Mercury in an Australian sclerophyll Eucalyptus forest and emissions from fuel reduction prescribed burning

Site description

The study area (38°03′S, 146°25′E) is classified ecologically as lowland wet sclerophyll forest in the Highlands – Southern Fall bioregion of Gippsland, Victoria (EVC 16). Located in the Boola Boola State Forest between the townships of Walhalla and Tyers (Fig. 1), the forest overstorey is dominated by silvertop ash (Eucalyptus sieberi). Messmate–stringybark (E. obliqua), narrow leaf peppermint (E. radiata) and broad-leaved peppermint (E. dives) comprise the rest of the overstorey.

Fig. 1.

Map of the study area (right panel) highlighting collection points with blue markers for unburned sites and brown markers for burned sites. The hatched area indicates the extent of the planned burn; however, burning extended beyond these boundaries, accounting for the presence of brown markers outside the shaded region. The extent of spread outside the planned area is unknown. The lower left panel presents the study site’s position relative to the towns of Walhalla and Tyers. The upper left panel provides a broader view of the study site’s location within Australia. The map was created using ArcGIS Pro (ver. 3.1.3, Environmental Systems Research Institute, Redlands, CA, USA).

Sclerophyll forests are characterised by trees with hard, leathery, evergreen foliage specially adapted to long periods of dryness and heat and for prevention of moisture loss (Heise-Pavlov and Procter-Gray 2021). The study site climate is temperate (type Cfb according to Köppen), cool-wet, with average annual precipitation 803 mm and temperature 11.8°C (Climate Data 2024). The area is exposed to moderate to high rainfall and weathering of tertiary sediments, resulting in moderately to strongly acidic, leached soils with well-developed horizons (Aldrick et al. 1992). The area is characterised by Yellow Dermosols (code DEAC in the Australian Soil Classification), that is, soils lacking a strong contrast between the A and B horizons (Isbell and National Committee on Soil and Terrain 2021). In this study, we defined the O horizon as the visible organic layer with distinct darker brown colouration and the A horizon as the 3-cm layer immediately below the O horizon.

The study site underwent a prescribed burn in March 2022, predominantly on the western side of Tyers–Walhalla Road (Fig. 1). This has offered a unique opportunity to conduct a paired study of total Hg concentrations in soil from inside and outside of the prescribed burn. An area of ~290 ha was burned as part of this March 2022 prescribed burning (the planned burn extent is marked on the map in Fig. 1 by the hatched area).

Prior to this 2022 prescribed burn, the most recent burn in the area occurred in autumn 2004, conducted as a prescribed burn by the Victorian Department of Energy, Environment and Climate Action (DEECA).

Sample collection and processing

Field sampling was conducted over two winter campaigns. Soil and leaf litter samples were collected in July 2022, whereas bark and twig samples were collected in June 2024. The second campaign was necessary because bark and twigs were not collected during the initial campaign, and it was subsequently determined that these data are essential in calculating Hg emissions from the burning of biomass.

This work was conducted under permit AA-0001030 on land managed by Parks Victoria.

Mercury analyses

For Hg analyses, 100 ± 20 mg of the sample was weighed into nickel boats and analysed using a Milestone Direct Mercury Analyser (DMA-80 Tricell; Milestone, Bergamo, Italy), following USEPA method 7473 (United States Environmental Protection Agency 1998). The DMA-80 thermally decomposes all species to Hg⁰, using amalgamation and then quantifies total Hg in solid samples by atomic absorption spectrometry. A pair of blanks and a pair of certified reference materials (SRM 2706, New Jersey soil, from the National Institute of Standards and Technology, USA) were analysed for every 36 soil samples. For litter samples, SRMs 1515 (apple leaves) and 1575a (pine needles) were used as reference materials. Care was taken to analyse additional SRM pairs with concentrations falling in the same spectrometer cell of the DMA-80. The results of the standard reference material are provided in Supplementary Table S1.

Duplicate samples were run for every five samples, results for these pairs varied less than 10% and were averaged and reported as the mean. When Hg concentration varied by greater than 10% for a duplicate pair, a triplicate was analysed.

