Phytoliths from Cloggs Cave, GunaiKurnai country (southeastern Australia) reveal the gathering of whole grasses for burning practices inside the cave

1 Introduction

Cloggs Cave, in GunaiKurnai Country, East Gippsland, southeastern Australia, contains an exceptional record of people–plant relations spanning at least the past 25,000 years, representing a thousand generations of Old Ancestors. The sedimentary sequence at Cloggs Cave contains high abundances of introduced (allochthonous) plant materials that have so far been studied through macro-plant remains, pollen analysis and micromorphology (David et al., 2024; Grono et al., 2024; Miller, 2022). These independent analyses found that people were the main drivers of plant material deposition within the cave, and that the study of this material provides information both on the activities of the people in the cave and of the cultural meaningfulness of the plants used at the site. Here, we continue these investigations by analysing the phytolith assemblage from the cave's stratified archaeological sediments.

1.1 Aims of the Cloggs Cave phytolith study

Following initial archaeological excavations at Cloggs Cave by Josephine Flood in 1971–1972 (Flood, 1973, 1980), new excavations were undertaken by the GunaiKurnai Land and Waters Aboriginal Corporation (David et al., 2021b; Mullett et al., 2021), representing the site's GunaiKurnai Traditional Owners and knowledge-keepers, in collaboration with university-based specialists. The aims of the new investigations were to reconstruct in finer detail than previously available how the cave was used by the Old Ancestors¹, utilising recent methodological developments in archaeology and in light of cultural knowledge recorded in nineteenth century ethnographic texts and held by present GunaiKurnai Elders (e.g., Howitt, 1887). The excavations clarified the chronostratigraphic sequence and established a continued cultural history of cave visitations from 27,390–24,760 cal BP through to the Late Holocene, and revealed a number of archaeological features characteristic of GunaiKurnai ritual installations and practices (David et al., 2021b, 2024; Mialanes et al., 2022; Mullett et al., 2021; Stephenson et al., 2020).

A focus of the renewed research at Cloggs Cave is to investigate the abundant plant material preserved in the stratified deposits. The study of plant remains provides valuable evidence on the collection and use of vegetation by people in the past, the nature of activities carried out by people inside the cave, and environmental conditions and change in the surrounding landscape over time. At Cloggs Cave, plant materials including ashes, charcoal and whole leaves were found through much of the excavated sequence and date from throughout most of the Holocene and extend back to the Late Pleistocene (for observations of their presence in the original 1970s excavation, see Flood, 1974, pp. 180–181).

A series of dense stratified ash layers relating to the repeated in situ burning of thin spreads of vegetation across the cave's palaeo-surfaces date from c. 4,400 to 1,460 cal BP (David et al., 2021b; Stephenson et al., 2020). Around 2,000 cal BP, a 28.1 cm-high standing stone was erected in the cave, with vegetation repeatedly accumulated and burned around the standing stone over a period of some 400 years (David et al., 2021b). The 73 well-differentiated but thin and mostly ashy layers at the base of and around the standing stone are all largely devoid of charcoal, suggesting that the materials burnt consisted of soft plant matter (e.g., grass, leaves and bark) (David et al., 2021b). This was corroborated by the micromorphological analyses that identified charred plant structures and phytoliths confirming the burning of soft plant material (Miller, 2022).

Deeper in the deposit, at levels dated to c. 11,000 and 12,000 cal BP respectively, two ritual installations each consisting of a well-defined miniature fireplace incorporating well-preserved trimmed Casuarina cunninghamiana or C. glauca sticks smeared with animal or human fat were revealed (David et al., 2024). Their structure and contents—including the smearing of the trimmed sticks with the visually invisible fat—closely match GunaiKurnai ritual installations described in nineteenth century ethnography (Howitt, 1887).

Pollen analysis revealed that the stratified deposits in the cave contained a suite of flowering plants known through GunaiKurnai knowledge and nineteenth century ethnography to have special cultural and medicinal uses (Grono et al., 2024). Clumps of single-taxon pollen that had not dehisced were observed, indicating in situ deposition of flowering plant material and suggesting that people carried plant material into the cave (Grono et al., 2024).

Taken together, the archaeological features and multiple lines of archaeobotanical evidence at Cloggs Cave corroborates traditional GunaiKurnai knowledge and nineteenth century ethnography of ritual practices involving the use of plant resources, indicating that the cave was a special, secluded place used by mulla-mullung, powerful learned men and women, for the practice of magic, healing, cursing and spiritual practices.

Pollen and micromorphological analyses undertaken at Cloggs Cave independently observed a high abundance of phytoliths on the pollen slides and micromorphological thin sections, where they were associated with plant material transported into the cave by people (Grono et al., 2024; Miller, 2022). While the pollen and micromorphological analyses enabled observation of phytoliths, these techniques are limited in their ability to adequately study the phytolith assemblage. While micromorphology can characterise the contextual arrangements and depositional processes of phytoliths, phytoliths are often obscured by sediments and are unable to be rotated to observe three-dimensional bodies, preventing morphotype identification. Meanwhile, pollen analysis is ideal for vegetation reconstruction of a wide range of plants, especially woody and herbaceous plants. However, it is unable to discriminate the taxonomy of grasses, which are high phytolith producers. Furthermore, pollen does not preserve when subjected to temperatures exceeding 350 °C (Ghosh et al., 2006; Sengupta, 1975), and is thus unsuited to studying burnt contexts at archaeological sites.

Phytoliths are microscopic, silt-sized silica bodies that form when plants absorb soluble silica from the soil and deposit it within their structures. Due to the biological silica composition of phytoliths, they often preserve well in soils, sediments and archaeological sites. In most environments, phytoliths are not prone to long distance transport, and are generally considered a highly localised element of plant decay (Piperno, 2006). Biogenic silica precipitates in different parts of a plant's structure, enabling a degree of correlation between phytolith morphotype and anatomical part of the plant represented. Phytoliths are often found articulated in their original, in situ botanical structure following the decay of associated organic matter, whereas isolated phytoliths can reflect the breakdown of plant structures and liberation of single cells due to reworking and other processes. When subjected to heating, phytoliths preserve evidence of alteration, ranging from colour changes to morphological warping and melting with increasing temperatures (Fritzsch and Röpke, 2025; Wu et al., 2012). When found at archaeological sites, phytoliths can be a useful proxy for cultural activities involving the in situ deposition of plant material by people (e.g., Albert et al., 2008; Cabanes et al. 2012; Esteban et al., 2023; Regev et al., 2015; Shillito, 2011; Tsartsidou et al., 2009; Wadley et al., 2011). These characteristics of phytoliths make them an ideal proxy to further investigate the evidence of in situ plant deposition at Cloggs Cave. We undertook phytolith analysis to try to: A complicating factor about the Cloggs Cave phytolith-rich sediments is that they not only hold much evidence of human activities, including stone tools, standing stones, ritual fireplaces and their associated wooden objects, but they also contain large quantities of whole and largely intact Common Brushtail Possum (Trichosurus vulpecula) scats. To ensure that we studied phytoliths from plants introduced to the site by people, we needed to identify and differentiate phytoliths introduced by possums through their scats. We undertook two additional analyses towards this aim. First, we analysed the phytolith content of individual whole Common Brushtail Possum scats excavated from the site, to provide a ‘fingerprint’ of phytoliths introduced into the cave by the possums (for an overview of their herbivory, see Grono et al., 2024, pp. 5–6). The scats were recovered from the stratified deposits during excavation and indicate that possums repeatedly frequented the cave. Second, ash pseudomorphs were analysed to assess the representation of woody elements and indications of anthropogenic burning. Ash pseudomorphs are calcium oxalate crystals abundant in trees and shrubs that transform to calcium carbonate when heated at temperatures above 450–500 °C (Brochier and Thinon, 2003; Canti, 2003; Gur-Arieh and Shahack-Gross, 2020). As woody taxa are low phytolith producers, this analysis helps quantify the effects of differential phytolith production by the woody vs. non-woody plant taxa.

