Rethinking Termite Methane Emissions: Does the Mound Environment Matter?

ABSTRACT

Termites are important decomposers in tropical ecosystems and emit methane (CH₄) as they digest plant material. Global estimates of termite‐derived CH₄ are calculated using termite emission factors (TEF, measured from individuals) and estimated biomass. However, this approach overlooks how the termite mound, via internal and external factors, may influence emissions to the atmosphere. Termite feeding habits, mound methanotrophs and mound structure (internal environment), as well as temperature and season (external environment) can influence net CH₄ emission but remain unparameterized. We investigated how these factors shaped CH₄ emissions from three dominant mound‐building termite species (Coptotermes acinaciformis, Nasutitermes magnus, and Amitermes laurensis) in a northern Australian savanna across four seasons. We compared species‐level TEFs and emissions at the mound‐ and landscape‐scales to evaluate relative species contributions, both with and without accounting for the internal and external environment. We hypothesized that larger, thinner‐walled mounds would emit greater CH₄, and that emissions would be higher at high temperatures and during wet seasons. We expected greater emissions with lower abundances of methanotrophs and pmoA gene copies (involved in CH₄ oxidation) in mound material. Coptotermes acinaciformis individuals had the highest TEFs (1.07 μg CH₄ g⁻¹ termite h⁻¹), N. magnus mounds emitted the most CH₄ (3426 μg CH₄ h⁻¹ m⁻²) and A. laurensis had the highest emissions at the landscape scale (1.04 × 10⁻⁹ Tg CH₄ ha⁻¹ year⁻¹). CH₄ emissions increased with temperature and were highest in the wet‐to‐dry transition season. Mound structure, bacterial methanotroph communities, and pmoA abundance had no effect on CH₄ emissions. Our results highlight the limitations of relying solely on TEFs to estimate contributions of termites to global CH₄ emissions and emphasize the importance of incorporating external environmental conditions, while further exploring internal mound processes. This information allows more accurate parameterization of termite CH₄ contributions to savanna carbon and global CH₄ budgets.

Many factors can influence the amount of methane (CH₄) that is released from a termite mound. In this study, we tested how the external and internal environment of a termite mound impacts CH₄ emission using field measurements of mounds in a Northern Australia savanna. We explored external environmental factors, such as temperature and season, as well as internal factors, such as termite feeding habits, mound structure, and methanotrophic microbial communities. Ultimately, we found that external environmental factors played a stronger role in predicting CH₄ emissions, and this information can be used in upscaling the impact of termite CH₄ emissions at global scales.

1. Introduction

Termites are key decomposers, especially in the tropics, dominating terrestrial arthropod biomass at an estimated global biomass of 100 Mt. (Rosenberg et al. 2023) and cycling significant quantities of terrestrial carbon (Seibold et al. 2021; Griffiths et al. 2019). As a by‐product of carrying out decomposition, termites release carbon dioxide (CO₂), but also methane (CH₄). CH₄ is produced by symbiotic methanogens (Archaea) in termite guts as they break down wood, organic matter in soils, and dead plant matter (Zimmerman et al. 1982; Brune and Dietrich 2015). CH₄ is a powerful greenhouse gas that contributes to climate warming (IPCC 2021), but there exist many uncertainties in quantifying natural CH₄ sources (i.e., emissions) and sinks (i.e., removal, conversion, or storage) that comprise the CH₄ budget (Kirschke et al. 2013; Saunois et al. 2020). Termite CH₄ is one such poorly parameterized component of the global CH₄ budget (Jamali, Livesley, Grover, Dawes, et al. 2011; Khalil et al. 1990) current estimates show that termites emit 8–21 Tg CH₄ per year (Ito 2023). However, this value is highly debated (Zimmerman et al. 1982; Saunois et al. 2020), since variation in termite ecology is not captured in present estimates (Law, Allison, et al. 2024). To resolve such uncertainties, a more nuanced understanding of the underlying factors contributing to termite CH₄ emissions is required (Law, Allison, et al. 2024).