Organic matter analyses of soil

Organic matter was analysed using the loss on ignition analysis of organic matter content described by Wang Q et al. (2011), which comprises heating ~1 g of sample in a muffle furnace (LABEC, model CEMLL) at 550°C for 8 h. Samples were left in the furnace to cool for 4 h and then weighted. The difference between initial and final weight was used to calculate the percentage of organic matter loss.

Emission factor and biomass combustion calculation

The Hg emission factor (EF) from biomass combustion in each compartment (i) (EF_i, g Hg ha⁻¹) was calculated according to Eqn 1:

where DB_i is the dry biomass in a specific compartment (leaf, bark or twig) (Mg biomass ha⁻¹), C_Hgi is the Hg concentration (ng g⁻¹) in the biomass of the compartment i, and CC is the combustion completeness, which is defined as the ratio of fuel consumed by fire to total available fuel in the biomass category.

The dry litter biomass for the sclerophyll forest was estimated from published and unpublished studies of litter and combustion in prescribed fires (Keith 1991; Volkova and Weston 2015; Neumann et al. 2021). The total aboveground litter dry mass for E. sieberi foothill forest in Victoria was 13.5 Mg ha⁻¹ (Volkova and Weston 2015). The range in litter mass for sclerophyll open forest in south-east Australia includes 11–19 Mg ha⁻¹ (Keith 1991; Neumann et al. 2021). Based on the range of available data, we used a conservative estimate of litter mass of 12.5 Mg ha⁻¹. The overall EF from the biomass combustion (EF_bio, g Hg ha⁻¹) was then calculated according to Eqn 2:

The EF_bio (ng Hg g⁻¹) for each compartment i, was also calculated as the mass of Hg emitted per unit of dry biomass burned (Andreae and Merlet 2001) according to Eqn 3:

where CM_i corresponds to the component of the combustible material that makes up the litter. The overall EF_bio (ng Hg g⁻¹) was then calculated with Eqn 4:

In specific research studies, combustion factors for litter in prescribed fires have been reported as 60% for forests and woodlands in Victoria (Volkova et al. 2019), 70% in E. sieberi open forest in Victoria (Volkova and Weston 2015), 66–84% in E. sieberi forest in coastal Victoria (H. Keith, unpubl. data) and 68% in E. pauciflora forest in the the Australian Capital Territory (Keith 1991). We estimated an average combustion efficiency of 66%.

Our study site is classified as a cool wet forest based on the forest litter type categorisation by Neumann et al. (2021). To calculate Hg emissions, we used data from Neumann et al. (2021), which was gathered from cool wet climates in the south-eastern mountains of Australia. Additionally, we incorporated field data from alpine ash (E. delegatensis, a species closely related to silvertop ash) in the southern tablelands of New South Wales. The litter composition was estimated as 20% leaves, 30% twigs, 20% bark and 30% other materials, with the ‘other’ category referring to unidentified fragmented material present in the lower layers of the litter.

Prescribed fire burns all fine litter components, including bark and twig components (NSW Rural Fire Service 2005). For Hg emission calculations, all components are included in the percentage of burned fuel. The proportion of other unidentified fragmented material was reallocated to the leaves, twigs and bark categories to give proportions of the burned material, estimated to be 29% leaves, 43% twigs and 28% bark. It is important to note that only fallen leaves were sampled as the canopy should not burn during a prescribed fire, provided the burn is conducted according to the prescribed guidelines.

Statistical analysis

All statistical analyses were performed using the R statistical package (ver. 4.4.0, R Foundation for Statistical Computing, Vienna, Austria, see https://www.r-project.org/). Prior to analyses, all Hg concentration and OM percentage data were log-transformed to meet the assumption of normality. Where this assumption was not met, non-parametric analyses were used. An alpha level of 0.05 was applied for significance in all statistical tests. The full results of all statistical tests are presented in the supplementary data.

