For peat’s sake! Peat type influences critical moisture thresholds that prevent combustion of organic soils in Western Australia

Introduction

Peat fires pose considerable hazards to firefighters, biodiversity, greenhouse gas concentrations and public health, and thus constitute a serious national and international issue. Peatlands combine flammable vegetation over organic substrates that may extend several centimetres to metres deep in a low-oxygen environment, representing a sizeable pool of stored carbon (e.g. Page et al. 2002; Grishin et al. 2009; Beaulne et al. 2021). When these organic-rich soils ignite, smouldering ground fires produce toxic emissions and promote char formation, high temperatures and a propensity for flare-ups and reignition of above-ground fuels over many months (Jones 2005; Huang and Rein 2014; Hu et al. 2018). Australian peatlands are also ecologically susceptible to fire due to their restricted extent, long recovery times and provision of specialised habitat to migratory or endemic species (e.g. Horwitz 1997; Pemberton 2005; Whinam and Hope 2005; Benson and Baird 2012; Winkle et al. 2021). It is unequivocal to avoid peatlands catching alight for these reasons, but avoiding peat fires is difficult to achieve.

Organic soils can become flammable when dry, with the degree of flammability dependent on the composition and the moisture content of the soils (Frandsen 1987; Semeniuk and Semeniuk 2005; Reardon et al. 2007; Huang and Rein 2015; Prior et al. 2020). Moisture thresholds for peat ignition are known to vary as a function of bulk density, organic carbon and inorganic content (e.g. Frandsen 1997; Reardon et al. 2007; Benscoter et al. 2011; Huang and Rein 2015; Huang et al. 2015; Prat-Guitart et al. 2016; Prior et al. 2020). Moisture thresholds around 110% have been shown to prevent ignition of disaggregated peat (e.g. nursery peat) compressed to a bulk density akin to peat turves (Frandsen 1987; Huang and Rein 2015; Prat-Guitart et al. 2016; Huang and Rein 2017). Higher moisture thresholds were necessary to prevent ignition of peat turves cut retaining much of their original structure, cut from peatlands in North America (120%, Frandsen 1997) and the United Kingdom (115–135%, Rein et al. 2008). This difference between compressed disaggregated peat and peat turves is likely due to differences in bulk density or other structural factors. Organic carbon content varies between peatlands, affecting moisture-holding properties, emissions and combustion dynamics measured as different mass and volume loss (Hu et al. 2018). Tropical peats typically have 12% more organic carbon content than temperate peats, but organic carbon content also varies between more localised peatlands. Inorganic content also varies between peatlands according to their geoformation and other processes (e.g. Semeniuk and Semeniuk 2005) and has an inverse effect on moisture thresholds to prevent peat ignition. For example, peat samples collected across the United States (Frandsen 1997) and Tasmania, Australia (Prior et al. 2020) showed lower moisture thresholds when the peatland soils had higher inorganic content. These natural variations of bulk density, organic and inorganic content across peatlands makes it unclear whether a single moisture threshold is sufficient to prevent ignition of all peats.

The hydrology of peat systems can also affect the rate of peat formation, nutrient status, inorganic content and physical structure (i.e. bulk density) (Gore 1983; Moore 1987). In a Floridian cypress bog study, soil moisture content was found to vary according to a sample’s distance to the edge of a peatland, and soil moisture alone did not predict depth of peat burning (Watts 2013). Studies from temperate peatlands in North America and the United Kingdom have shown botanical composition of peat affects bulk density, combustion rate and the heat released during combustion that can affect moisture thresholds (Frandsen 1987; Grishin et al. 2009). Higher lignin contents can reduce the rate of decomposition and peat compaction, affecting relative bulk density (Collinson and Scott 1987). A recent study using structurally-intact soil turves from Tasmanian peatlands found the moisture required to prevent ignition ranged from 25 to 95% gravimetric moisture, depending on organic content and bulk density of the sample, which related to vegetation types ranging from sphagnum mosses to woody scrub (Prior et al. 2020). Most peat formations in southwest Western Australia (SWWA) originate from sedges and restionaceous species rather than sphagnum mosses, thus moisture contents needed in SWWA peatlands may vary from boreal and some Tasmanian Sphagnum moss peatlands (SWWA; Pemberton 2005; Whinam and Hope 2005).