• Determine what kinds of plants had been deposited in the ashy layers where pollen had not survived;
• Record which plant parts had been deposited, to provide a complimentary line of evidence of people–plant relations in the cave.
• Investigate the processes of transport and deposition of the microbotanical remains in the cave, to better understand what had been brought in by the Old Ancestors.

2 Site setting

Cloggs Cave is a domed limestone cave whose entrance lies 72 m above sea level in the Buchan karst in the southern foothills of the Australian Alps (Figure 1). The cave opening is located 17 m above the valley floor on the southern floodplain of the Buchan River, c. 4 km northwest of the confluence of the Buchan and Snowy Rivers. The region has a temperate climate and environmental conditions inside the cave are dry and cool (c.15 °C) year-round, with little seasonal variation. The neutral to slightly alkaline pH (pH values of 7–8) of the sediments provide excellent preservation conditions for organic materials and their remains including phytoliths, which preserve well above acidic conditions of pH = 3 and below alkaline conditions of pH = 8 (Cabanes et al., 2011).

The vegetation surrounding Cloggs Cave today is mainly farmland. Pre-European settlement vegetation along the Buchan River valley bottom consisted of Riparian Forests with moist soils, tall eucalypts (e.g., Eucalyptus globoidea, E. viminalis), wattles (e.g., Acacia dealbata, A. melanoxylon), ferns (e.g., Calochlaena dubia, Cyathea australis) and grasses (e.g., Carex breviculmis, Microlaena stipoides var. stipoides) [Ecological Vegetation Class (EVC) 18 of the Victorian government's vegetation classification categories]; and open Valley Grassy Forest with tall eucalypts (e.g., Eucalyptus globoidea, E. viminalis), acacias (e.g., Acacia melanoxylon, A. mearnsii) and grasses and sedges (e.g., Themeda triandra, Poa australis, Carex spp.) (EVC 47) (DEECA, 2025). On the valley sides and crests surrounding Cloggs Cave, the vegetation was an open Grassy Woodland, with tall eucalypts (e.g., Eucalyptus melliodora, E. viminalis) and a diverse ground layer of herbs (e.g., Hypericum gramineum) and grasses (e.g., Poa labillardierei), including the native cereal, kangaroo grass (Themeda triandra), still growing abundantly on the valley sides (EVC 175) (DEECA, 2025).

Cloggs Cave measures 12 m deep, 7 m wide and 5 m high, and has two outside areas (termed the ‘Porch’ and ‘Rockshelter’) before a narrow (0.85 m-wide) and 5 m-long, curving Entrance Passage that ends with a 1.2 m drop to the cave floor of the Main Chamber. Inside the cave, where both Flood's 1970s and David et al.'s 2019–2020 archaeological excavations took place, the Main Chamber is 5 m long, 3.5 m wide and 5 m high. Three 50 × 50 cm squares were excavated in 2019–2020, representing two separate chronostratigraphic sequences. All the squares were excavated against the cleaned walls of the original 2 × 2 m, still-open pit of Flood's 1971–1972 excavation. In 2019–2020, contiguous squares P34 and P35 were excavated against the southeast wall of Flood's pit, and R31 against its northeast wall. R31 was positioned 2–2.5 m to the north of P34–P35. Both the 1971–1972 and 2019–2020 excavations were positioned near the middle of the Main Chamber, some 2 m from the inner end of the Entrance Passage (Figure 2).

2.1 Stratigraphy and chronostratigraphy

The archaeological excavations in both sets of squares from the 2019–2020 excavations were undertaken in excavation units (XUs) that sub-divided the stratigraphy (stratigraphic units, SUs) into finer units. It is fundamental in archaeology to excavate in thin levels that are finer than the sediment strata and that follow their contours and boundaries, so that the internal as well as between-SU (including stratigraphic interfaces) depositional and taphonomic history of the strata can be investigated, as well as to analyse buried materials in thinner and shorter-term units than stratigraphic units alone would permit. The P34–P35 sequence was dated through 40 accelerator mass spectrometry (AMS) radiocarbon ages, and R31 through 69 AMS radiocarbon ages, each on a single piece of charcoal, wood, bark or possum scat (for the radiocarbon results, see David et al., 2021a: Table 1; 2024 Stephenson et al., 2020);. Additionally, 10 single-grain optically stimulated luminescence dates were obtained from the walls of Flood's exposed pit (David et al., 2021a: Table 1) (for the Bayesian modelling of R31, see David et al., 2021a: Supplementary Tables 2, 3). The major SUs across both pits were standardised (e.g., SU2 in P34–P35 is the same layer as SU2 in R31), but the sub-SUs were not (because of the complex and discontinuous inter-leafing of individual sub-layers and lenses in each square; e.g., SU2A in P34–P35 is not the same sub-layer as SU2A in R31).

Table 1. Phytolith concentrations of sediment samples and possum scat samples, and ash pseudomorph concentrations of sediment samples from P35 and R31, Cloggs Cave. Samples are listed according to Stratigraphic Unit (SU) and depth (for P35)/Excavation Unit (XU for R31) and are accompanied by associated archaeological evidence and radiocarbon dates recovered from within the respective excavated units (David et al., 2021a, 2021b, 2024).