Soil

Mercury and organic matter concentrations in unburned soils

Mean Hg concentrations in unburned soils were highest in samples collected from the base of trees, followed by O horizon soils, and A horizon soils (143 ± 61, 112 ± 71, 56 ± 30 ng g⁻¹ respectively) (Table 1, see Fig. 2 for medians). Soil organic matter is highest at the base of tree stems due to the accumulation of litterfall and consequently high microbial bacterial activity (Ryan and McGarity 1983). Therefore, Hg concentrations from the base of trees were analysed only in unburned sites to characterise soil Hg levels in the sclerophyll forest, rather than to assess emissions from burning. These high Hg levels are important to measure and record because they contribute to the total Hg load in the soil.

Fig. 2.

Median Hg concentrations in tree base, A and O horizon soils. A and O horizon soils are presented for both burned and unburned sites. Note that soil samples from the tree base were collected only from unburned sites. The central line within each box represents the median Hg concentration, whereas the edges of the box denote the first and third quartiles (Q1 and Q3). The whiskers extend to the smallest and largest values within 1.5 times the interquartile range (IQR). Outliers beyond this range are marked as individual points.

One-way ANOVA indicated a significant main effect of soil layer (unburned O and A horizons, and base of the tree) on Hg concentration (F_{2, 45} = 16.45, P < 0.0001) (Fig. 2, Supplementary Table S6). Tukey’s HSD post hoc testing revealed statistically significant differences between the unburned O and A horizons (P < 0.005) and soils from the base of trees and the unburned A horizon (P < 0.0001) (Fig. 2, Supplementary Table S6). No significant difference was found between unburned soil from the O horizon and the base of trees (P = 0.0936) (Supplementary Table S6).

Mean OM content in unburned soils was highest in samples collected from the base of trees, followed by O horizon soils, and A horizon soils (50 ± 29, 28 ± 18, 9 ± 2% respectively) (Table 1). Median OM content followed the same pattern (49, 24, 9%) (Table 1, Fig. 3). This difference is statistically significant, as indicated by the Welch’s ANOVA (F_{2, 19.445} = 23.699, P < 0.0001) (Supplementary Table S6). Post hoc testing with the Holm-adjusted Games–Howell test yielded significant differences between OM concentrations in O and A horizon soils (P < 0.05), tree base soils and A horizon soils (P < 0.005), and between tree base soils and O horizon soils (P < 0.005) (Supplementary Table S6).

Fig. 3.

Median percentage of organic matter (OM) in soil across different locations: tree base, organic horizon (O) and A horizon (A). For the O and A horizons, data are presented for both burned and unburned areas. Note that soil samples from the tree base were collected from unburned sites exclusively. The central line within each box represents the median OM concentration, whereas the edges of the box denote the first and third quartiles (Q1 and Q3). The whiskers extend to the smallest and largest values within 1.5 times the interquartile range (IQR). Outliers beyond this range are marked as individual points.

Organic matter and mercury correlations in soil

Pearson correlations were calculated between organic matter content and mercury concentration for all soil types. Since no significant difference in Hg concentrations was observed between burned and unburned sites, correlation analyses were performed using all soil samples combined (burned and unburned) to increase the statistical power of the analysis.

A positive correlation was observed between OM and Hg concentrations across the study sites. The regression analysis yielded a significant relationship (P < 0.05, Fig. 4, Supplementary Table S8), with the model for all samples combined explaining 60% of the variance in Hg concentrations (R² = 0.6). This indicates that sites with higher OM content tend to have higher Hg concentrations.

Fig. 4.

Assumed linear relationships between log transformed mercury concentration and organic matter in; (a) all soil combined, (b) tree base soil, (c) A horizon soil, and (d) O horizon soil.

Mercury concentrations in litter

The litter samples collected from both burned and unburned sites consisted of unburned leaves, twigs and bark that either remained unburned or had fallen after the fire. Consequently, all litter samples from burned and unburned sites in this study were combined, and no statistical analyses were performed to differentiate between the two prescribed burning contexts.