Peats are commonly associated with wetlands, which have been classified in southwestern Australia using a combined hydroperiod and geomorphic approach (Semeniuk and Semeniuk 1995). Hydroperiod refers to the duration and timing of waterlogging or inundation, which may be permanent or seasonal. Geomorphologies include basins, flats, channels, slopes and highlands. Seasonally waterlogged basins, slopes, flats and highlands may contain peat-forming sedge species including Empodisma gracilliumum and Reedia spathacea (Winkle et al. 2021). Slow-moving smouldering fires that burn the organic substrate within these communities can cause substantial damage to root zones, which hinders or even prevents recovery of peatlands (Rein et al. 2008; Horwitz et al. 2009; Watts 2013; Turetsky et al. 2015; McDougall et al. 2023). Ecological effects of peat fires are exacerbated by the Mediterranean climate in SWWA that includes dry summer conditions not suited to re-establishing wet-adapted species.

The moisture content of peatlands varies seasonally with fluctuations in rainfall (Prior et al. 2020). Annual rainfall has demonstrated a decreasing trend over the past several decades in SWWA, leading to reduced recharge of groundwater systems important to sustain wetlands (Pemberton 2005; Semeniuk and Semeniuk 2005; Department of Environment and Conservation 2012; IPCC 2021). Recent summer bushfires have led to some deep peatland systems burning away to mineral earth, indicative of unprecedented drying. Fire managers advocate for the exclusion of fire from wetlands when possible, but the lack of sufficient recharge due to climate change increasingly puts peatlands at risk from prescribed burning operations. Recently, prescribed burning operations have ignited some deep peatland systems while leaving others nearby intact. These experiences have highlighted the need to identify critical moisture thresholds that influence the potential for peatlands to sustain smouldering combustion.

Given the mixed susceptibility of neighbouring peatlands to burn when exposed to the same fire, it is unclear if a single threshold is sufficient for all peatlands, or if moisture thresholds to prevent sustained combustion vary between peat types. The aim of this study was to determine (1) if peatlands within different geomorphologies exhibited different critical moisture thresholds, and (2) if peat classification alone would provide a sufficient key to determine operational moisture thresholds or additional factors were necessary to predict susceptibility to ignition. A further aim was to determine the mathematical relationship between gravimetric moisture and a soil moisture probe likely to be used in an operational setting.

Methods

Study region and sampling design

We compared moisture thresholds to prevent sustained combustion between peatlands from four different geomorphologies located within the Warren bioregion in SWWA, a Mediterranean climate region (Rundel et al. 2016) (Fig. 1). Gravimetric moisture content was used throughout this study. Wetland geomorphologies containing peatlands in SWWA include seasonally waterlogged basins (dampland), slopes (paluslope), highlands (palusmont) and flats (palusplain) (Semeniuk and Semeniuk 1997). Historic annual rainfall averages 1400 mm within the Warren bioregion, primarily occurring within the six consecutive wettest months (May–October). Minimum rainfall coincides with the three summer months when maximum temperatures often exceed 35°C and frontal systems bring dry lightning storms (Dowdy and Kuleshov 2014; Bates et al. 2018). Prescribed burning is regularly used to reduce fire hazard and promote conservation within SWWA, including the Warren bioregion (Burrows and McCaw 2013; Howard et al. 2020). A decreasing trend in rainfall in SWWA has been ongoing since the 1970s (IPCC 2021), contributing to less recharge of wetlands, inclusive of peatlands.

Fig. 1.

Map indicating the location of 16 peat sampling sites. Sites included four representatives from four seasonally waterlogged geomorphologies. Damplands (basin) shown by circles, Palusplains (flat) shown by squares, Palusmonts (highlands) shown by diamonds and Paluslopes (slope) shown by triangles. Red square in map inset indicates general location of sampling area within Western Australia.