P34–P35 were excavated in 130 XUs of average 1.8 cm thickness, down to a depth of 228 cm below surface. The squares were positioned over what was later identified as an infilled sinkhole with a post-sinkhole capping of sediment layers. The lowermost 170 cm (SU3B–SU3G, excavated as XU52–XU130) of the excavation had accumulated rapidly through torrential downslope collapse just after the sinkhole formed c. 6,000 cal BP. Over the infilled sinkhole, c. 20 cm of sediment (SU3A, excavated as XU31–XU55, with XU52–XU55 being the SU3A–SU3B interface) continued to accumulate between c. 6,000 and 4,400 cal BP. SU2 (excavated as XU31–XU5) then gradually accumulated in a c. 40 cm-thick series of 73 mostly fine ashy (‘hearth’) anthropogenic layers between 4,400 and 1,600 cal BP, particularly in the last 400 years of that period (David et al., 2021b). These mostly ashy layers then ceased, with the upper 10 cm of the sequence consisting of archaeologically sterile sandy loam (SU1, XU5–XU1, with XU5 being the SU1–SU2 interface) that accumulated over the past c. 150 years.

R31 was excavated in average 2.3 cm-thick XUs, down to XU56 at the SU5A–SU5B interface, at a depth of 153 cm. XU51–XU55, at 119–144 cm depth at the interface of SU4V–SU5A, consist of a mixture of fine and rocky deposits intermixed by roof-fall, with XU44–XU50 (up to the SU4U–SU4 V interface) having minor age reversals, but the overlying sediments (XU1–XU43, in SU2–SU4U) are made up of a sequence of 28 thin and well-defined layers dated through 57 radiocarbon ages that are in excellent chronostratigraphic order.

The lowermost stratigraphic unit (SU5) began to accumulate sometime before 46,580–43,230 cal BP and contains extinct megafauna remains (David et al., 2021a). SU5 is overlain by SU4, of which the lower part (XU43 and below) incorporates rockfall beginning shortly before 25,740–24,760 cal BP resulting in sediment mixing and corresponding radiocarbon age reversals. The upper part of SU4 (XU42 and above) has no such evidence of rockfall and good chronostratigraphic integrity (David et al., 2021a). Some of the upper SU4 sub-layers are small ashy lenses from fireplaces. SU4 is overlain by SU2, as SU3, the infilled sinkhole over which P35 was positioned, is absent in R31. SU2 is a c. 15 cm-thick deposit consisting of seven ash lenses from anthropogenic fires deposited sometime between 5,000 and 2,000 cal BP (David et al., 2021b). Capping the sequence, the uppermost 20 cm of deposit consisting of apparently archaeologically sterile sediment are absent, because they had been removed by Flood in 1971–1972. Flood had then covered the exposed surface of the underlying ashy deposit with thick plastic sheeting weighed down with some 20 cm of fine-grained sediment to protect the underlying deposits. These protected deposits were in excellent condition when the surface sediments and plastic sheeting were removed to permit excavation of R31 in 2019–2020 (for further descriptions, see David et al., 2021b, p. 7).

2.2 Sampling for phytolith analysis

Our phytolith study focuses on the same sediment samples from P35 and R31 as those analysed in the palynological study (Grono et al., 2024). The P35 samples for phytolith analysis span SU1 and SU2 and were removed by pushing 1.5 cm-diameter plastic tubes into the exposed wall of the pit at 5 cm intervals from 0.5 cm to 40 cm depth, from 10 to 18 cm southwest of the northwestern edge of P35, their collection locations being contiguous with the P35 chronostratigraphy (Figure 3). The P35 samples are referred to according to their depths in centimetres. The archaeological deposits of P35 sampled for phytolith analysis sit above SU3, the localised sinkhole that formed c. 6,000 cal BP and subsequently infilled with mixed, collapsed sediments between c. 6,000–4,400 cal BP (Delannoy et al., 2020). The lower part of P35 and P34 are thus not analysed in this study.

The R31 samples for phytolith analysis were taken from XUs starting at XU1 at the top to XU43 at the SU4U–SU4V interface (Figure 3), and does not include the deeper and older mixed sediments (XU44–XU56 from SU4V and SU5) with age reversals caused by rockfall. The R31 samples are referred to according to their respective XUs.

The P35 samples and the top sample of R31 (XU2) overlap in age, dating to after 5,000 cal BP, whereas there is no overlap in age between the P35 samples and the deeper R31 sequence dating back to c. 25,000 cal BP (XU6–XU43) (Table 1). Together, the P35 sequence dating the Mid to Late Holocene and the R31 sequence representing the terminal Pleistocene to Late Holocene represent a long and continuous cultural sedimentary record extending from the Last Glacial Maximum (locally spanning 28,600 ± 2,800 BP to 17,700 ± 2,200 BP: Cadd et al., 2021) to the Late Holocene. The oldest definitive archaeological evidence from a secure chronostratigraphic context at Cloggs Cave are stone artefacts from XU45 (SU4 V) dating to sometime between 27,390 and 24,760 cal BP (Mialanes et al., 2022). Human activities in the excavated part of the cave persist until c.1460 cal BP (David et al., 2021b).

The stratigraphy, uncalibrated radiocarbon ages and locations of the P35 and R31 phytolith samples of each square are illustrated on Figure 3 and summarised in Table 1.

3 Methodology

We analysed phytolith concentrations from nine sediment samples in P35 and 24 in R31. Phytolith extraction and quantification followed Katz et al. (2010). Approximately 50 mg of dry sediment per sample were weighed using an analytical balance, pre-treated with 50 µL of 6 N HCl to remove carbonates, and subjected to a heavy density separation using 450 µL of a sodium polytungstate (SPT) solution with distilled water calibrated to 2.4 specific gravity (g/mL). Particles were disaggregated using sonication for 10 min and the heavy and light fractions were separated by centrifugation at 5,000 rpm for 5 min. For microscopic examination, 50 µL per sample was pipetted onto a glass slide and phytoliths were counted in 30 random fields of view at ×400 magnification under an Olympus CX31 optical polarising microscope. Phytolith concentrations are calculated based on slide area counted, total volume of solution and initial weight of the sample using the formula in Katz et al. (2010). Phytoliths are reported in millions per 1 g of sediment, with an assumed general error of 30% that takes into account both counting (e.g., misidentification, fragments, multicellular structures) and analytical (e.g., differences between counting fields arising from uneven distributions of phytoliths on a slide) errors (Katz et al., 2010).

Following the quantification of phytolith concentrations, phytolith morphotype analysis was undertaken on four samples from P35 and 13 samples from R31 to identify plant taxonomy and plant anatomical parts. Samples with high phytolith abundances determined in the preceding concentrations procedure were selected at regular depth intervals from P35 and R31 to provide comprehensive stratigraphic representations across the two excavation squares. Following the extraction procedure for phytolith concentrations outlined above, the samples were subjected to multiple washes in distilled water and mounted on microscope slides in glycerol for morphotype analysis. The generation of a phytolith reference collection of modern vegetation on GunaiKurnai Country was beyond the scope of this study, so morphotype identifications were made through published modern phytolith reference resources for Australian plants (Bowdery, 1998; Boyd et al., 2024b, 2024a; Hart, 1988, 1990; Parr and Watson, 2007; Wallis, 2000, 2003, 2013) and international phytolith literature (Albert and Weiner, 2001; Neumann et al., 2019; Piperno, 2006; Rapp and Mulholland, 1992). Terminological descriptions of morphotypes follow the International Code for Phytolith Nomenclature (ICPN) 2.0 (Neumann et al., 2019). Systematic counts of a minimum of 200 single-cell phytoliths were made for each sediment sample (Albert and Weiner, 2001; Piperno, 2006), and morphotypes within multicell forms were added to count totals after the counts were completed.