Estimates for global termite CH₄ emissions are typically derived by multiplying a termite emission factor (TEF: average CH₄ emission per termite biomass, μg CH₄ g⁻¹ termite h⁻¹) by regional‐scale estimates of termite biomass (Khalil et al. 1990; Ito 2023; Saunois et al. 2020). Species variation in TEFs spans a wide range, from negligible emissions to more than 25 μg CH₄ g⁻¹ termite h⁻¹, which is largely attributed to differences in feeding groups (Sanderson 1996; Zhou et al. 2022) and termite gut microbial composition (i.e., the presence of CH₄‐producing Archaea, Rouland et al. 1993). It has been shown that soil feeders (Cubitermes sp.) and grass feeders (Trinervitermes sp.) have high TEFs (Sanderson 1996; Zhou et al. 2022), potentially due to higher archaeal‐to‐bacterial ratios in their guts (Rouland et al. 1993; Arora et al. 2022). Additionally, accurately estimating global termite biomass is a challenging venture, due to the diverse sizes and constructions of termite colonies, including epigeal (above ground) mounds, subterranean colonies, arboreal nests, and colonies in deadwood (Law, Allison, et al. 2024).

Termite contributions to global CH₄ emissions are therefore estimated using an overly simplistic model that fails to capture the complexities of how termite ecology influences net CH₄ emission to the atmosphere (Law, Allison, et al. 2024). This failure is especially true when considering mound‐building termites, which are important decomposers of wood, soil, plant litter, and grass, particularly in savanna ecosystems (Zanne et al. 2022; Bunney et al. 2024). Tropical savannas often feature large epigeal termite mounds (in Africa: Meyer et al. 1999; Davies et al. 2014, in Australia: D'hont et al. 2021, in South America: Neto et al. 1986, in India: Jouquet et al. 2015), sometimes covering up to 5%–7% of the land area (Levick et al. 2010; Holdo and McDowell 2004). In this way, termite mounds represent a concentration of termite biomass and CH₄ emission (Jamali, Livesley, Dawes, Cook, et al. 2011; Brümmer et al. 2009; Van Asperen et al. 2021; Jamali, Livesley, Grover, Dawes, et al. 2011), which can overcome the oxidative capacity of soils, which are recognized as CH₄ sinks (Eggleton et al. 1999; Dunfield 2007). However, current CH₄ calculations fail to consider mounds at all, and it is therefore necessary to determine how internal components of the termite mound (structure, mound microbial communities) and local external environmental conditions shape mound CH₄ release.

Structural traits of termite mounds define the internal mound environment and shape the capacity for internal gas exchange and diffusion to the atmosphere (Ocko et al. 2019). Wall thickness is one such trait that varies widely across species and locations (Korb 2011). Greater wall thickness, while more protective from the outside environment and organisms trying to access the mound, can limit diffusion and result in an accumulation of gases, such as CO₂, inside the mound (Korb 2003). Further, when termite mounds grow, they can support a larger termite population (Josens and Soki 2010), with the potential to generate higher total mound CH₄ emissions (Van Asperen et al. 2021; Jamali, Livesley, Dawes, Hutley, and Arndt 2011; Vesala et al. 2023). Therefore, identifying how wall thickness and mound size influence CH₄ emission will provide insight on the importance of incorporating these factors in CH₄ upscaling efforts.

