Abstract
In Australia, the drivers of precolonial fire regimes remain contentious, with some advocating an anthropogenic-dominated regime, and others highlighting the importance of climate, climatic variability or alternatively some nexus between climate and human activity. Here, we explore the inter-relationships between fire, humans and vegetation using macroscopic charcoal, archaeology and palynology over the last ~5430 cal. year BP from Broughton Island, a small, near-shore island located in eastern Australia. We find a clear link between fire and the reduction of arboreal pollen and rainforest indicators on the island, especially at ~4.0 ka and in the last ~1000 years. Similarities with comparable palaeoenvironmental records of fire in the region and a record of strong El Niño (dry, fire-prone) events supports the contention that climate was a significant influence on the fire regimes of Broughton Island. However, two periods of enhanced fire activity, at ~4000 years BP and ~<600 years BP have weaker links to climate, and perhaps reflect anthropogenic activity. Changes to the fire regime in the last ~600 years corresponds with the earliest evidence of Indigenous archaeology on the island, and coincides with implications that Polynesian people were present in the region. After the mid-Twentieth Century a human-dominated fire regime is also an obvious feature of the reconstructed fire record on Broughton Island.
Introduction
This paper describes a palaeoenvironmental investigation of the last ~5430 calibrated years Before Present (cal. year BP) from Broughton Island, a near-shore island located in humid subtropical eastern Australia. The aim was to use high resolution macroscopic charcoal analysis and palynology to reconstruct past fire activity and determine any inter-relationships between fire and vegetation on the island. The work also aimed to examine how fire and vegetation related to archaeological information and past climatic variability in the region to allow exploration of the potential drivers of fire in this landscape.
Previous research conducted on the central eastern coast of Australia (Black et al., 2007; Black and Mooney, 2006, 2007; Mooney et al., 2007) and in a compilation of sites across Australia (Williams et al., 2015) and Australasia (Mooney et al., 2011, 2012) has suggested that fire over long temporal timeframes can be best related to climatic drivers. Williams et al. (2015), in particular, found little evidence for any statistical links between compiled archaeological information and charcoal data, arguably reflecting little correlation between human and fire activity. At best, the data indicated that anthropogenic fire was likely to have been undertaken only at small scales and during climatic or socio-econonomic transitions when more complex procurement techniques were required for survival. They concluded that this was unlikely to result in ecological or landscape change.
Here, instead of focusing on compilations at a continental or regional scale, we focus on one particularly well-defined, island landscape to further explore these broader conclusions. In this research we were particularly interested in the question as to whether fire in the landscape might be related to shorter-scale climatic variability such as El Niño-Southern Oscillation (ENSO). Climatic variability on the ENSO time scale appears as a feature throughout the Holocene (Rodó and Rodriguez-Arias, 2004), but there is considerable evidence for an increase in ENSO strength and/or variability after the mid-Holocene (see e.g. Andrus et al., 2002; Conroy et al., 2008; Koutavas et al., 2012; Moy et al. 2002; Rein et al., 2005; Rodbell et al., 1999; Tudhope et al., 2001).
Lynch et al. (2007) argued that ENSO derived climatic variability is an important driver of fire in Australia and thus attributed low fire activity in the mid-Holocene followed by higher levels of biomass burning after ca. 4 ka to ENSO. Mooney et al. (2011, 2012) and Williams et al. (2015) found no clear relationship between changes in an ENSO-sensitive proxy (e.g. Moy et al., 2002) and patterns of biomass burning in either the entire continental or in smaller, latitudinal-based compilations. This is perhaps unsurprising given that this mode of climatic variability only affects some regions of Australia in a relatively complex spatial pattern.
Black and Mooney (2006, 2007) argued that in some locations in the Sydney Basin, including at Kings Waterhole located to the south of this research, people strongly influenced fire in the late-Holocene due to the increased risk of severe fire as an ENSO-dominated climate became established. Black and Mooney (2006, 2007) thus proposed that fire activity in the landscape resulted from a complex nexus between climate and humans. As noted, Williams et al. (2015) found few synchronous responses between archaeological radiocarbon data and a synthesis of charcoal data in southern (25 – 45oS) Australia but notably there were relatively short-lived exceptions to this, and this included a significant correlation during significant transitions of ENSO, including in the period ~5-4 ka and 1.2-0.8 ka. They hence argued that Aboriginal adopted or increased the management of fire during societal stress resulting from climatic variability.
Site description
Broughton Island (32°37’S, 152°19E) (Figure 1) is located on the New South Wales central coast, approximately 230 km north of Sydney and 16 km north-east from the entrance to Port Stephens. The island is 3 km offshore from Dark Point, which is an important archaeological site and a declared Aboriginal Place (DEC, 2007; Thorpe, 1929; White et al., 2014). The palaeoenvironmental analyses described herein were completed on a small limnogeneous minerotrophic peatland located approximately 45 m from the high tide mark at the far eastern end of Providence Beach (Figure 1). It is hereafter referred to as Grey Pole Swamp.

The location of Broughton Island in eastern Australia, showing the region, with insets depicting Broughton Island and Grey Pole Swamp. The doted line on the Broughton Island inset is the approximate water catchment for Grey Pole Swamp. An asterisk (*) marks the position of the sampled sediment core used in this resarch. The regional map was drafted using Google Map Styling Wizard (https://mapstyle.withgoogle.com/ accessed March 2019). The insets are geo-registered (to Geographic Datum of Australia 1994 Map Grid Zone 56) extracts from the Port Stephens (Run 4 Photo 0135 taken on the 03/01/2001) aerial photograph taken by NSW Land and Property Information.