Nine sediment samples from each of P35 and R31 were also analysed for ash pseudomorph concentrations following the methodology of Gur-Arieh et al. (2013). Following sieving through a 150 µm sieve to eliminate grains larger than ash pseudomorphs, the ash pseudomorph extraction procedure is similar to the above protocol outlined for phytoliths, omitting pretreatment of HCl and centrifugation as these steps are destructive to the fragile calcitic wood ash pseudomorphs (Gur-Arieh et al., 2013). Ash pseudomorphs were counted in 30 random fields of view per sample at ×400 magnification under plane-polarized light and cross-polarized light using an Olympus CX31 optical polarizing microscope. Concentrations are calculated using the formula outlined in Gur-Arieh et al. (2013) and the results are reported in thousands per 1 g of sediment, with an assumed general counting and analytical error of 30% (Gur-Arieh et al., 2013).

In addition to the sediment samples, phytolith concentrations were also analysed from four whole Common Brushtail Possum scats excavated from P35 and nine from R31, to assess whether possums introduced phytoliths into the deposits through their scats. One possum scat from P35 and four scats from R31 containing the highest phytolith concentrations were then analysed for phytolith morphotypes. The same analytical methods were used for the scats as for the sediment samples.

4 Results

4.1 Phytolith concentrations

The P35 sediment samples contain generally low abundances of phytoliths, ranging from 0.1 to 0.6 million phytoliths per 1 g of sediment (Figure 4 and Table 1). However, this is not so for R31, which exhibits fluctuating levels of phytoliths signalling a clear relationship to stratigraphic depth. The top R31 sample (XU2) overlaps in age with the samples from P35 and share similar low phytolith abundances. With increasing depth in R31, phytolith abundances gradually increase from 0.2 to 11.4 million phytoliths per 1 g of sediment between XU2 and XU20, peaking at 13.1 million phytoliths per 1 g of sediment in XU21, before gradually decreasing from 10.6 to 1.8 million phytoliths per 1 g of sediment between XU22 and XU43 (Figure 4 and Table 1).

The possum scat phytolith concentrations vary unevenly through the P35 and R31 sequences, with peak concentrations higher than any seen in the sediment samples. The P35 scat samples contain 0.6 to 4.6 million phytoliths per 1 g of faecal material, peaking in the scat from XU4 (Figure 5 and Table 1). P35 scat samples XU2, XU3 and XU4 date to the nineteenth to twentieth century and are equivalent in age to the P35 sediment samples 0.5 cm, 5 cm and 10 cm.

Phytoliths are even more abundant in the scats from R31. Most of the R31 scats have between 3.0 and 8.5 million phytoliths per 1 g of faecal material, with the scats from XU4 and XU21 having extremely high concentrations of 19.2 and 20.6 million phytoliths per 1 g respectively (Figure 5 and Table 1).

4.2 Phytolith morphotypes

The phytolith morphotypes identified from the Cloggs Cave sediment samples and possum scats are listed in Table 2. Representative photomicrographs of the major types are presented in Figure 6. The proportions of morphotypes in each sediment sample and possum scat are shown in Figure 7 (see Supplementary File 1 for the raw counts). Overall, excellent preservation conditions are indicated by the high phytolith abundances, the presence of delicate morphologies (e.g., Acute bulbosus, decorated margins of Elongate types, Papillate, Stomata) and multicellular silicified structures, and minimal observations of weathering. In only two of the sediment samples (P35: 35 cm; R31: XU7), the count target of 200 single cells could not be reached (Supplementary File 1).

Table 2. Phytolith morphotypes identified in the Cloggs Cave samples and their taxonomic and plant anatomical attributions.

Nomenclature follows ICPN 2.0 guidelines (Neumann et al., 2019) and taxonomic and anatomical derivations follow published literature and modern reference studies (Twiss et al., 1969; Hart, 1990; Bozarth, 1992; Ollendorf, 1992; Rosen, 1992; Ball et al., 1999; Albert and Weiner, 2001; Albert et al., 2003; Wallis, 2003; Strömberg, 2004; Piperno, 2006; Neumann et al., 2009, 2019: Supplementary Data File 1; Watling and Iriarte, 2013; Ramsey et al., 2016; Wroth et al., 2019; Ge et al., 2020a, 2020b; Pearce and Ball, 2020; Piperno and McMichael, 2020; Liu et al., 2021; An and Xie, 2022; Chowdhary et al., 2023; Boyd et al., 2024b, 2024a).

4.2.1 Sediment samples

The phytolith assemblage from the sediment samples is relatively homogenous, with slight variations in the relative proportions of morphotypes through and between the P35 and R31 sequences (Figure 7). The dominant class of phytoliths are Grass Silica Short Cell Phytoliths (GSSCP), accounting for 22.6–54.2% of morphotypes in any sample (Figure 7). GSSCP are unequivocal diagnostic indicators of Poaceae (grass family) where they are exclusively produced in the epidermis, most abundantly in the leaves and inflorescences (Table 2). Other common morphotypes are Acute bulbosus (4.6–23.7% of all morphotypes in any sample), Blocky (0.6–12.3% of all morphotypes in any sample) and Elongate entire (9.4–26.0% of all morphotypes in any sample) (Figure 7). These types have limited taxonomic utility due to their occurrence in many plant families and are therefore poorly diagnostic or non-diagnostic', although we note that they are most commonly produced in the leaves and culms of grasses (Table 2). Acute bulbosus represents silicified epidermal hair cells or trichomes abundant in the leaves, inflorescences and, less often, the culms of grasses (Neumann et al., 2019: Supplementary Data File 1), but are not exclusive to grasses as they also occur in woody and herbaceous taxa (Piperno, 2006; Gao et al., 2018; Neumann et al., 2019: Supplementary Data File 1; An and Xie, 2022; Boyd et al., 2024b). Blocky (0.6–12.3% in any sample) phytoliths often form bulliform cells in the leaves of grasses but can also be produced in other plant families, thus their overall diagnostic value is low (Neumann et al., 2019: Supplementary Data File 1). Single and multicell stacked forms (Figures 6I,J) of Bulliform flabellate (0.3%–5.5% in any sample) morphotypes relate to bulliform cells in the leaves of Poaceae and Cyperaceae (Neumann et al., 2019: Supplementary Data File 1). Elongate entire phytoliths are produced in many plant families including woody plants (Boyd et al., 2024a; Gao et al., 2018; Ge et al., 2020b; Liu et al., 2021), but are most common in the stems/culms of grasses and are generally used as an indicator of such (Ramsey et al., 2016). Diagnostic morphotypes of grass inflorescences (single cell and occasional multi-cell Elongate dentate/dendritic and Papillate phytoliths) are present in low proportions (≤2.3% dendritic and ≤1.7% Papillate in any sample) (Figure 7). While Acute bulbosus, Elongate dentate (≤7.0% in any sample) and GSSCP can all occur in the inflorescences, they can also form in other parts of the grass plant (Table 2) (Rosen, 1992; Neumann et al., 2019: Supplementary Data File 1; Ge et al., 2020b) and for this reason are not explicitly designated to inflorescences. Elongate sinuate phytoliths (0.7%–6.5% in any sample) (Figure 7) are another common margin type of elongate forms that are predominantly produced in the leaf epidermis of grasses (Table 2).