Mean Hg concentrations in litter were highest in leaves, followed by bark and twigs (71 ± 11, 21 ± 13, 7.5 ± 3.9 ng g⁻¹ respectively). Median Hg concentrations followed the same pattern (69, 16, 6.3 ng g⁻¹ respectively) (Table 2, Fig. 5).

Fig. 5.

Median Hg concentrations (ng g⁻¹) in bark, leaf and twig litter components. The central line within each box represents the median Hg concentration, whereas the edges of the box denote the first and third quartiles (Q1 and Q3). The whiskers extend to the smallest and largest values within 1.5 times the interquartile range (IQR). Outliers beyond this range are marked as individual points.

A significant difference in Hg concentrations between litter types was identified using a Kruskall–Wallis test (χ²(2) = 69.6, P < 0.0001). Post hoc Dunn testing found significant differences between bark and leaf litter, bark and twig litter, and leaf and twig litter (Z = −3.8, P ≤ 0.0001; Z = 5.1, P ≤ 0.0001; Z = 8.1, P ≤ 0.0001 respectively) (Supplementary Table S9).

Calculated emissions from the prescribed burn

The Hg burden in litter in unburned areas is 374 g Hg ha⁻¹ (Table 3). The emission factors for biomass burning (EF_bio), calculated for each type of biomass, are shown in Table 3. Soil Hg emissions were not calculated, as the results showed no significant difference in Hg concentrations between burned and unburned areas. Based on the estimated total burned area of 290 ha, the EF_bio was calculated as 0.25 g Hg ha⁻¹ for the entire burned area.

Mercury concentrations in bulk soil and near trees of sclerophyll forest

Our study revealed a wide range of Hg concentrations in soils of sclerophyll forest, with 143 ± 62 ng g⁻¹ measured at the base of trees, 100 ± 65 ng g⁻¹ in the organic layer and 51 ± 25 ng g⁻¹ in the mineral layer. These concentrations are among the highest recorded in Victoria (Table 4). For comparison, soil Hg concentrations in Pumphouse, central Victoria, were measured at 29.4 and 25.3 ng g⁻¹ in the topsoil (0–2 cm) and lower (5–10 cm) soil layers respectively (Howard et al. 2019). In the Latrobe Valley, located south of our study site, the mean soil Hg concentration was reported as 53.5 ± 25.7 ng g⁻¹ (Schneider et al. 2021). Similarly, in a study conducted in Victoria, a comparable mercury concentration of 90 ng g⁻¹ was documented (Packham et al. 2009).

Soils representing environmental background Hg concentrations were sampled around Glenbawn Lake in New South Wales, measuring at 18.2 ± 1.1 ng g⁻¹ (Schneider et al. 2021). By comparison, background Hg concentrations in coastal soils from Queensland and New South Wales have been reported to range from 10 to 50 ng g⁻¹ (Rayment et al. 1998). When comparing our findings to a broader study encompassing the entirety of Australia, our results are consistent with those of Reimann and de Caritat (2017), who documented the highest median Hg concentrations (~40–50 ng g⁻¹) in Victoria and Tasmania relative to other Australian states. However, caution should be exercised when comparing our data with other studies in Australia, as different analytical methods were employed. Additionally, our study focused on the <63-µm soil fraction, whereas other studies analysed bulk soil (Table 4).

Another important consideration regarding the higher concentrations found in this study is the forested area and type of soil where the samples were collected. In forest ecosystems, soil Hg concentrations are generally higher than in open terrestrial ecosystems due to the efficiency of forest canopies in trapping atmospheric Hg (Wang X et al. 2019; Zhou et al. 2021). Broadleaf evergreen forests have been shown to be particularly effective at transferring atmospheric Hg (in part due to the long lifetime of leaves), and consequently at transferring atmospheric Hg to the soil through litterfall deposition, resulting in higher soil Hg concentrations compared to deciduous forests (Wang X et al. 2016b; Chen et al. 2024). Although the role of wet sclerophyll forests in the Hg cycle has been understudied, it is likely that forest structure similarly contributes to increased atmospheric Hg uptake, thereby elevating Hg concentrations in the soil. Furthermore, the soils in the wet sclerophyll forests of Victoria are predominantly loamy (Gullan 2017), a soil type known for its high efficiency in retaining Hg. This is because Hg, like other trace elements, tends to concentrate on the surface of finer particles, such as silt and clay (Ravichandran et al. 1995; Boszke et al. 2004).

