. 205–245.

Blasco

Saenger

Janodet

(1996) Mangroves as indicators of coastal change. Catena 27: 167–178.

Chappell

(2005) Geographic changes of coastal lowlands in the Papuan past. In: Pawley

Attenborough

(eds) Papuan Pasts: Cultural, Linguistic and Biological Histories of Papuan-speaking Peoples. Canberra: ANU Press, pp. 525–539.

Chappell

Grindrod

(1984) Pollen analysis: A key to past mangrove communities and successional changes in northern Australian coastal environments. In: Bardsley

Davie

JDS

Woodroffe

(eds) Coasts and Tidal Wetlands of the Australian Monsoon Region. Canberra: ANU Press Mangrove Monograph 1, pp. 225–236.

10.

David

McNiven

Leavesley

. (2012) A new ceramic assemblage from Caution Bay, south coast of mainland PNG: The Linear Shell End-Impressed Tradition from Bogi 1. Journal of Pacific Archaeology 3: 73–89.

11.

David

McNiven

Richards

. (2011) Lapita sites in the Central Province of mainland Papua New Guinea. World Archaeology 43: 580–597.

12.

Denham

Haberle

(2008) Agricultural emergence and transformation in the Upper Wahgi Valley, Papua New Guinea during the Holocene: theory, method and practice. The Holocene 18: 481–496.

13.

Duke

(1991) A systematic revision of the mangrove genus Avicennia (Avicenniaceae) in Australasia. Australian Systematic Botany 4: 299–324.

14.

Duke

Bunt

(1979) The genus Rhizophora (Rhizophoraceae) in northeastern Australia. Australian Journal of Botany 27: 657–678.

15.

Ellison

(2005) Long-term retrospection on mangrove development using sediment cores and pollen analysis: A review. Aquatic Botany 89: 93–104.

16.

Ellison

(2008) Holocene palynology and sea-level change in two estuaries in Southern Irian Jaya. Palaeogeography, Palaeoclimatology, Palaeoecology 220: 291–309.

17.

Enright

Gosden

(1992) Unstable archipelagos – Southwest Pacific environment and prehistory since 30,000 BP. In: Dodson

(ed.) The Naïve Lands: Prehistory and Environmental Change in Australia and the Southwest Pacific. Melbourne: Longman-Cheshire, pp. 160–198.

18.

Fairbanks

Mortlock

Chiu

T-C

. (2005) Marine radiocarbon calibration curve spanning 0 to 50,000 years B.P. based on paired ²³⁰Th/²³⁴U/²³⁸U and ¹⁴C dates on pristine corals. Quaternary Science Reviews 24: 1781–1796.

19.

Fitzpatrick

(1965) Climate of the Port Moresby-Kairuku area. In: Mabbutt

Heyligers

Scott

. (eds) Lands of the Port Moresby-Kairuku Area. Melbourne: CSIRO Land Research Series No. 14, pp. 83–97.

20.

Furukawa

Wolanski

Mueller

(1997) Currents and sediment transport in mangrove forests. Estuarine, Coastal and Shelf Science 44: 301–310.

21.

Galipaud

J-C

(2006) The first millennium B.C. in remote Oceania: An alternative perspective on Lapita. In: Lilley

(ed.) Archaeology of Oceania: Australia and the Pacific Islands. Melbourne: Blackwell Publishing, pp. 228–240.

22.

Gillison

(1983) Tropical savannas of Australia and the southwest Pacific. In: Bourliére

(ed.) Ecosystems of the World 13: Tropical Savannas. Amsterdam: Elsevier, pp. 183–244.

23.

Grimm

(1991) Tilia Program Ver. 2.0 B4. Springfield: Illinois Museum.

24.

Grindrod

(1985) The palynology of mangroves on a prograded shore, Princess Charlotte Bay, north Queensland, Australia. Journal of Biogeography 12: 232–348.

25.

Grindrod

(1988) The palynology of Holocene mangrove and saltmarsh sediments, particularly in northern Australia. Review of Palaeobotany and Palynology 55: 229–245.

26.

Grindrod

Rhodes

(1984) Holocene sea-level history of a tropical estuary: Missionary Bay, north Queensland. In: Thom

(ed.) Coastal Geomorphology in Australia. Sydney: Academic Press, pp. 151–178.