The headlands and bed-rock islands of the Port Stephens – Myall Lakes region are Carboniferous sediments with interbedded Nerong Volcanics, and the intervening embayments have extensive sand deposits of Quaternary age (Matston and Chesnut, 1975; Patteson, 2003; Thom et al., 1981). Broughton Island is one such rocky outcrop, with the geology including toscanite, dacite, andesite and ignimbrite and sedimentary units of agglomerate, conglomerate, sandstone and siltstone (NPWS, 2002, Patteson, 2003).
The varying resistance of the geology that makes up Broughton Island has resulted in an irregular shape with a major axis, running east-west, of approximately 3.2 km, and a width of up to 1.6 km. The majority of the island is a slightly undulating plateau and it attains a maximum altitude of 94 m a.s.l. (Dodkin, 1981; EJE Group, 2000; NPWS, 2002; Patteson, 2003). Most of the shoreline is rocky, with 25% of the shoreline being steep cliffs but sandy beaches occur, including at Providence Beach (EJE Group, 2000).
The soils of island are similar to those of the northern sections of Port Stephens (the ‘North Arm Cove’ soil type) but the higher altitude eastern portion of the island is mapped as the Gan Gan soil type (Murphy, 1995). The relatively high nutrient status of the Nerong Volcanics formation has produced soils that support rainforest communities on comparable landforms such as Yacaaba and Tomaree Head, at the entrance to Port Stephens and Cabbage Tree Island (NPWS, 2002). Grey Pole Swamp is immediately adjacent to outcrops of sedimentary units which have a lower nutrient status (Murphy, 1995). Wind-transported beach sands also contribute to the soil mantle across the island.
Broughton Island experiences a humid subtropical climate (Köppen climate classification) characterised by mild winters and hot summers (BOM, 2019a). Precipitation tends to occur all year round however considerable variability occurs from year to year. The mean annual rainfall for Nelson Bay (located approximately 18 km south-west of Broughton Island; Figure 1) is 1342 mm. The mean annual daily temperature ranges from 13.3°C to 23.0°C. In winter, the average daily minimum and maximum temperatures are 8.6°C and 19.3°C respectively and in summer are 17.5°C and 26.9°C (BOM, 2019a; averages from 1881 to 2004). In spring and summer, winds are typically from the north and north-east and in late winter wind has a stronger westerly component. In the morning the wind is usually (>40% of days) from the west and the north-west but by the afternoon the wind swings to the south-east. The mean annual wind speed ranges from 13.9 km/h at 9am and 18.4 km/h at 3pm (BOM, 2019a).
Broughton Island is located within a zone of transition between subtropical, summer maximum rainfall patterns to the north and temperate, winter maximum rainfall patterns to the south. Historically this generates the most extreme fire weather during late winter and spring when moderate temperatures can combine with strong winds and low relative humidity. In most years summer rains mark the end of the fire season however when these fail it may last until March (NPWS, 2014). The average impact of strong El Niño events on rainfall in coastal eastern Australia is relatively weak in winter/spring and any effect on summer rainfall is minimal (BOM, 2019b). In contrast, in strong La Nina years, rainfall deciles are well above average across a broad austral warm season (November to March) in some areas of coastal eastern Australia, and this includes near Broughton Island (BOM, 2019b).
Broughton Island forms part of Myall Lakes National Park (MLNP, Figure 1): on the mainland the spatial distribution of vegetation is controlled by soils, and particularly the age of the substrates (the inner barrier is of Pleistocene age and the outer barrier is Holocene) and hence the degree of podzolisation, but also topography, exposure, drainage and nutrient status (Myerscough and Carolin, 1986; Thom et al., 1992). Local fire regimes also determine floristic composition, diversity and community structure (Fox, 1988; Fox and Fox, 1986; Hunter, 2000; Myerscough and Carolin, 1986).
The extant vegetation of Broughton Island is described, based on field observations, in Results below. A limited number of terrestrial faunal species are found on Broughton Island (White-Tailed Rats? Uromys hadrourus, reptiles and a colony of the Endangered or Vulnerable Green and Golden Bell frog Litoria urea) (NPWS, 2002). In contrast, a number of bird species nest in the island including the Little Penguin (Eudyptula minor), Sooty Shearwater (Puffinis griseus), Wedge-Tailed Shearwater (Puffinis pacificus), Short-Tailed Shearwater (or Mutton birds Puffinus tenuirostris), the Silver Gull (Chroicocephalus novaehollandiae) and Crested Tern (Thalasseus bergii spp. cristata) (Patteson, 2003). Mutton bird burrows occur on some of the sandy sections of the island and they nest in spring and early summer. Europeans have introduced rabbits (Oryctolagus cuniculus), black rats (Rattus rattus) and a herd of goats (Capra hircus) also existed prior to 1972 (Dodkin, 1981).
The people and archaeology of the region and of Broughton Island
The Myall Lakes and Port Stephens region is encompassed within the boundaries of the Worimi people (Garland and Wheeler, 1982). Geographically, the Worimi occupied the land between Barrington Tops and Forster in the north and Maitland and the Hunter River in the south (Marr, 1997; Sutherland, 2011). Evidence of the occupation and land use by the Worimi people is abundant in the region and includes ceremonial sites (stone arrangements, bora grounds, carved trees and burial sites), campsites, middens, canoe and otherwise scarred trees, several wells, a fish trap, abraded grooves and quarries. Occupation of the area is thought to have been more permanent compared to the seasonal use of some other areas of southern Australia and the coastal environment, favourable climate and reliable freshwater sources probably supported a relatively large Aboriginal community (Garland and Wheeler, 1982).