A range of morphotypes that are more commonly produced in woody plants and traditionally considered as indicators of woody taxa are also present in the assemblage. Conventional indicators of the leaves of woody plants include epidermal tissue [Epidermal tabular polyhedral and Epidermal tabular sinuate (Jigsaw)], Mesophyll and Tracheary annulate/helical (accounting for ≤3.3%, ≤0.7% and ≤3.2% in any sample, respectively) (Figure 7). However, these morphotypes can also occur in grasses and herbaceous taxa such as Asteraceae and have low diagnostic taxonomic value (An and Xie, 2022; Pearce and Ball, 2020; Watling and Iriarte, 2013) (Table 2). Opaque perforated platelets (Figure 6Z), diagnostic of the inflorescences of Asteraceae (daisy) (Bozarth, 1992; Piperno, 2006), were provisionally identified in low quantities (<2.3% in any sample) throughout the sequence (Figure 7 and Table 2). Stomata (accounting for ≤1.8% in any sample) (Figure 7) occur in woody, herbaceous and grass plants and have limited taxonomic value but some anatomical significance, occurring in the epidermis of leaves (Table 2). Sclereids (≤2.3% in any sample) and Irregular elongate (≤3.1% in any sample) (Figure 7) types are produced in sclerenchyma of woody plants, especially in the bark, fruits, leaves and seeds (An and Xie, 2022; Boyd et al., 2024b; Watling and Iriarte, 2013) (Table 2).

Articulated phytoliths, otherwise referred to as multicells or ’silica skeletons’, represent multicellular or anatomically connected phytoliths preserved in their original anatomical positions (Figure 8). High proportions of articulated phytoliths in the Cloggs Cave assemblage (up to 35.1% of total phytoliths in any sample) suggest in situ deposition of plant material and low taphonomic reworking of the deposits. Articulated phytoliths provide additional diagnostic information about the anatomical part of the plant represented. Most multicell forms observed in the sediment samples are articulated Elongate cells (mainly with entire, sinuate or dentate margins), or Elongate entire/sinuate phytoliths in association with GSSCP, regular rows of Stomata and Acute bulbosus phytoliths along the surfaces of such structures (Figures 8A–E, 9E,F). The association and arrangement of these morphotypes are diagnostic of the epidermis of grass leaves and culms. In particular, chains of Bilobates in association with Elongate entire/sinuate cells are very common, indicating the leaves of Panicoid grasses (Figures 8D,E). While most multicellular structures appear to derive from the leaves, rare articulated Elongate dendritic/dentate and Papillate (Figures 8F,G, 9C,D) also point to the presence of grass inflorescences.

GSSCP morphologies can be differentiated into grass subfamilies that in turn can be used as environmental proxies (Twiss et al., 1969). The dominant grass subfamilies in the Cloggs Cave sediment samples are Pooideae, the major C3 grass family whose species commonly grow in temperate, cool climatic conditions, and Panicoideae, C4 grasses of tropical to sub-tropical origin that are generally associated with warm, wet and humid climate environments (Piperno, 2006; Twiss et al., 1969). Pooideae, represented by Rondel as well as Trapeziform and Crenate morphotypes (32.0–58.8% of all the short cells) and Panicoidaeae, represented by Bilobate as well as occasional Cross and Polylobate morphotypes (33.3–62.4% of all the short cells), are generally found in similar proportions. A slight increase in Panicoideae relative to Pooideae is detected in the middle of the R31 sequence (XU11B–XU26), while Pooideae dominates most of the P35 samples as well as the upper (XU4–XU7) and lower (XU30–XU42) parts of the R31 sequence (Figures 7, 10). Chloridoideae, a subfamily of drought-resistant C4 grasses that grow well in warm, dry arid to semi-arid climates (Alexandré et al., 1997; Twiss et al., 1969), are represented by Saddle morphotypes that contribute a minor element (≤9.2% of all the short cells, Figures 7, 10) in the P35 and R31 sequences.

Single cells of Hat-shaped (also referred to as cone-shaped/conical) phytoliths are occasionally present (≤1.4% in any sample), although no contiguous articulated lines of Hat-shaped phytoliths were detected in the sediment samples. The Hat-shaped phytolith morphotype is identified as deriving from Cyperaceae (sedges) based on worldwide designations (Ollendorf, 1992), although Australian phytolith analysts have described similar Hat-shaped types in non-Cyperaceae taxa including Acacia schinoides, Banksia oblongifolia. and Allocasuarina distyla (Hart, 1990). For this reason, the Hat-shaped morphotype is tentatively assigned to Cyperaceae in this study. In addition to phytoliths, sponge spicules and diatoms are also present, albeit in low numbers (Supplementary File 1).

In all the sediment samples, a proportion of the phytolith assemblage, including single and multicell forms, appear blackened from burning (Figure 9), and in some samples melted phytoliths were occasionally observed (Figure 9H). The highest proportion of burnt phytoliths is in XU11B, in the upper part of the R31 sequence, where 18.3% of all phytoliths are burnt (Figure 7). Other sediment samples from the upper part of the R31 sequence, and some from P35, also have elevated amounts of burnt phytoliths (P35 25 cm: 3.5%; P35 35 cm: 10.9%; R31 XU4: 12.8%; R31 XU7: 3.9% of all phytoliths), suggesting an intense period of burning of the plant parts that contained the phytoliths at this time (Figure 7). In the middle and lower parts of the R31 stratigraphic sequence (XU14–XU42) burning continued, albeit at lower frequencies, based on the sustained detection of burnt phytoliths (0.4% to 5.1% of all phytoliths in any sample) (Figure 7).