Combustion

The protocol used by Prior et al. (2020) and others (see Rein et al. 2008) was modified slightly to combust peat samples and provide a conservative estimate of critical moisture thresholds for sustained combustion. Ten separate combustion chambers were assembled using clay bricks on a concrete paver within a fume cupboard to fully enclose samples on four sides, and the bottom with a heat-resistant barrier (Fig. 2). Peat samples were removed from vacuum sealer bags, weighed in the foil trays and then placed within an individual chamber, before foil trays were re-weighed to account for any remaining moisture or peat not retained in the sample block. These weights were used to assess mass loss due to combustion and to calculate actual pre-burn moisture content compared to target moisture. A ruler was used to vertically measure each corner of the sample to the top of the adjacent brick.

Fig. 2.

Combustion chambers for peat samples showing top-down (a) and lateral (b) view. Note lateral view has had the end brick removed following a combustion stimulus. Clay bricks were used on all four sides and samples sat on a concrete paver within a fume cupboard. Photographs taken by V. Densmore.

A flat heating element (350 mm long, 4.2 mm wide, 2.2 mm deep; 240 V, 200–250 W, J type inbuilt thermocouple formed to M shape; Yueqing Langchen Electrical Co, Hongqiao Town, China) was inserted between the brick and one side of the sample. A percentage voltage controller (Megawatt Heating Elements, Somerville, Victoria) was used to supply 100 W of power to the heating coil for 30 min, after which time the coil was switched off and gently extracted to minimise disturbance to samples, which remained in situ overnight to allow any smouldering combustion to complete, further evidenced by a lack of smoke or heat associated with the samples.

Once smouldering of the organic matter ceased, the vertical height from the top of the brick to the top of each of the four corners of the sample was remeasured. Samples were removed from the combustion chamber into pre-weighed foil trays, weighed and oven dried at 105°C for another 24–48 h to distinguish mass loss due to combustion rather than evaporation.

Results

The gravimetric moisture content needed to prevent successful combustion of peat samples varied significantly among both geomorphologies (P = 0.033, F = 3.516) and peat types (P = 0.044, F = 3.249) (Table 2, Fig. 3). Binomial models predicted 63.1% gravimetric moisture would prevent sustained combustion for paluslopes, while basins required 80.6% moisture and palusplains would need 101% moisture (i.e. g H₂O/100 g dry soil). Palusplains also showed the highest measured gravimetric moisture to prevent ignition (131.3%), compared to 100.6% for basins and 77.7% for paluslopes (Table 2). Geomorphologies did not help define a clear pattern of mass loss across combustion classes (Table 3, Fig. 3c).

Fig. 3.

Probability of peat samples combusting (a, b) or percentage mass loss of samples (c, d) as gravimetric moisture content increased compared across geomorphologies (a, c) and peat types (b, d). Binomial generalised linear models were used to calculate the probability that a sample would demonstrate moderate to strong combustion, as per Table 1. Percentage mass loss calculated using measurements taken before and after combustion trials.

Binomial modelling predicted sapric peats would require 115.7% gravimetric moisture to resist sustained combustion, significantly more than fibric peats (68.2%) (P = 0.006, F = 3.585), but not hemic peats (79.7%) (Table 2, Fig. 3b). No sapric samples were classified as nil combustion, thus the moisture content required to prevent ignition was not determined for sapric peats. Gravimetric moisture to prevent ignition was measured as 89.5% for hemic peat and 98.3% for fibric peat (Table 2, Fig. 3b). Sapric peat types also lost significantly more mass than hemic or fibric types in strong and weak combustion classes (Table 3, Fig. 3d, P < 0.001, t ratio = 4.59).

Peat type predicted combustion more accurately than geomorphology when either was combined with gravimetric moisture content in proportional odds logistic regression (OLR) GAMs. EC, total organic carbon (TOC) and total nitrogen also improved models based on moisture alone (Fig. 4). The most accurate and parsimonious model combined gravimetric moisture, peat type and EC (Table 4), with a 28% greater predictive ability than pre-burn moisture alone. When peat types were modelled separately, pre-burn moisture was sufficient to predict combustion class for sapric peats, but EC improved the models for hemic and fibric peats by 4.7 and 18.5%, respectively (Table 4).