Despite extensive ethnographic accounts of Aboriginal occupation (e.g. Enright, 1932, 1936; Threlkeld, 1834), there have been few detailed archaeological studies in the region to the immediate north of the Hunter River. Amateur collectors visited Dark Point in 1928 (Thorpe, 1929; White et al., 2014). Modern investigations began in the 1970s by Dyall (1971, 1972) and included early and late-Holocene midden assemblages from Swansea and Port Stephens, respectively. In the sand dune country of the Newcastle Bight numerous sites are evident (e.g. Dean-Jones, 1990) but their ages are constrained to about the last 4500 years by their location on the outer barrier dune sands.
This lack of deep-time or other stratified sites led to a general shift of archaeological focus away from the region. Instead, the well-documented Upper Hunter Valley (Moore, 1981) and Upper Mangrove Creek (Attenbrow, 2003, 2004) have been used as a proxy for past occupation in this region, with the exception of Moffats Swamp (Figure 1), near Medowie, is often referenced as one of the earliest sites of Aboriginal activity for the region (Baker, 1994). The basal age of ~16,500 cal. year BP at Moffats Swamp (test pit A1, spit 6 – NZA-3016: 14,750 ± 130) must, however, be treated with caution, as several other dates from the same basal level showed much younger ages, and the charcoal for the dates was, in some cases, from unsecure contexts.
More recent archaeological excavations in the Newcastle CBD described three phases of Aboriginal occupation at ~6.5 ka, ~3.5 ka and ~2.0 ka (AHMS, 2011). The artefact assemblage demonstrated the ongoing exploitation of locally derived lithic materials and the intensification of coastal exploitation in the more recent period of use (~50% of the assemblage coming from the upper 40 cm). Further inland, archaeological investigation of an elevated sand sheet in Tarro overlooking Hexham Swamp recovered ~4000 artefacts hinting at a much longer history of visitation and occupation (AHMS, 2017). The assemblage, constrained to the upper 1 m, suggest an initial visitation may have occurred as early as 33 ka, although the majority of the artefacts were recovered from a period dating to 13-2 ka and centred on ~7 ka when the swamp was initially formed. Interestingly, the site appears to be largely abandoned by 2 ka, and may suggest a shift in hunting and procurement strategies, coinciding with increasing archaeological sites on the nearby coast.
Therefore, despite considerable investigation of the region over the last 50 years, the presence of deep-time or stratified sites is still relatively sparse, with most current evidence indicating only a sporadic use of the region before the early Holocene, and then increasing focus on coastal environments in the last few thousand years.
The use of near-shore islands by Australian Indigenous people has been the focus of considerable research (e.g. Sullivan, 1982) and evidence suggests that they were first used in the mid-to-late-Holocene (e.g. Rowland et al., 2015 and references therein). The maximum water depth between Broughton Island and the mainland is approximately 15 m (Wright, 1975) and so this landscape only became a near coastal island during the marine transgression of the early Holocene (~9 ka) (Thom et al., 1981, 1992; Williams et al. 2018). The island is within sight of the mainland, particularly from Dark Point which forms an easy access point for visiting Broughton Island. Dark Point has been used by people for at least 4000 years (DEC, 2007).
There are several features which provide evidence for Aboriginal utilisation of Broughton Island (Patteson, 2003; Sullivan, 1982; Wright, 1975). Enright (1935) described an arc-shaped, stone-wall fish trap on Broughton Island and this was further described by Godwin (1988). While such a feature cannot typically be directly dated, the marine transgression at ~9 ka (Williams et al., 2018) gives a terminus post quem age for the trap. Wright (1975) described a midden adjacent to Grey Pole Swamp and this included flaked stone artefacts (mostly large scrapers), burnt stone, charcoal, shells (limpets, periwinkles and abalone), fish bones and scales, numerous bird bones and occasional mammal bones (seal and dog). Based on their contents, Sullivan (1982) concluded that Broughton Island was visited in spring and early summer to exploit birds and young seals. Wright (1975) also described the remains of two human burials in close proximity to this midden.
Wright (1975), Gillespie (1977) and Gillespie and Temple (1977) reported radiocarbon ages associated with this midden. A charcoal sample (SUA-402C) from the upper layer of the midden is described as ~440 uncalibrated years before present (bp) (440 ± 180 bp in Wright, 1975; 445 ± 170 bp in Gillespie and Temple, 1977; 445 ± 179 bp in Gillespie, 1977) and shells (SUA-402/S1, SUA-402/S2) provide a similar date to a maximum of 600 ± 85 bp (Gillespie and Temple, 1977). This suggests that the midden can be dated to ~405 cal. year BP (AD 1540 or 1550 using CALIB v7.10 (Stuiver et al., 2019) and SHcal13 (Hogg et al., 2013) and the median probability).
The European history of the region and Broughton Island
European excursion into the Myall region began with the northward migration of timber getters from Newcastle in 1816 (Garland and Wheeler, 1982). European settlement of the region dates to the late 1820s when a garrison was established at Soldiers Point and early historic houses in Port Stephens date to the 1830s (PSC, 2020). With an increasing European presence, the Aboriginal population rapidly declined and the traditional lifestyle of the Worimi people ended ~1830 (Garland and Wheeler, 1982). Despite the adversities of the historic period, the Worimi people are extant, and since 2007 traditional lands at Stockton Bight (Worimi State Conservation Area, Worimi National Park and Worimi Regional Park) maintain the cultural and natural heritage of the Worimi people (Sutherland, 2011).