4.2.2 Possum scat samples

The same morphotypes that dominate the sediment samples also dominate the possum scats, namely GSSCP (16.2–66.5% in any sample), Acute bulbosus (4.1–51.4% in any sample), Elongate forms (mainly Elongate entire, 6.6–19.8%; Elongate dentate, 0.9%–5.6%; and Elongate sinuate, ≤7.8% in any sample), Blocky (1.9%–6.5% in any sample) and Bulliform flabellate (≤4.6% in any sample) (Figure 7). In contrast to the sediment samples, the possum scats generally appear to be dominated by a single morphotype whose frequency varies between samples. For example, in R31, Bilobate accounts for 39.4% of all phytoliths in XU4, while Acute bulbosus comprises 37.0% of phytoliths in XU21 and 51.4% in XU31 (Figure 7). Indeterminate phytoliths more commonly produced in woody taxa are present in relatively low abundances (Elongate irregular ≤3.0%; Epidermal tabular forms ≤2.3%; Mesophyll ≤0.4%; Sclereid ≤1.9%; Tracheary annulate/helical ≤13.2% of phytoliths in any sample). Hat-shaped phytoliths (≤0.5% in any sample) tentatively assigned to Cyperaceae and Elongate dendritic phytoliths (≤0.7% in any sample) from the grass inflorescences are rare. No burnt or melted phytoliths were observed (Figure 7). Articulated phytoliths were frequently observed, comprising up to 24.4% of the total phytoliths in any sample (Figure 7). A few pollen grains were also observed in the scat samples, but sponge spicules and diatoms were not observed.

4.3 Ash pseudomorph concentrations

Ash pseudomorphs are negligible to absent in the sediment samples from P35 and R31. Fourteen sediment samples had no ash pseudomorphs (Figure 11 and Table 1). Ash pseudomorphs occur in extremely low abundances (11,000–20,000 ash pseudomorphs per 1 g of sediment) in one sample from P35 (25 cm) and three samples from R31 (XU2, XU21 and XU42) (Figure 11 and Table 1).

5 Discussion and conclusions

5.1 Depositional pathways of the Cloggs Cave phytoliths

Determining the pathways and processes by which phytoliths entered the deposits is essential for understanding the cultural and palaeoecological implications of the phytolith assemblage. In contrast to pollen, which is released from the plant flower heads, often in a position well above the ground surface, phytoliths accumulate at the interface between plants and the soil substrate where the plants die. Phytoliths are therefore considered a localised in situ element of plant decay (Piperno, 2006). The main mechanisms by which phytoliths enter archaeological sites are (1) vegetation growing on-site, (2) transport by geomorphic processes such as wind and water, (3) animals via faecal matter, and (4) cultural activities (Wallis, 2001).

At Cloggs Cave, insufficient sunlight penetrates the main chamber, where the excavations took place, and plants do not currently grow inside the cave away from the photic zone. On this basis, all plant materials recovered from the excavations are therefore allochthonous, introduced either by animals or geomorphic processes. While under the right circumstances phytoliths can be transported over long distances by wind or water (Strömberg et al., 2018), such geomorphic processes are very unlikely to have been responsible for the accumulation of the large abundances of phytoliths deep inside the cave. The 5 m-long, curving and narrow (0.85 m-wide) entrance passage into the cave, and the position of the cave's entrance on the lee-side of predominant winds restricts wind flow, and, while the cave receives percolating water as attested by the cave's speleothem formations, there is no sign of flowing water in the cave (Delannoy et al., 2020). Moreover, the phytoliths do not show evidence of having been physically abraded by long-distance transport (Osterrieth et al., 2009; e.g., Madella and Lancelotti, 2012).

In the absence of localised vegetation growth and geomorphic transport from further afield, we conclude that the abundant phytoliths detected in the sediment samples accumulated through the activities of animals or people. The high rates of articulated phytoliths (up to 35.1% of phytoliths in any sample) indicate that they were not introduced as single cells but, rather, in situ from decayed plant matter. Before attributing the origins of the phytoliths to the activities of people, we first need to consider whether animals that consume phytolith-producing plants could have excreted the phytoliths, a non-digestible element of plants, in their faeces.

Scats of the Common Brushtail Possum were found in abundance in the cave, including at depth in the stratified deposits. The Common Brushtail Possum is a generalist herbivore that feeds on leaves (including those of Eucalyptus spp.), flowers and fruits of a wide range of plants including grasses, herbs, trees and shrubs (mainly Acacia spp.) (Fitzgerald, 1984; Kerle, 1984). The palynology of the Cloggs Cave possum scats indicates that they consumed a broad spectrum of grasses, herbs including daisies, and woody taxa including Myrtaceae (Grono et al., 2024).

While the phytolith concentrations in the cave sediments vs. the scats are not directly comparable due their different compositions (the sediments being generally heavier than the largely organic composition of the dry scats), the phytolith concentrations in the two assemblages nevertheless show similar stratigraphic trends through the deposits. For example, P35 sediments and scats show low phytolith abundances (with the exception of the scat from XU4 P35), while in R31, the highest sediment and scat phytolith concentrations are found in XU21 (Figures 4, 5). Moreover, the same phytolith morphotypes dominate the sediment samples and the possum scats, including morphotypes of grasses and, rarely, morphotypes more common in woody plants (Figure 7). Similarities in phytolith concentrations and morphotypes between the cave sediments and possum scats could suggest that the phytoliths from the sediments derived from the disintegration of possum scats. However, the morphotypes shared between the sediments and the scats are among the most common phytoliths produced by plants and thus tend to be well-represented in phytolith assemblages generally. The individual possum scats are generally dominated by a single morphotype, possibly reflecting a single night of foraging and consumption. In contrast, the cave sediments, reflecting a gradual sedimentation rate, contain diverse morphotypes in varying abundances (Figure 7).

Independent stratigraphic evidence from Cloggs Cave of dozens of well-stratified ash lenses indicate that people brought large quantities of plants into the cave and burnt them. Unequivocal evidence for the deposition of phytoliths through the activities of people is provided by blackened phytoliths that were liberated into the sediments through the localised burning of plant matter in those layers (Figures 7, 9, 12). Blackened phytoliths are entirely absent in the possum scat samples. We therefore use burnt phytoliths that are detected throughout the stratigraphy as indicators of people bringing plants into the cave and burning them. While at least some of the Cloggs Cave phytoliths thus accumulated through the activities of people, the source of the unburnt phytoliths is more ambiguous. The unburnt phytoliths in the Cloggs Cave sediments probably represent vegetation introduced into the cave by both people and other agents, including disintegrated possum scats.

The micromorphological analysis at the site characterised Cloggs Cave as a relatively dry environment with minimal water percolation and localised (authigenic) mineral precipitation (Miller, 2022). Phytolith weathering and dissolution is not pronounced at Cloggs Cave, and the preservation of stratified plant material is generally excellent, including phytoliths, pollen, ashes, two trimmed Casuarina cunninghamiana or C. glauca sticks and whole Banksia sp. Leaves (Flood, 1974, pp. 180–181; David et al., 2021b, 2024; Miller, 2022; Grono et al., 2024). The finely layered stratigraphy of the upper levels of P35 and XU2–XU5 (SU2) in R31 (David et al., 2021a, 2021b)—the assemblages analysed here—signals that physical processes such as sediment reworking and vertical translocation are unlikely to have affected the phytoliths in any significant way. A very low incidence of post-depositional reworking is supported by the high rates of phytolith articulation (up to 35.1% in the sediment samples, and 24.4% in the possum scats) (Figures 7, 8).