The Hg concentrations observed in this study exceed previously published data for Australian soils (Table 4), yet they are consistent with the typical range found in uncontaminated podzol–spodozol soils from temperate and boreal regions (Grigal 2003; Obrist et al. 2011; Kronberg et al. 2016; Gómez-Armesto et al. 2020). The wide range of variance in Hg concentrations across the sites within the sclerophyll forest is likely attributable to variations in environmental factors that influence soil organic matter content and atmospheric Hg deposition rates.

Vertical mercury distribution in soils

The vertical distribution of Hg in sclerophyll forest soils in this study closely resembles the patterns observed in temperate and boreal forests, with greater accumulation in the surface layer where Hg is predominantly bound to OM (Schwesig and Matzner 2000; Grigal 2003; Obrist et al. 2016; Landis et al. 2024). The positive correlation between the Hg and OM content supports this hypothesis. The higher Hg concentrations observed in the O horizon compared to the A horizon can therefore be directly linked to the elevated OM content in the surface organic layer. In forest ecosystems, the decomposition of litterfall enriches the superficial soil in OM. The thickness and composition of the O horizon is driven by multiple factors including climate, soil properties, microtopography, pedoturbation and decomposition of large biomass components (Obrist et al. 2016; Wang X et al. 2019; Landis et al. 2024).

In the studied sclerophyll forest, the combination of high humidity during winter and relatively low temperatures facilitates the accumulation of OM in the surface soil. Wet sclerophyll forests typically have deep litter layers and very large trees (Ashton 1975) compared to other Australian forest types. The litter on the forest floor is made up of portions of fern fronds, large leathery leaves from shrubs and eucalypts, twigs, branches and strips of bark. None of these break down quickly, despite the usually moist conditions at ground level, so the litter and the organic soil beneath it becomes very deep and friable.

This indicates that sites with higher OM content tend to have higher Hg concentrations, indicating that OM plays a key role in Hg retention and accumulation in soils of sclerophyll forest in Victoria. The positive correlation between Hg and OM in our study is likely a result of Hg’s strong affinity for OM due to its ability to form stable complexes with organic ligands (Gabriel and Williamson 2004, Ravichandran 2004). The binding of Hg to OM is a well-documented process that facilitates its retention in soil and sediment matrices (Grigal 2003; Miretzky et al. 2005; Skyllberg 2010). The vertical distribution profiles of Hg and OM are often closely linked, with higher concentrations typically found in surface soil. This relationship is commonly observed in podzol soils of boreal and temperate regions where the highest Hg concentrations generally occur in the uppermost soil horizon (Grigal 2003; Obrist et al. 2016) – a pattern that is also evident in our study. The relationship between Hg and OM is often used to assess Hg dynamics in soil (Grigal 2003).

In our study, the amount of Hg in OM was calculated as 0.28 µg Hg g⁻¹ OM at the tree base, 0.45 µg Hg g⁻¹ OM in the O horizon, and 0.56 in the A horizon. For surface organic soils of temperate and boreal forests an average of 0.22 µg Hg g⁻¹ OM was reported by Grigal (2003). An increasing stoichiometry is also frequently observed with depth, which is attributed to less loss of Hg than C during the mineralisation of OM (Grigal 2003).

Mercury in soil at the base of trees

Mercury concentrations at the base of trees were significantly higher than in soil samples taken in between trees. This difference is likely related to the higher OM content also observed in the tree base region. Both the increased OM content and elevated Hg concentrations near the tree base that has high root density have been linked to the activity of soil fungi mycelium (Read and Perez-Moreno 2003). Previous studies have demonstrated that arbuscular mycorrhizal fungi can enhance Hg sorption in soil, suggesting that mycorrhizal associations may play a role in the higher Hg accumulation observed at tree bases (Liu et al. 2010).