Additionally, external local environmental conditions such as temperature and season influence termite activity and CH₄ emissions. Termite wood decomposition rates are highly sensitive to increasing temperatures, which could result in increased CH₄ emissions (Zanne et al. 2022; Law, Flores‐Moreno, et al. 2024). Direct relationships between termite CH₄ emission, decomposition activity, and temperature remain unexamined, but previous research suggests that mound CH₄ emissions are highest at the warmest part of the day (Jamali, Livesley, Dawes, Cook, et al. 2011; Räsänen et al. 2023). Seasonality has also been shown to affect termite mound CH₄ emissions: previous research in an Australian tropical savanna showed significantly higher mound CH₄ emission during the wet season than for other times of the year (Jamali, Livesley, Dawes, Cook, et al. 2011), which could be due to changes in termite biomass (Jamali, Livesley, Dawes, Hutley, and Arndt 2011) or temperature. However, other studies found mound emissions to be highest in the dry season (Quevedo et al. 2021; Räsänen et al. 2023), which could result from decreased oxidation of methane by microbial activity resulting from moisture limitations. Determining the role of temperature and seasonality in termite mound CH₄ emissions is therefore necessary for accurate annual CH₄ emission estimates.

Finally, CH₄ released from individual termites residing within a mound may not reach the atmosphere due to the presence of methanotrophic (CH₄‐consuming) bacteria and archaea within termite mounds. Methanotrophic bacteria have been shown to reduce the net emission of CH₄ from the mound to the atmosphere by 50% (Nauer, Hutley, and Arndt 2018). The mound microbial community can therefore influence net CH₄ emission, which is known to vary across termite species (Ho et al. 2013; Chiri et al. 2020). The gene encoding the particulate methane monooxygenase subunit A (pmoA) is involved in the CH₄ oxidation pathway for most bacterial methanotrophs (Cupples and Thelusmond 2022). The pmoA gene plays an important role in CH₄ metabolism, and its abundance in termite mound material is a standard method for identifying methanotrophic microbe communities (Chiri et al. 2020). However, direct associations between methanotroph community composition, pmoA gene abundance, and termite mound CH₄ emission are yet to be tested.

In this study, our goal was to refine our understanding of termite contributions to CH₄ emissions by parsing out various factors that influence net mound CH₄ emission to the atmosphere. To do so, we combined previous measures of mound abundances (Clement et al. 2021) with new measures of species‐level TEFs and mound‐level CH₄ emissions from three dominant mound‐building termite species, with varied feeding habits and mound traits, in North Queensland, Australia tropical savannas. We asked the following questions: 1. How do species‐level TEFs contrast with average mound‐level CH₄ emissions and landscape‐scale CH₄ emissions? 2. What is the relative influence of species‐specific mound structural traits (mound size and wall thickness) and external environmental drivers (temperature and season) on mound‐level CH₄ emissions? 3. Do methanotrophic microbial communities (archaea and bacteria) and methanotroph gene (pmoA) abundance in the mound material influence CH₄ emission within and across termite mounds?

2. Methods

2.1. Study Site

Termite mounds were sampled at the Australian Wildlife Conservancy (AWC) Brooklyn Sanctuary in the Station Creek region (−16.61 S, 145.24 E, Figure 1). The Station Creek region comprises a wet savanna ecosystem with distinct wet and dry seasons, experiencing 1728 mm of rainfall annually (Cheesman et al. 2018). Floristically, the region is dominated by Eucalyptus cullenii, Corymbia clarksoniana, Acacia disparrima subsp. Calidestris, and Larsenaikia ochreata (Flores‐Moreno et al. 2024). Termite diversity and abundance in this region are relatively high for savanna ecosystems, with 14 species described to occur in the area (Clement et al. 2021). Amitermes laurensis, Coptotermes acinaciformis, and Nasutitermes magnus are abundant mound‐building termite species (Clement et al. 2021) and were selected as the focal species of this study. Coptotermes acinaciformis feeds on wood and typically builds medium to large‐sized mounds with a dome shape (mean height = 87 cm, Clement et al., unpublished data), which are often built at the base of trees which they frequently hollow (Yatsko et al. 2024, 2025). Nasutitermes magnus feeds on grass and builds large mounds (mean height = 92 cm, Clement et al. unpublished data). Amitermes laurensis feeds on plant litter and builds small conical mounds (mean height = 59 cm, Clement et al. unpublished data) that have high abundance in the landscape (see Figure 1 for mound morphologies).