The first European to distinguish the island itself was Captain William Broughton in 1795. Its environmental value was realised as early as 1862 when it was declared a water and wildlife reserve, a declaration that prevented proposed tourist resorts and fish canneries in the 1930s (NPWS, 2002). On the mainland timber milling, fishing, boat building small scale farming and sand mining characterised the land use of the region in the Twentieth Century. In 1972, the ecological, scientific and recreational values of the region were officially recognised with the formation of Myall Lakes National Park which included Broughton Island.
The first European settlement on Broughton Island was in the late 1800s when Italian fisherman constructed huts behind the dunes at Providence Beach (DEC, 2006). Around the time of World War I, Broughton Island was further colonised by Greek fishermen. A second settlement was constructed at Esmeralda Cove in the 1930s and the huts that remain there today have historic and social significance (EJE Group, 2000). Fisherman introduced goats to the island for milk and meat sometime in the pre-WWII period (DEC, 2006). During the 1960s the area between the Esmeralda Cove huts and Coal Shaft Bay was excavated for a mining exploration project and it was also used for military artillery activities (Dodkin, 1981).
There is no information available regarding the historic use of fire on Broughton Island before about the middle of the Twentieth Century. Anecdotal evidence suggests that from 1950 to 1990 fires were lit on an annual or near-annual basis to facilitate walking around the island (pers. comm., T. Wade, 30/03/2005). These human induced fires often covered much of the island. This purported annual burning regime ceased in 1990. There are conflicting reports about recent fires on Broughton Island: NPWS (2002) has recorded fires in 1985, 1992 and 1994, however regular visitors to the island claim that since 1990 there has been only one fire, in 1995, which burnt about half of the island (pers. comm., J. Gibb, 30/3/2005).
Previous local investigations
How humans may have influenced fire and vegetation on Broughton Island has been the subject of considerable speculation. Unlike Broughton Island, nearby equivalent environments, including islands and headlands (e.g. John Gould Island Nature Reserve on Cabbage Tree Island and Yacaaba Headland), support littoral rainforest and coastal vine thickets, a vegetation community that is Critically Endangered (e.g. under the Australian Federal Government Environment Protection and Biodiversity Conservation Act 1999 and Littoral Rainforest is an Endangered Ecological Community in the NSW North Coast, Sydney Basin and South East Corner Bioregions in the NSW Biodiversity Conservation Act 2016). Wright (1975: 19) speculatively linked the scarcity of trees and shrubs on Broughton Island to ‘periodic white settlement’. Dodkin (1981) hypothesised that the intermittent use of fire by Indigenous people on Broughton Island was responsible for the extant open vegetation and that patches of extant rainforest were ‘remnant’ in areas topographically protected from fire.
Comparison with palaeoenvironmental work on the adjacent mainland and regionally (Black and Mooney, 2007; Macphail, 1973; Mooney and Maltby, 2006; Mooney et al. 2007; Shimeld, 2004) may better describe the precolonial history/history of fire in this landscape and hence the environmental and human history of the region. In a study located on the adjacent mainland of MLNP Mooney and Maltby (2006) argued that the only discernible anthropogenic signal in the late-Holocene history of fire was in the late Twentieth Century. In contrast, fire in the precolonial period was rare, and periods of slightly increased evidence for fire were coincident with previously identified periods of climatic instability (Mooney and Maltby, 2006).
Methods
Field work was conducted in March and September 2005. Using a high tensile steel probe and coring equipment, several transects were completed across Grey Pole Swamp to determine the location of the deepest sediment accumulation. Once this was located a series of overlapping 50 cm cores were taken to a depth of 2.50 m using a Russian D-section corer (Jowsey, 1966). The cores were wrapped and refrigerated until analyses were conducted. A species list of the vegetation of Grey Pole Swamp and for the island (Thomas, 2006) was generated over various trips to the island.
In the laboratory the sediment cores from Grey Pole Swamp were photographed and described. Three (bulk carbon) sections (73–83, 140–150 and 225–235 cm) of the sediment were submitted for AMS 14C dating. Radiocarbon dates were calibrated with using Calib 7.10 (Stuiver et al., 2019) using ShCal13.4c (Hogg et al., 2013) datasets. All dates reported represent ‘calibrated years before present’ (cal. year BP).
To characterise the sediments accumulating in Grey Pole Swamp, bulk density and loss-on-ignition were determined on contiguous 1 cm samples along the entire core length, based on methods described by Bengtsson and Enell (1986). Bulk density was estimated by the mass of the volumetric samples after drying at 105oC for 24 hr and loss-on-ignition was determined from the mass lost after 4 hr in a muffle furnace of 550oC.
Charcoal was analysed using a modification of the wet-sieving method (Long et al., 1998; Mooney and Tinner, 2011). Contiguous volumetric subsamples were taken at 1 cm intervals and treated with 6% sodium hypochlorite (bleach) for 48 h. The subsamples were then gently washed through a 250 µm sieve and the collected material transferred to petri dishes, digitally photographed and analysed using image analysis software (Scion Image Beta 4.02 for Windows). Charcoal concentration was expressed as both a count (no./cm3) and as an area (mm2/cm3). Charcoal accumulation (CHAR) was calculated by dividing the charcoal concentration by the deposition time estimated from the age-depth relationship.