Fig. 4.

Partial dependence plots demonstrating influence of best performing factors to predict combustion class as per Table 4. (a) Role of peat types in model c.mp. (b) Role of electrical conductivity (EC) in model c.me. (c) Role of nitrogen (N) in model c.mn. (d) Role of EC in best-performing model, c.mep.

Average values for several physical and chemical attributes varied significantly across peat types (Supplementary Table S2). Fibric peat samples had 25–36% higher acid neutralising capacity, 18–22% lower bulk density and 5–39% higher clay content than hemic or sapric peat samples. Hemic peat samples had 19–28% lower silt content, 14–19% lower coarse sand and approximately 22% more fine sand than fibric or sapric peat samples. Sapric peat samples had 54–85% greater EC, 31–47% more total carbon and 85–112% more total nitrogen than fibric or hemic peat samples.

A strong relationship (R² = 0.86) was found between soil probe moisture readings and gravimetric soil moisture calculated using samples collected immediately after probe readings (Fig. 5a). The best fit line describing this relationship had an R-squared of 0.86 and equation y = 6.4374x + 0.0347, where y is the gravimetric moisture and x the soil probe moisture reading. This equation was used to calculate the soil probe readings required to prevent ignition or sustained combustion (Table 5; Fig. 5b). The data suggests a soil probe reading of 18% would be sufficient to minimise or negate peat combustion.

Fig. 5.

Soil moisture readings required to prevent peat from combusting. (a) Relationship between gravimetric soil moisture content and field measurements of the same samples using an Acclima SDI-12 Sensor. The R-squared value and equation for a line fitted to the points is shown. (b) Soil probe readings corresponding to threshold gravimetric moisture for peat types to remain unburned. Probe readings were calculated using the equation in A with mean plus standard errors in Table 2 ‘Unburned’ column. Actual values provided above respective bars.

Discussion

This study found that peat type, more than geomorphology, significantly affected gravimetric moisture thresholds to minimise combustion or prevent ignition. Moisture content was able to predict combustion to over 85% accuracy when peat type was known. Adding EC of soil improved the overall model and models for fibric and hemic peats, but no other factors improved combustion models for sapric peat types. This suggests that physical and chemical factors previously found to influence critical moisture thresholds may be inherently aligned with peat classification. These findings improve the ability to operationally assess the vulnerability of peats to ignition in a field setting, without supplementary laboratory testing.

Peat types are designated using physical tests to determine the degree of constituent plant matter decomposition, or humification, which would reflect other factors known to influence critical moisture thresholds (Taufik et al. 2019). As plant material decomposes from identifiable material to indistinguishable components, particle size decreases, allowing greater soil compaction and less soil porosity that would translate as higher soil bulk density (Collinson and Scott 1987; Davies et al. 2013). The percentage of organic content is higher in more decomposed substrates, such as sapric peats compared to fibric peats, or peats formed from less woody material like sphagnum moss compared to more fibrous sedges (Collinson and Scott 1987; Semeniuk and Semeniuk 2005; Hu et al. 2018; Taufik et al. 2019). Higher organic content increases the probability that smouldering combustion will occur at higher gravimetric moisture contents (e.g. Prior et al. 2020), thus increasing critical moisture thresholds needed to prevent ignition or sustained combustion. The heat of combustion is also directly related to organic content (Shafizadeh and Sekiguchi 1984). Peat types that have higher organic content would likely produce more heat during initial combustion increasing the likelihood that any moisture within peats would evaporate (Huang and Rein 2014), supporting a longer duration and further spread of smouldering combustion (Huang and Rein 2017).