27.

Haberle

(2005) A 23,000-yr pollen record from Lake Euramoo, wet tropics of NE Queensland, Australia. Quaternary Research 64: 343–356.

28.

Haberle

(2007) Prehistoric human impact on rainforest biodiversity in highland New Guinea. Philosophical Transactions of the Royal Society B: Biological Sciences 362: 219–228.

29.

Haberle

Hope

van der Kaars

(2001) Biomass burning in Indonesia and Papua New Guinea: Natural and human induced fire events in the fossil record. Palaeogeography, Palaeoclimatology and Palaeoecology 171: 259–268.

30.

Heyligers

(1965) Vegetation and ecology of the Port Moresby-Kairuku area. In: Mabbutt

Heyligers

Scott

. (eds) Lands of the Port Moresby-Kairuku Area, Territory of Papua and New Guinea. Melbourne: CSIRO Land Research Series No. 14, pp. 146–173.

31.

Hope

(2007) Palaeoecology and palaeoenvironments of Papua. In: Marshall

Beehler

(eds) The Ecology of Papua. Singapore: Periplus Editions, pp. 255–266.

32.

Krauss

Lovelock

McKee

. (2008) Environmental drivers in mangrove establishment and early development: A review. Aquatic Botany 89: 105–127.

33.

Kristensen

Bouillon

Dittmar

. (2008) Organic carbon dynamics in mangrove ecosystems: A review. Aquatic Botany 89: 201–219.

34.

Jorry

Droxler

Mallarino

. (2008) Bundled turbidite deposition in the central Pandora Trough (Gulf of Papua) since Last Glacial Maximum: Linking sediment nature and accumulation to sea level fluctuations at millennial timescale. Journal of Geophysical Research 113: 1–15.

35.

Lees

(1992) Geomorphological evidence for late Holocene climatic change in northern Australia. Australian Geographer 23: 1–10.

36.

Lentfer

Pavlides

Specht

(2010) Natural and human impacts in a 35,000-year vegetation history in central New Britain, Papua New Guinea. Quaternary Science Review 29: 3750–3767.

37.

Lewis

Sloss

Murray-Wallace

. (2012) Post-glacial sea-level changes around the Australian margin: A review. Quaternary Science Reviews https://dx-doi-org.web.bisu.edu.cn/10.1016/j.quascirev.2012.09.006

38.

Löffler

(1985) Landforms and landform development. In: Gressitt

(ed.) Biogeography and Ecology of New Guinea. The Hague: Junk Publishing, pp. 57–73.

39.

McAlpine

Keig

Falls

(1983) Climate of Papua New Guinea. Canberra: CSIRO in association with ANU Press.

40.

McNiven

David

Aplin

. (2010) Historicising the present: Late Holocene emergence of a rainforest hunting camp, Gulf Province, Papua New Guinea. Australian Archaeology 71: 41–56.

41.

McNiven

David

Aplin

. (2012) Terrestrial engagements by terminal Lapita maritime specialists on the southern Papuan coast. Terra Australis 34: 121–156.

42.

McNiven

David

Richards

. (2011) New direction in the human colonisation of the Pacific: Lapita settlement of the south coast New Guinea. Australian Archaeology 72: 1–6.

43.

Mabbutt

(1965) Geomorphology of the Port Moresby-Kairuku area. In: Mabbutt

Heyligers

Scott

. (eds.) Lands of the Port Moresby-Kairuku Area, Territory of Papua and New Guinea. Melbourne: CSIRO Land Research Series No. 14, pp. 104–128.

44.

Mabbutt

Heyligers

Pullen

. (1965) Land systems of the Port Moresby-Kairuku area. In: Mabbutt

Heyligers

Scott

. (eds) Lands of the Port Moresby-Kairuku Area, Territory of Papua and New Guinea. Melbourne: CSIRO Land Research Series No. 14, pp. 20–82.

45.

Nott

(1996) Late Pleistocene and Holocene sea-level highstands in northern Australia. Journal of Coastal Research 12: 907–910.

46.

Osborne

Humphreys

Polunin

NVC

(1993) Sediment deposition and late Holocene environmental change in a tropical lowland basin: Waigani Lake, Papua New Guinea. Journal of Biogeography 20: 599–613.