CHAR (based on counts, no./cm2/y) was also transformed (tCHAR) using methods similar to those developed by the Global Palaeofire Working Group (Power et al., 2010). The transformation involved three steps: (1) a minmax transformation (the minimum in the range was subtracted from each value and then it was divided by the range of the values) was used; (2) the minmax values were log transformed after the addition of a small positive integer to avoid problems with zero values; and (3) the data were then rescaled (using a base period of 0.2–1 cal. year BP) to yield z-scores. These transformations are used to convert individual charcoal records towards normality and to homogenise inter-site variance, allowing comparison between two or more sites.
Fire frequency for Grey Pole Swamp was calculated using the program CharAnalysis (Higuera et al., 2010). The input parameters used in CharAnalysis were as follows: CHAR was log transformed, resampled on the basis of an average deposition time of 22 years and the background component was defined using a Lowess smoothing function which was robust to outliers over a 600-year period. Fire events were identified using a locally derived threshold to determine the peaks component of CHAR – we used a 99% percentile cut-off derived from the local noise distribution determined using a Gaussian mixture model. Fire frequency was calculated over a 500-year smoothing period. This is a relatively conservative approach, with fire frequency based on peaks in CHAR that clearly exceed the long-term, background component.
Twenty-five pollen slides were prepared at 10 cm intervals down the sediment column using standard palynological techniques (Faegri and Iversen, 1975) consisting of an alkali treatment (10% NaOH at 100°C for 10 mins), sieving (125 µm), acetolysis and mounting in silicone oil. Where sand grains were present the residue was re-suspended in tertiary-butyl alcohol and the sand removed by settling and decanting. The alcohol washes were then repeated and samples re-mounted. Not all pollen was identified, and instead counts (n ⩾ 200) of palynomorphs were used to calculate several indices as described in Table 1.
The pollen indices from Grey Pole Swamp sediments used to examine the history of vegetation on Broughton Island.
Results
The vegetation of Broughton Island is dominated by Imperata cylindrica var. major (Blady Grass), Lomandra longifolia (Spiny-headed Mat Rush) and Pteridium esculentum (Bracken) with associated dense patches of Monotoca elliptica (Tree Broom-heath) and Leucopogon parviflorus (Coastal Bearded-heath) occupying the cliff-lines. A few scattered Melaleuca quinquenervia (Paperbark) and Leptospermum polygalifolium (Yellow Tea-tree) trees surround areas of lower topography where semi-permanent wetlands and swamps, like Grey Pole Swamp, occur. Common exotic species on Broughton Island include Crofton Weed (Ageratina adenophora), Prickly Pear (Opuntia stricta var. stricta) and Coastal Morning Glory (Ipomoea cairica).
At the time of field sampling (2005) Grey Pole Swamp was vegetated primarily with Jointed Twig rush (Baumea articulate) and Soft Twig Rush (Baumea rubiginosa). On the upper, drier end of the swamp (away from the site of sediment coring) Batswing Fern (Histiopteris incisa) and False Bracken (Culcita dubia) formed a dense colony. Prostrate Knotweed (Persicaria prostrata) was also very common on the western edge of the swamp. A number of species characteristic of a littoral rainforest (particularly, but not only, understorey vines, creepers etc) were found in the small catchment of Grey Pole Swamp, and this included Black Apple (Planchonella australis), Lilly Pilly (Acmena smithii), scrambling forest Clematis, Veiny Wilkiea (Wilkiea huegeliana), Native Guava (Rhodomyrtus psidioides), Rasp Fern (Doodia aspera), Whalebone Tree (Streblus brunonianus) and Sickle Fern (Pellaea falcate).
The sediment accumulating in Grey Pole Swamp consists mainly of sandy humus, with increasing clay and decreasing humus towards the base. The upper 35 cm was mostly fibrous plant and root material and the base of the analysed sequence was sand. Results from the radiocarbon dating are given in Table 2 and provide the basis for the age-depth model shown in Figure 2a. The close fit of the average to the dated intervals describes a near-linear sedimentation rate at the site (Figure 2a). This is described by the highly significant (r2 = 0.999) age-depth relationship:
This formula is used for all age calculations and suggests that the analysed core (250 cm) represents 5430 cal. year BP. It also suggests that the 1 cm increment of all analyses conducted herein represents an average temporal resolution of ~22 years (viz. the deposition time). This is longer than the potential return period of fire, meaning that charcoal peaks could potentially represent more than one fire event. Pinus pollen, an exotic genus, was found to a depth of 9 cm, equivalent to ~142 cal. year BP (1807) in our model, which is in reasonable agreement with the European arrival to the Port Stephens area.
The results of the radiocarbon dating of samples from Grey Pole Swamp, Broughton Island.
the calibration used Calib 7.10 (Stuvier et al., 2019) and shcal13.4c (Hogg et al., 2013).

The results of the Broughton Island palaeoenvironmental analyses, graphed against depth in the Grey Pole Swamp sediment core. (a) the age-depth relationship based on the radiocarbon analyses; (b) bulk density and loss-on-ignition of the sediments; (c) charcoal (>250 µm) accumulation (CHAR); and (d) transformed charcoal accumulation (tCHAR).
Bulk density and loss-on-ignition values for the core analysed from Grey Pole Swamp are shown in Figure 2b. Loss-on-ignition is generally high in the section below a depth of 150 cm, low between 140 and 50 cm, and high in the uppermost (recent) sediments. Bulk density is often the mirror image of loss-on-ignition, and this proves the case here.