Further work is needed to explore the extent of regions with elevated OM and Hg concentrations around the base of trees in sclerophyll forests of Australia. Not only is this information needed for accurately estimating Hg emissions from prescribed burns, but it is also integral in assessing wildfire emissions, which may significantly affect Hg concentrations in soils.

Mercury concentrations in litter

The mean Hg concentrations in leaf litter, bark and twigs from this study are in good agreement with the few data previously published for Eucalyptus in Victoria (Table 5). The mean Hg concentration in leaves in this study is 64 ± 9 ng g⁻¹; this value exceeds both the global mean (54 ± 22 ng g⁻¹) and median (46 ng g⁻¹) reported by Wang X et al. (2016b) across 168 forest sites worldwide.

Mercury concentrations in both foliage and litter are often strongly correlated with atmospheric Hg levels. In vascular plants, ~90% of the Hg found in leaves and needles is reported to be derived from the atmospheric uptake of gaseous elemental Hg (Zhou et al. 2021). Besides atmospheric Hg concentration, various environmental factors such as solar irradiation, air temperature, altitude and biological factors, including plant species, leaf age and leaf position, significantly influence Hg uptake by foliage (Ericksen et al. 2003; Blackwell and Driscoll 2015; Zhou et al. 2016).

Given that atmospheric Hg concentrations in Australia are relatively low compared to global averages (United Nations Environment Programme and Arctic Monitoring and Assessment Programme 2019; Fisher and Nelson 2020), the elevated Hg concentrations observed in Eucalyptus species may be attributed to unique biological factors, linked to the unique climate and soils in Australia. This influence of biological factors on Hg uptake is exemplified by studies reporting elevated Hg concentrations in evergreen broadleaf forest biomes (Wang X et al. 2016a; Zhou and Obrist 2021). Notably, the highest Hg concentrations in leaves and litterfall have been recorded in the Amazonian tropical forest, with mean and median concentrations of ~70 ng g⁻¹ (Fostier et al. 2015; Wang X et al. 2016a), and in the subtropical evergreen Atlantic Forest, where annual concentrations range from 167 to 334 ng g⁻¹, with a mean of 238 ± 52 ng g⁻¹ (Teixeira et al. 2012). Eucalypt forests are also evergreen and broadleaf. There is a critical need for further research on Hg dynamics in Eucalyptus species in Australia, as well as a deeper understanding of atmospheric Hg patterns across the country.

Emission factor

For the burned litter, the overall Hg EF (19.8 ng Hg g⁻¹ burned litter) was low when compared to EF for wildfires, which range from 315 ng Hg g⁻¹ for Boreal and Equatorial Asia to 41 ng Hg g⁻¹ for grasslands (Kumar et al. 2018). However, it is important to note that the intensity of wildfires is usually much higher than that of prescribed fires, and thus results in soil Hg emissions which were not observed in the present study. Conversely, our estimate is in the same range as the EF previously reported by Howard et al. (2019) (28.7 ± 8.1 ng g⁻¹) as the upper estimate of Hg emission from Australian dry sclerophyll surface fuels for wildfires. During forest fire, Hg is mainly emitted as gaseous elemental mercury (GEM) and particulate bound mercury (PBM), however, many factors can influence their proportion such as moisture and fuel nature and fire intensity (Obrist et al. 2018). Furthermore, part of GEM can be absorbed on ash and PBM is easily deposited at short distance (De Simone et al. 2017; Richter et al. 2023), therefore limiting long distance atmospheric transport. However, fate and transport of Hg emitted during biomass burning are understudied and more work needs to be done to improve this understanding.

Future directions

To improve our understanding of Hg emissions from forest burning in sclerophyll ecosystems, future research should address several key areas:

By addressing these gaps, future research can enhance the precision of Hg emission estimates and contribute to a more comprehensive understanding of Hg dynamics in Australian forests. The relatively high Hg concentrations found in litter leaves, highlights the need for further research on Hg uptake mechanisms in Australian sclerophyll forests.