Bulk density has shown a direct relationship to critical moisture thresholds. More porous peats must be drier than denser peats to ignite (Schulte et al. 2019; Prior et al. 2020). Porous materials are able to bond water in hygroscopic form, creating a thin surface film that must be vapourised to be removed (Huang and Rein 2014). Studies on Indonesian peatlands found that more humidified peats (i.e. sapric and hemic) retained less water (Taufik et al. 2019), decreasing the amount of evaporation required. Evaporation is an essential first step before pyrolysis and char formation can occur, while the latter two can occur simultaneously (Shafizadeh and Sekiguchi 1984; Huang and Rein 2014). The greater heat generated through pyrolysis and char formation compared to an external stimulus, like a heating coil, has influenced several studies to use a dry conduit such as smouldering charcoal or dry peat between the heating coil and peat samples being tested for critical moisture content (Frandsen 1997; Prat-Guitart et al. 2016; Pastor et al. 2017; Bates 2018; Prior 2020).

In this study we applied the heating coil directly to peat samples without a dry intermediary and found that sapric peats required significantly higher moistures to resist ignition compared to fibric or hemic peats. Sapric peats had the highest carbon and nitrogen contents and EC, indicating more organic matter. Although adding EC slightly improved the ability of models to predict combustion of fibric and hemic peats, EC values were not statistically different between fibric and hemic peats. Sapric peats also had significantly higher bulk density than fibric, but not hemic, peats. Fibric peats had the highest acid neutralising capacity, indicating significantly higher calcium carbonate content (Semeniuk and Semeniuk 2005), which may have decreased the moisture required to prevent ignition or sustained combustion (Semeniuk and Semeniuk 2005). Notably, moisture thresholds did not differ significantly between fibric and hemic peats at any combustion class, but the moisture required to prevent sustained combustion after ignition (e.g. weak combustion class) in hemic peats was statistically intermediate between fibric and sapric peats. Further testing with larger sample sizes is needed to establish whether this trend is significant.

This study was designed to test potential differences between geomorphologies rather than peat type, thus using the latter to model combustion probabilities had fewer samples than originally intended. Sites representing different geomorphologies could be chosen prior to sampling, unlike peat type, which required physical examination of samples. The act of sampling peat in Western Australia required a disturbance permit, thus constraining the design of this study. Moisture thresholds to prevent ignition or sustained smouldering combustion did differ significantly for palusplains relative to basins and paluslopes, but modelling showed peat type was a more accurate predictor of moisture thresholds than geomorphology. In addition, several peatlands selected and sampled failed to show sustained combustion even in dry soils, and were predominantly found to have carbon contents below 10%. Some sites did show evidence of prior fire damage, which may have caused degradation of the peats (Turetsky et al. 2015). Although this reduced the number of peat sites to be tested, using peat type reduced the groups from four to three, thus compensating for some loss of sites. Further, BIC were used during modelling to address small sample sizes, and this study’s findings make ecological sense. However, we recommend follow-up testing to confirm the hypothesis that peat type will affect critical moisture content.

Previous studies have identified multiple factors affecting ignition and sustained combustion of peat, andrequire further testing of peat samples often using specialised equipment. This presents a difficulty to fire managers who may lack the time and resources necessary to determine the gravimetric moisture, bulk density, organic and inorganic contents of peats. Source of peat has also been found to affect moisture thresholds (e.g. see Huang and Rein 2014, 2017 versus Lin et al. 2024). Testing flammability of peat turves from various sources have identified moisture, organic matter, inorganic content and bulk density to affect the probability of ignition, but critical moisture thresholds vary (e.g. Frandsen 1997; Reardon et al. 2007; Davies et al. 2013; Watts 2013; Prat-Guitart et al. 2016; Pastor et al. 2017; Prior et al. 2020).

This study has identified that peat type may be sufficient to determine critical moisture thresholds to prevent combustion in temperate ecosystems, which has considerable implications for operational use. Peat type can be classified fairly easily using the Australian Soil Classification guidelines (Isbell and National Committee on Soil and Terrain 2024 “Peat” in ASC-Glossary 1993). Robust relationships are also evident between soil moisture probes and gravimetric moisture (Prior et al. 2020, this study). These findings have importance to translating basic science into applied functionality, at least within southwest Western Australia. Future research to examine whether these principles apply more broadly may have great benefit to peatland conservation, particularly in light of changing rainfall patterns due to climate change.

Supplementary material

Supplementary material is available online.