The influx of charcoal (⩾250 µm) in the sampled core is quite dynamic, as shown in Figure 2c. Charcoal influx, expressed as a number and as an area, are in good agreement, with low amounts of CHAR in most of the samples. There are two very large ‘peaks’ in CHAR centred on ~70 (1480 cal. year BP) and ~185 cm (4005 cal. year BP), and charcoal increases towards the top of the core. Figure 2d depicts the transformed CHAR (tCHAR) for Grey Pole Swamp and depicts these trends in more detail.
In Figure 3, tCHAR is graphed against time (cal. year BP) and suggests five ‘phases’ on Broughton Island: (1) a period before 4150 cal. year BP when charcoal influx is low; (2) 4150 – 3840 cal. year BP when charcoal accumulation is high; (3) 3840 – 1750 cal. year BP when accumulation is again low; (4) 1750 – 1510 cal. year BP when accumulation is high; and (5) 1510 cal. year BP – present when charcoal accumulation is variable, but generally elevated. In the period after the first record of Pinus pollen tCHAR is high, but notably this recent maximum is exceeded several times during the last 1000 years, and especially during the phases of high charcoal accumulation centred on ~1600 and ~4000 cal. year BP.

Two measures of fire activity through time and the pollen indices as defined in Table 1. (a) transformed charcoal accumulation (tCHAR) and relative fire frequency; (b) the ratios of grass pollen and trees and shrubs to total pollen; (c) the ratios of total rainforest pollen and Blechnum to total pollen.
Figure 3 also shows the calculated fire frequency on Broughton Island through time. As noted in the Methods, this was based on a conservative approach to identifying fire events, and so trends, rather than absolute numbers are highlighted here. Fire frequency was relatively low in the period before about 4200 cal. year BP, was relatively high from about 4000 to 3700 cal. year BP, and again from 3000 to 2800 cal. year BP and in the last ~1000 years. The calculated fire frequency of the Twentieth Century is the highest for the entire 5430-year record.
The pollen indices are also depicted in Figure 3. The grass/total pollen index is generally stable until a dramatic increase in the last 600 or 700 years. The trees and shrubs/total pollen index is essentially the reverse of the openness index – while variable, there is a noticeable decline in the last ~1000 years. The rainforest/total pollen index is generally low and only exceeds 5% at ~4300, 3600 and 1400 cal. year BP. The Blechnum fern spore/total pollen index was extremely high in the periods from 3800 to 3200 cal. year BP and from 1200 to 600 cal. year BP.
Discussion
In the contemporary environment, fire is managed on Broughton Island with the objective of minimising the ecological impacts of inappropriate fire regimes (NPWS, 2002, 2014). The implementation of an appropriate fire regime is hindered by a lack of records with sufficient temporal depth, and a meagre understanding of the relationship between people, fire and vegetation change, which has not previously been considered beyond speculative suggestions regarding the impact of precolonial and more recent historic anthropogenic fire.
The distance between Broughton Island and the mainland (~3 km) means that it is unlikely any charcoal quantified in this analysis were derived from fire on the adjacent mainland as charcoal particles of this size (>250 µm) are not transported far from their source (e.g. Clark, 1988; Whitlock & Millspaugh, 1996; Higuera et al., 2005). Australian pollen (Dodson, 1983) and particularly pollen of rainforest genera (Kodela, 1990a, 1990b), tends to be poorly dispersed and hence strongly suggests the local presence of the taxa that produced it. Together, these suggest that the palaeoenvironmental information outlined here (Figures 2 and 3) represents the local Broughton Island environment and events. CHAR is likely to represent a complex spatial aggregation of fire across Broughton Island, with a bias towards fire events in the Grey Pole Swamp catchment (Hallett et al., 2003).
Fire is often described as a check to the distribution of littoral rainforest in eastern Australia and an inappropriate fire regime is an identified threat for this community. Despite these generalisations, this community is not particularly flammable compared with other vegetation types in eastern Australia (e.g. eucalypt-dominated vegetation communities), as the generally moist conditions do not support ignition or fire spread. Nonetheless, during extended periods of drought, conditions might allow for fire and this is especially true along ecotone edges. On Broughton Island, the proportion of rainforest pollen to the total vegetation of Broughton Island is relatively low for the majority of the last 5405 cal. year BP, however the representation of rainforest is notoriously low (Kodela, 1990a, 1990b). The pollen of rainforest taxa and other arboreal taxa appears to have been reduced by fire at about 4000 cal. year BP but recovered, strongly in the case of rainforest taxa, soon afterwards. Peaks in the rainforest/total pollen index at ~4300, 3600 and 1400 cal. year BP also all occurred during periods of relatively low CHAR and inferred fire frequency.
The pollen record also shows that the proportion of trees and shrubs has undergone a clear decline (below the range of variability) in the period from 1500 cal. year BP to the present, and this is commensurate with an increase in grass. This change becomes very clear in the last millennium and specifically in the last ~600 years when tree and shrub representation is extremely low. When these vegetation changes are compared to transformed charcoal influx (tCHAR) and to the calculated fire frequency (Figure 3) it is apparent that these vegetation changes occurred when fire frequency increased. This suggests that fire, together with exposure, has contributed to the nature of the vegetation of Broughton Island which is now dominated by treeless communities including grasslands, fern-lands and localised coastal shrubs. This mirrors ecological work on the adjacent mainland where fire is a major control of the plant communities (Fox, 1988; Fox and Fox, 1986; Hunter, 2000; Myerscough and Carolin, 1986).