5.2 Comparison of the phytolith and pollen records from Cloggs Cave

The accumulation of pollen and phytoliths in sediments are controlled by different production rates, dispersal and depositional processes, and preservation conditions (Bryant and Holloway, 1983; Piperno, 2006). Together, they provide complimentary datasets that broaden the recovery of plant remains at archaeological sites. A major difference between the two types of microfossils is their differential production in plants. Monocotyledons such as grasses and sedges produce more phytolith than other kinds of plants and thus tend to be overrepresented in phytolith assemblages. Eudicots and conifers produce few and sometimes no phytoliths, and the phytoliths they do produce include many redundant morphotypes (i.e., the same morphotypes are produced in a range of plant families), limiting taxonomic resolution (Piperno, 2006). In contrast, pollen production and dispersal in arboreal vegetation provides well-resolved reconstructions of forest communities, yet pollen grains of Poaceae (grass family) are notoriously difficult to differentiate below the family level. Thus pollen lacks the precision necessary for grass vegetation histories (Bremond et al., 2004; Piperno, 2006). A comparison of vegetation representation between the phytolith and pollen assemblages at Cloggs Cave demonstrates these differences. While the pollen data provides broad vegetation reconstruction across woody and herbaceous elements, aquatic plants and grasses, the phytolith data is overwhelmingly dominated by grasses with very low representation of aquatic plants, herbs and woody taxa (Figure 12).

Phytolith analysis at Cloggs Cave corroborates the dominance of grasses established in the pollen analysis (Grono et al., 2024) (Figure 12), and provides more detailed insights of the occurrence of grasses in the cave. GSSCP, unequivocal diagnostic indicators of grasses, are the dominant class of phytoliths in the assemblage, and other common morphotypes (e.g., Acute bulbosus, Blocky, Elongate entire, Elongate sinuate), although formally redundant due to their production in many plant families, are predominantly produced in grasses (Figure 7 and Table 2). In phytolith studies, GSSCP morphologies are used to discriminate between major grass subfamilies, especially between Pooideae, reflecting cool, temperate C3 grasses, Panicoideae, reflecting warm, humid C4 grasses, and Chloridoideae, representing warm, arid C4 grasses (Piperno, 2006; Twiss et al., 1969). In present-day Australia, C4 grasses dominate the northern regions of the continent that experience warm, humid, tropical and sub-tropical climates, whereas C3 grasses become prevalent in the southern part of the continent, in which Cloggs Cave is situated, where conditions are cooler and temperate (Hattersley, 1983). Based on GSSCP morphologies, the landscape surrounding Cloggs Cave appears to have been populated by a mixture of C3 and C4 grasses, with slight fluctuations detected throughout the sequence (Figure 10). However, despite the accepted GCCSP subdivisions, there exists major overlap in GSSCP morphologies between grass subfamilies—for example, the Rondel morphotype can occur in grass subfamilies other than Pooideae and Bilobate can occur in grass subfamilies other than Panicoidaeae. Due to these reasons, and as detailed GSSCP classification schemes have not yet been developed for the local vegetation in Gippsland (e.g., Alexandré et al., 1997; Barboni and Bremond, 2009; Boyd et al., 2024a; Piperno and Pearsall, 1998), we cannot refine palaeoecological inferences and we do not further advance the interpretation of C3 and C4 grass distributions based on the phytolith data.

In contrast to grasses, woody vegetation is poorly represented in the phytolith assemblage. Morphotypes that can occur in woody taxa (e.g., Epidermal tabular, Irregular elongate Mesophyll, Tracheary annulate/helical, Sclereids) are present in low amounts (Figures 7, 12), yet these can also occur in monocotyledons and have low taxonomic value (Table 2). Notably, morphotypes diagnostic of woody taxa such as Spheroid, Ellipsoid and Irregular bodies and Polyhedral types (An and Xie, 2022; Boyd et al., 2024b; Piperno, 2006; Piperno and McMichael, 2020; Wallis, 2003; Watling and Iriarte, 2013) were not observed. The pollen analysis established the presence of woody taxa (Figure 12), including Acacia spp., Casuarina spp., Myrtaceae, Proteaceae and Sapindaceae (Grono et al., 2024), taxa which are non-producers or producers of redundant morphotypes of indeterminate taxa (Boyd et al., 2024b; Piperno, 2006; Wallis, 2003).

Among herbaceous taxa, phytolith production in the Asteraceae family, an abundant taxa in the pollen sequences (Grono et al., 2024), varies from no or very few phytoliths to morphotypes including Acute bulbosus, Elongate forms, Epidermal and Tracheary annulate/helical that are not exclusive to Asteraceae (Katz et al., 2013; Pearce and Ball, 2020; Wallis, 2003) (Table 2). Opaque perforated platelets, provisionally identified in low numbers throughout the sequence, are distinctive to Asteraceae and are produced in the inflorescences of the plant (Bozarth, 1992; Piperno, 2006). This morphotype is the only type attributed to herbaceous taxa in the phytolith assemblage; in contrast, the pollen assemblage is dominated by herbaceous taxa (Figure 12) (Grono et al., 2024). Herbs other than Asteraceae of note in the pollen assemblage include Plantago spp., Onagraceae and Thymelaeaceae, plants which produce negligible and/or non-diagnostic phytoliths and/or have not been studied previously (Piperno and McMichael, 2020).

In addition to production, two other important differences between pollen and phytoliths are dispersal/transport and depositional processes. Animals and people were the main agents for the introduction of both phytoliths and pollen into the cave (see above for the phytoliths; for the pollen, see Grono et al., 2024). Both the phytoliths and pollen came into the cave when plants were brought in, as evidenced by abundances of articulated phytoliths (up to 35.1% in any sample, Figures 7, 8) and single-taxon pollen clumps that were introduced into the cave without having been separated through aerial pollen dispersal, indicating that they had arrived in the cave while the pollen was still attached to the parent plant (Grono et al., 2024). Yet while the distinct pollen spectra of the cave sediments and possum scats enable us to differentiate between people and animal accumulations, inherent problems of multiplicity (i.e., different morphotypes produced within a single taxon), redundancy (i.e., the same morphotypes produced in different plant taxa) and differential production (i.e., different rates of phytolith production among plant taxa) in phytolith analysis have obscured ecological signals and produced unresolvable similarities in possum and sediment phytolith assemblages (see above for details).