FIGURE 1.

(a) Study location in far North Queensland, Australia (white square). (b) Location of sampled termite mounds repeatedly measured across all four seasons (colored by species, n = 19 total, n = 7 A. laurensis, n = 7 C. acinaciformis , n = 5  N. magnus ) in the greater Station Creek region. Termite mound morphologies common to (c) A. laurensis, (d) C. acinaciformis , (e) N. magnus . White scale bars on each species mound morphology image indicate a reference height measurement of 50 cm. Mean dimensions of each species' mound can be found in Table S13.

2.2. Determining Species‐Level Termite Emission Factor With Laboratory Measurements

In February 2024 we collected mound material from the field and conducted laboratory measurements to calculate a TEF for each species. To do so we removed entire A. laurensis mounds (n = 5) from the field due to their small size. For C. acinaciformis (n = 5) and N. magnus (n = 5), mounds were too large, so we subsampled mound pieces using a saw and chisel. Mound samples were placed in a bucket and transported back to laboratory conditions where they equilibrated to room temperature (24°C) for a minimum of 12 and a maximum of 24 h prior to CH₄ emission measurement. We cut mound samples into smaller pieces (approximately 5 × 5 cm) and removed all termites from the mound piece using entomological forceps (mean termite biomass = 2.2 g) to measure CH₄ emissions directly from termite individuals. CH₄ emissions were measured from combined samples of worker and soldier termites. We note that each species had a different ratio of workers to soldiers (A. laurensis: 99% workers, 1% soldiers. N. magnus : 75% workers, 25% soldiers. C. acinaciformis : 56% workers, 42% soldiers). We measured CH₄ emissions using a Los Gatos Research (LGR) Ultraportable Greenhouse Gas Analyzer (UGGA, San Jose, CA, United States) connected to a closed sampling chamber made from PVC and plexiglass (LI‐COR 8200–104, Lincoln, Nebraska, United States) with a volume of 1570.8 cm³. Measurements were taken over the course of 300 s following a 60 s calibration (Figure S3). We divided emissions by termite biomass (g, wet mass) to calculate the species‐level TEF, defined as mean μg CH₄ g termite⁻¹ h⁻¹ (Zhou et al. 2022).

2.3. Termite Mound CH₄ Emission Field Sampling and Species Identification

Termite mound CH₄ emissions were sampled in the field (separate from laboratory measurements described above) at four times between May 2022 and February 2024 to capture the range of seasonal conditions and temperatures characterizing the savanna ecosystem (Table S1). We selected mounds for each of the three termite species based on the following criteria: (a) mounds were accessible by vehicle across weather conditions, (b) mounds were active and occupied by termites at the time of measurement, (c) mounds were representative of the range of sizes for a given species, and (d) mounds were in locations with a low likelihood of human interference. The location of each mound was recorded with GPS coordinates.

The initial survey in May 2022 (wet‐to‐dry transition season) targeted a larger number of mounds per species (A. laurensis: n = 14, C. acinaciformis : n = 25, N. magnus : n = 21, Table S2). Then, eight mounds from each species were randomly selected from the May 2022 campaign for remeasurement in November 2022 (dry‐to‐wet season transition), August 2023 (mid‐dry season), and February 2024 (mid‐wet season). Over the course of sampling, there were five instances where mounds died or were destroyed and could not be re‐measured (Table S2). When this was the case, the mound was replaced with a mound closest in size from the original May 2022 sampling pool. In February 2024, we were only able to re‐measure seven N. magnus mounds due to logistical constraints of accessing mounds following Cyclone Jasper which affected the region in December 2023. At the end of the experiment, seven A. laurensis, seven C. acinaciformis, and five N. magnus mounds were re‐measured during all four seasons (Table S2, Figure 1). These mounds were used to analyze the effect of season and temperature on mound CH₄ emission.