In the recent historic period of the mid-to-late Twentieth Century fire was very frequent on Broughton Island, resulting from deliberate ignitions by visitors and semi-permanent fisher-people. Any conclusion regarding whether this recent fire regime has contributed to the further loss of rainforest and sclerophyll elements is constrained by the resolution of the palynology and tempered by the fact that the pollen spectrum of the surface sample is not particularly different from those of the last ~600 years. The very high CHAR and the inferred fire frequency in the recent sediments (e.g. above the first record of the exotic genus Pinus at a depth of 9 cm) suggests that they are accurately reflecting the recent fire regime. This is despite suggestions by Duffin et al. (2008) and Saiz et al. (2018) that macroscopic charcoal of this size (>250 µm) does not capture fire in grass-dominated vegetation communities, which dominates the majority of the Broughton Island vegetation during this time (and in the contemporary environment: Thomas, 2006). However, the palynology (Figure 3) clearly demonstrates that rainforest and sclerophyll communities were not lost as a result of fire in the historic period as suggested by Wright (1975), as this decline clearly predates the historic period.
This observation, that fire and vegetation are apparently linked on Broughton Island leads to an obvious question as to whether this was an anthropogenic-dominated fire regime, or otherwise. The precolonial Worimi people were clearly capable of exploiting this coastal environment. They constructed substantial (~4.5 m long) canoes from the bark of several trees, used large spoon-shaped paddles and fishing lines were utilised at the time of European settlement (Hunterlink, 2009). Broughton Island has several middens, human remains, and a fish trap that provide tangible evidence for Aboriginal utilisation of the island (Patterson, 2003; Sullivan, 1982; Wright, 1975) and hence the (>3 km) crossing from the mainland was not a barrier. All dated archaeological materials on the island are however less than ~600 years old which is consistent with the general pattern of coast use (e.g. Sim and Wallis, 2008), but with the absence of a systematic archaeological survey it is impossible to know if these young dates represent an initial use of the island.
It is often assumed that the application of fire to the landscape was a fundamental aspect of the Worimi people’s lifestyle (e.g. Hunter, 2000; NPWS, 2002). In 1818, John Oxley noted that ‘the natives were extremely numerous along this part of the coast. . .In the morning their fires are to be observed in every direction’ (Hunter, 2000). Beyond that the pre-contact culture and lifestyle of the Worimi people is only sparsely described and early ethnographies (e.g. Enright, 1932, 1936) make no reference to the use of fire in the landscape. The Garoowa Coastal Sea Country Report (Sutherland, 2011) also makes no mention of fire beyond cooking fires. It is hence feasible that Oxley’s observations describe domestic hearths rather than fire in the landscape. Benson and Redpath (1997) also raise the possibility that fire was used to alert other Aboriginal groups about the incursion of explorers.
The speculative suggestion by Dodkin (1981) that the use of fire by the Worimi people on Broughton Island was responsible for the extant open vegetation has become more emphatic through time: without any empirical data subsequent publications have linked the species composition and structural characteristics of the vegetation of Broughton Island to precolonial anthropogenic fire (DEC, 2006; EJE Group, 2000; Patteson, 2003). As an example, NPWS (2002: 34) argued that people “altered their natural environment” on Broughton Island with the use of fire.
One way of addressing this question is by comparing palaeoenvironmental data representing fire activity from different sites across the region, with the assumption that similarities are more likely to represent regional drivers of fire, such as climate or climatic variability as this is likely to better match that scale. Conversely, regional variation could perhaps better reflect the different cultural practices of people, given that different language groups and populations were operating in the landscape in the precolonial period.
Figure 4 compares the accumulation of charcoal (tCHAR, using a common base period) on Broughton Island through time with several comparable, published records. This shows several consistent trends, and these are most notable in the last 3000 years or so. The site closest to Broughton Island, Worrimi Swamp (Mooney and Maltby, 2006), is ~2800 cal. year BP long and hence does not cover the base of the Broughton Island sequence, nonetheless there is a remarkable concordance between the overlapping periods, especially in the last 1300 years. The peak in tCHAR at Broughton Island between ~1450 and 1700 cal. year BP is not a feature of the Worimi Swamp record but it is reflected in the Griffith Swamp (Mooney et al., 2007) record. Similarly, the peak in tCHAR at Broughton Island between ~3800 and 4100 cal. year BP is not apparent in the other regional records, and hence might, conceivably relate to local anthropogenic fire. Rainforest and arboreal taxa were reduced during this time but recovered soon after.

A comparison of charcoal accumulation through time at Grey Pole Swamp (Grey Pole Swamp) on Broughton Island and other published records of charcoal from sites in the region. In this figure charcoal accumulation is transformed (tCHAR) data using a common base period. The Broughton Island record is in black, and then the shade of grey indicates proximity to Grey Pole Swamp: Worimi Swamp (Mooney and Maltby, 2006) is 10.9 km in dark grey; Griffith Swamp (Mooney et al., 2007) is 114.5 km, in grey; Kings Waterhole Swamp (Black and Mooney, 2007), light grey, is 162.2 km away from Grey Pole Swamp.