47.

Paijmans

(1976) New Guinea Vegetation. Canberra: ANU Press.

48.

Paijmans

Rollet

(1977) The mangroves of Galley Reach, Papua New Guinea. Forest Ecology and Management 1: 119–140.

49.

Pain

Swadling

(1980) Sea level changes, coastal landforms and human occupation near Port Moresby – A pilot study. Science in New Guinea 7: 57–68.

50.

Pearsall

(2000) Paleothnobotany: Handbook of Procedures. San Diego: Academic Press.

51.

Perry

Smithers

(2011) Cycles of coral reef ‘turn-on’, rapid growth and ‘turn-off’ over the past 8500 years; A context for understanding modern ecological states and trajectories. Global Change Biology 17: 76–86.

52.

Prebble

Kennedy

Southern

(2010) Holocene lowland vegetation change and human ecology in Manus Province, Papua New Guinea. Terra Australis 32: 299–321.

53.

Prebble

Sim

Finn

. (2005) A Holocene pollen and diatom record from Vanderlin Island, Gulf of Carpentaria, lowland tropical Australia. Quaternary Research 64: 357–371.

54.

Ramsey

(2009) Bayesian analysis of radiocarbon dates. Radiocarbon 51: 337–360.

55.

Rowe

(2006) A Holocene history of vegetation change in the western Torres Strait region, Queensland, Australia. PhD thesis. School of Geography and Environmental Science, Monash University.

56.

Rowe

(2007) Vegetation change following mid-Holocene marine transgression of the Torres Strait shelf: A record from Mua. The Holocene 17: 917–927.

57.

Rowe

(2012) Modern surface pollen from the Torres Strait Islands: Exploring north Australian vegetation heterogeneity. Terra Australis 34: 413–434.

58.

Sand

(1997) The chronology of Lapita ware in New Caledonia. Antiquity 71: 539–547.

59.

Schulmeister

Lees

(1995) Pollen evidence from tropical Australia for the onset of an ENSO dominated climate at 4000 yr BP. The Holocene 5: 10–18.

60.

Shearman

(2010) Recent change in the extent of mangroves in the northern Gulf of Papua, Papua New Guinea. AMBIO 39: 181–189.

61.

Sloss

Murry-Wallance

Jones

(2007) Holocene sea level change on the southeast coast of Australia: A review. The Holocene 17: 999–1014.

62.

Stevenson

(1999) Human impact from the palaeoenvironmental record in New Caledonia. In: Galipaud

Lilley

(eds) The Pacific from 5000 to 2000 BP: Colonisation and Transformation. Paris: Editions de IRD, pp. 251–258.

63.

Thanikaimoni

(1987) Mangrove Palynology. UNDP/UNESCO Regional Project on Training and Research on Mangrove Ecosystems, RAS/79/002, and the French Institute, Pondicherry.

64.

Thom

Wright

(1983) Geomorphology of the Purari Delta. In: Petr

(ed.) The Purari-Tropical Environment of a High Rainfall Basin. The Hague: Junk Publishing, pp. 47–65.

65.

Tomlinson

(1986) The Botany of Mangroves. Cambridge: Cambridge University Press.

66.

Whitlock

Larson

(2001) Charcoal as a fire proxy. In: Smol

Birks

Last

(eds) Tracking Environmental Change Using Lake Sediments. Volume 3: Terrestrial, Algal and Siliceous Indicators. Dordrecht: Kluwer Academic Publishing, pp. 75–97.

67.

Wightman

(2006) Mangroves: Identification and Traditional Use. Darwin: Northern Territory Government, Department of Natural Resources, Environment and the Arts.

68.

Woodroffe

(1990) The impact of sea-level rise on mangrove shorelines. Progress in Physical Geography 14: 483–520.

69.

Woodroffe

(1995) Response of tide-dominated mangrove shorelines in northern Australia to anticipated sea-level rise. Earth Surface Processes and Landforms 20: 65–85.

70.

Woodroffe

Thom

Chappell

(1985) Development of widespread mangrove swamps in mid-Holocene times in northern Australia. Nature 317: 711–713.

71.

Woodroffe

(2009) Testing models of mid to late Holocene sea-level change, north Queensland, Australia. Quaternary Science Reviews 28: 2474–2488.