5.3 People–plant interactions at Cloggs Cave

The P35 sequence, dated to between 4,400 cal BP and 1,600 cal BP, and the R31 sequence, dated to between 25,740–24,070 cal BP and 3,360–1,460 cal BP, together represent some 25,000 years of GunaiKurnai ancestral story and history [together referred to as ‘(hi)story’] of the cave (Table 1). While we conclude that some of the phytolith assemblage probably derived from the disintegration of possum scats, the presence of burnt phytoliths (Figures 7, 9, 12) is a secure indicator that a significant proportion of the phytoliths were brought into the cave by people via the plants they carried. At an archaeological timescale, the presence of burnt phytoliths through much of the deposit shows that people brought and used plants into the cave through the entire span of the past 25,000 years. The burnt phytoliths are numerous at 25–35 cm depth in P35 (3.5–10.9% of all phytoliths) (Figures 7, 12), where several superimposed stratified ash lenses collectively date from c. 5,000 to 1,600 cal BP (David et al., 2021b). Burnt phytoliths are also abundant in XU4–XU11 in R31 (3.9–18.3% of all phytoliths), dated to between 7,290–2,950 cal BP and 12,950–11,420 cal BP (Figures 7, 12).

Total phytolith concentrations, including burnt and non-burnt fractions of the assemblage, are highest between XU18 and XU30 in R31, dated to 15,000–12,000 cal BP (Figures 4, 12), potentially indicating the relative intensity of human visitations to the cave and activities involving plant material deposition in this period. However, this inference is tentative as it is unknown whether the timeframes of cultural and natural sedimentation are equal across all excavation units, and thus whether individual excavation units represent equivalent or non-comparable windows of time within which plant material accumulated. A further limitation of using phytolith concentrations as a proxy of relative intensities of human activity inside the cave is that the proportion of non-burnt phytoliths potentially derived from the activities of people and/or possums (see Section 5.1).

Independent lines of archaeological evidence from the site corroborate that people indeed brought plant material into the cave and that key activities involved the use of plants and fire (David et al., 2021b, 2024; Grono et al., 2024; Miller, 2022). Two trimmed and deliberately laid Casuarina cunninghamiana or C. glauca sticks dating to c. 11,000 and 12,000 cal BP, each associated with its own miniature fireplace (each fireplace measuring 15–20 cm in diameter) (David et al., 2024), finely stratified superimposed ash layers largely devoid of charcoal indicative of repeated ‘cool’ burns around a standing stone between c. 2,000 and 1,600 cal BP (Stephenson et al., 2020), and pollen grains of locally ethnographically known edible and medicinal plants dating from c. 25,000 cal BP onwards (Grono et al., 2024) have been found in the layers of greatest human activity when animals contributed little to sedimentary inputs. In contrast, the relatively vegetation-poor pre-human occupation and hiatus layers are rich in owl bone deposits and phosphates derived from other animal activities (Miller, 2022).

The dozens of plant ash layers in the cave would have required the selection and transport of plant material as fuel. As only a proportion (≤18.3% of phytoliths in any sample) of the assemblage appears burnt or melted, the fires would have burnt at relatively low temperatures preventing most phytoliths from blackening or melting, which tends to occur at temperatures above 450 °C and 600–800 °C, respectively (Fritzsch and Röpke, 2025; Wu et al., 2012). If the phytoliths represent plants that were used mainly for fuel, the phytolith morphotypes identified in this study suggests that it was grasses that were intentionally selected for burning. Expansive surface fires of soft plant matter inside the cave are consistent with GunaiKurnai traditional practices of using ash to perform magic (e.g., Pepper, 1980, p. 57). More widely across Australia, ‘cool’ surface fires were, and continue to be (including among GunaiKurnai) widely used for the deliberate production of smoke in ceremonial and medicinal practices by First Nations peoples (Ward and Friesem, 2021). In GunaiKurnai culture, the men who carried out the initiation ceremony, the jeraeil, were smeared with ash and charcoal powder, and tufts of grass crowned their heads and were pushed through the perforation of the septum (Smyth, 1878, pp. 63, 274; e.g., Fison and Howitt, 1880, p. 196; Howitt, 1885, p. 305, 1904, p. 619). Ash, charcoal and grass were used in special GunaiKurnai ritual performances as recorded in the mid- to late 1800s.

Phytoliths of the leaves and culms (GSSCP, Blocky, Elongate entire, Elongate sinuate) and inflorescences (Elongate dentate/dendritic and Papillate) (Figure 7) of grass plants indicate that whole plants were collected and transported into the cave, including the inedible parts of grasses. This suggests that grass plants were used for purposes beyond the exclusive exploitation of their grains for consumption, and supports previous archaeological and ethnographic evidence that the cave was visited for special practices rather than routine subsistence activities such as food preparation (David et al., 2021b).

Understanding the representation of woody plants in the assemblage is more problematic. Woody morphotypes attributed to woody taxa are negligible to absent in both the sediment samples and possum scats. However, this could reflect the low phytolith production of trees and shrubs. Ash pseudomorphs, produced when woody plants are burnt at or above 450–500 °C (Brochier and Thinon, 2003; Canti, 2003; Gur-Arieh and Shahack-Gross, 2020), are mostly absent from the sediment samples, except in four samples that all display additional, independent evidence of burning (Figures 9, 12). Two of the four samples—in P35 at 25 cm depth and in R31 XU2—are from the dense ash layers dating to the past 5,000 years; one is from R31 XU21, being from the layer of peak phytolith abundance and numerous blackened phytoliths; and one is from R31 XU42, which also has burnt articulated phytoliths, charcoal and what appears to be an ash clump, as determined by micromorphology (Miller, 2022). While ash pseudomorphs were recovered in low to absent amounts overall (Figures 7, 12), this does not necessarily indicate the absence of woody plants or their burning. Ash pseudomorphs are vulnerable to postdepositional diagenesis and dissolution (Canti and Brochier, 2017; Karkanas, 2021), and micromorphological analysis indicates decalcification, localised dissolution and diagenetic alteration of ashes (Miller, 2022) that may have adversely impacted the identification of ash pseudomorphs in this study.

Despite these limitations of the analyses, it remains clear from both the phytolith and pollen records that grasses were collected outside, then brought and used in abundant quantities inside the cave. We interpret this as evidence that people intentionally selected grasses over woody plants for cultural burning activities that left shallow layers or lenses of ash in the cave. The paucity of charcoal in the ashy sediments observed during excavations (David et al., 2021b), the rare, isolated charcoal fragments observed during micromorphological analysis (Miller, 2022) and the low amounts of ash pseudomorphs from woody taxa reported in this study all support this interpretation.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fearc.2026.1871928/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Francesco Berna, Simon Fraser University, Canada

Reviewed by: Yile Chen, Macau University of Science and Technology, Macao SAR, China

Konglan Shao, Nanjing University, China

Footnotes

¹The terms ‘Old Ancestors’ and ‘Old People’ are the respectful and culturally appropriate ways to refer to past generations of GunaiKurnai, as advised by the Elders today.