As an alternative to anthropogenic fire, an increase in more severe fire weather is often associated in eastern Australia with ENSO (e.g. Dowdy, 2018) or other sources of climatic variability (e.g. Cai et al., 2009). Counts or colorimetric analysis of clastic layers in the Ecuadorian Laguna Pallcacocha, described by Rodbell et al. (1999) and Moy et al. (2002), are widely used as a proxy of moderate to strong El Niño events in the Holocene (cf. Schneider et al., 2018). There is a close correspondence between tCHAR on Broughton Island and the Moy et al. (2002) interpretation of (the number of) strong El Niño events at Laguna Pallcacocha (Figure 5). These records follow each other remarkably well, with exception of the peak in tCHAR at Broughton Island between ~3800 and 4100 cal. year BP, suggested above to potentially be anthropogenic, and in the Twentieth Century when fire on Broughton Island is also related to recent human visitation. This clearly suggests that the long-term trends in fire activity on the island are responding to some aspect of Pacific Ocean climatic variability, and this observation then is the more likely explanation for the regional similarities in fire activity (Figure 4).

A comparison of the transformed charcoal accumulation (tCHAR) through time at Grey Pole Swamp (Grey Pole Swamp) on Broughton Island and the record of strong El Niño events (as the number of events in non-overlapping 100 year windows) at Laguna Pallcacocha in Eucador (Moy et al., 2002).
The clear increase in fire activity on Broughton Island over the last 600 or 500 years coinciding with the appearance of tangible archaeology on Broughton Island at about the same time is an enticing link suggesting an anthropogenic fire regime after this time. There are however other suggestions of changes in hydroclimate at about that time: a reconstruction of the Southern Oscillation Index (SOI) by Yan et al. (2011) shows a distinct change from persistent negative values to positive values at AD 1400. Consistently more positive SOI is generally conducive to rainfall above the long-term median in eastern Australia, whereas previous discussion regarding climatic variability has focused on enhanced fire conditions during El Niño.
As suggested by Black et al. (2007), a complex nexus between climate and humans must also be entertained, and in this case a more favourable hydroclimate perhaps saw human populations increasing and/or expanding into previously marginal landscapes. It is interesting to note that the apparent increase in fire at Grey Pole Swamp in the last ~600 cal. year BP is also mirrored at a much wider spatial scale in the synthesis of fire in southern Australia (25–45oS) by Williams et al. (2015).
Equally tantalising is consideration of a ‘Polynesian-looking’ basalt adze blade that was found in 1928 at Dark Point, which is immediately adjacent to Broughton Island. Thorpe (1929: 123) described the original collection of this adze and thought it was “of undoubted Polynesian origin”. White et al. (2014) used pXRF analyses of the adze, other artefacts and Pacific basalts and concluded that it was most likely from a Polynesian voyage from Norfolk Island. The Polynesian settlement of Norfolk Island was apparently brief and ended about 600 years ago (Anderson and White, 2001) and Polynesian settlement is often associated with widespread alterations to fire (e.g. the Initial Burning Period following Polynesian settlement of New Zealand: McWethy et al., 2010; Perry et al., 2012). It is an interesting, albeit highly speculative, hypothesis that the Polynesian settlement on Norfolk Island occurs at the same time that fire increases significantly on Broughton Island and the two might be related.
The local Indigenous people, the Worimi, were dispossessed about 1830 and a European presence was not established on Broughton Island until late in the Nineteenth Century. Notably, during this gap in occupation there is no discernible decrease in charcoal accumulation or reconstructed fire frequency, and this could, perhaps, be used to argue that the previous precolonial fire regime was not anthropogenic. There is however evidence of a very high frequency of anthropogenic fire in the second half of the Twentieth Century, and this is reflected in the accumulation of charcoal and reconstructed fire frequency, even though this frequency is well below the resolution of this analysis.
Conclusions
There have been several studies in eastern Australia where the intermittent occupation of near-shore islands by precolonial people have been compared to more continuous mainland sites. As an example, Hope (1999) compared dune swale swamp sites on Hunter Island and the adjacent mainland Sandon Point on the west coast of Tasmania. Insight into environmental change, coastal people and their land use of near coastal islands have also come from Queensland, including from North Stradbroke Island and the Gulf of Carpentaria (e.g. Barr et al. 2017; Mackenzie et al. 2017; Moss et al., 2015; Tibby et al., 2017). This is perhaps not surprising given that islands have fascinated scientists for centuries.
This research describes a high resolution record of charcoal accumulation and other palaeoenvironmental proxies including those representing vegetation on Broughton Island, covering approximately 5430 cal. year BP. On the island, several changes in vegetation are clearly associated with changes in fire activity, and this is particularly the case for the reduction in arboreal and rainforest elements at ~4000 cal. year BP and the reduction in trees and shrubs and an increase in grass from ~1500 to 1000 cal. year BP.
On the island, the accumulation of charcoal closely follows proxies of El Niño through time suggesting that the fire regime is predominantly responding to some aspect of climatic variability centred on the Pacific Ocean. Similarities between the accumulation of charcoal at the Broughton Island site and similar records on the nearby mainland and in the region further supports the idea that fire represents climatic drivers, despite Indigenous populations being active along this coastline during this time. One exception to this is centred on about 4000 cal. year BP when fire events on Broughton Island were perhaps related to human activity. A clear increase in fire activity on Broughton Island over the last 600 or 500 years also coincides with dated archaeological materials on the island and more likely reflect an anthropogenic management of the landscape. Speculatively, this latter timing is concordant with the voyaging of Polynesian people in the region, and which may have played a role. These hypotheses can only be tested with additional systematic archaeological investigation of the island to further understand its past history of human occupation and use.
Footnotes
Acknowledgements
Dr Sean Ulm (University of Queensland) kindly supplied additional information on the archaeology of Broughton Island. Stuart Pearson, Lucy Gayler, Peter Shimeld and David Edwards generously helped with field work. Alison Thomas helped with the description of the vegetation used in this work and contributed to the drafting of Figure 1.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