Two soldiers and five workers were sampled from each mound in the May 2022 campaign and stored in 95% ethanol. Soldier sample morphology was used to confirm species‐level identification (P. Eggleton, personal communication, May 2023). When visual identification was not possible, we followed the protocol in Clement et al. (2021) for species identification using DNA barcoding of the COII gene.

2.4. Field‐Based Measurements of Termite Mound CH₄ Emissions

We sampled CH₄ emissions from termite mounds in the field using a semi‐closed chamber‐based sampling system connected to the LGR UGGA (Figure S1). For each mound, 4–5 surface measurements were taken by attaching the LGR UGGA to a semi‐closed PVC sampling chamber (Jeffrey et al. 2020) via gas tubing. Sampling chambers were placed on the mound surface without breaking the mound wall (Figure S1a) and secured with an external airtight seal using a ring of inert potting clay (Walker Ceramics, Bayswater, VIC, Australia; Jeffrey et al. 2020, Figure S1b).

Sampling chambers were fixed on the flattest surface of the mound to accurately calculate chamber volume which was approximated as a cylinder. Different sized chambers were used for the three species, as A. laurensis mounds were smaller (chamber volume = 2960.1 cm³, surface area = 183.9 cm²) and C. acinaciformis and N. magnus mounds could fit medium chambers (chamber volume = 899.9 cm³, surface area = 83.3 cm²). We standardized the mound sampling points to optimize for the unique geometries of each mound. Coptotermes acinaciformis and N. magnus mounds were sampled in five locations: north‐facing side at 50% mound height, east‐facing side on the flattest surface closest to the mound base, south‐facing side at 50% mound height, west‐facing side on the flattest surface closest to the base, and the top of the mound at the flattest surface (Figure S1c). Due to their conical shape, A. laurensis mounds were sampled in the first four locations but with the top measurement omitted, since there was no flat surface available. Individual sampling points were marked so that the sampling position could be repeated during remeasurement.

After chambers were fixed to the mound and sealed with clay, CH₄ emission was recorded for 120 s and internal chamber air temperature (°C, temperature at the mound surface) was simultaneously recorded every minute using a thermocouple probe (PerfectPrime TC41, 4‐Channel K‐Type Digital Thermometer Thermocouple Sensor). All measurements were made between 08:00 and 17:00. Measurement quality assessed using two criteria: (1) no humming noise coming from the sampling chamber, which indicated that the clay ring seal was airtight and free of leaks (Jeffrey et al. 2020), and (2) real‐time measurements on the LGR UGGA displayed positive, linear slopes, indicating constant production of CH₄ (indicating termite occupation) and no leaks. Following CH₄ emission measurement, a small area of the mound was broken to access galleries and confirm termite occupation.

2.5. CH₄ Emission Calculation

CH₄ emissions from termite mounds and individual termites (μg CH₄ h⁻¹) were calculated using linear regression slopes (x = time, y = [CH₄] in chamber headspace) following the ideal gas law and using average chamber air temperature for each measurement and assuming ambient pressure. R ² and p‐values were calculated for each measurement, and samples with R ² < 0.4 were manually inspected for erroneous points. Erroneous points often were present at the beginning or end of the measurement due to the chamber opening or closing and were omitted when present.

Positive slopes indicated net CH₄ emission, while negative slopes indicated that CH₄ oxidation exceeded production. In our data collection and analysis, we only considered termite mounds to be alive if the average mound emission was a positive value. However, within an individual mound it was possible for some subsamples to show negative slopes, indicating net CH₄ oxidation. For analyses where emissions from subsampled areas of the mound were the unit of measure (i.e., mound microbial communities, see below), negative values for CH₄ emission were included in the analysis. For analyses run at the mound‐level (i.e., structural equation modeling, see analyses below), we only considered live mounds and therefore CH₄ emission was always positive. CH₄ emissions for each subsample were divided by the surface area covered by the sampling chamber to derive emissions per unit mound surface area (μg CH₄ h⁻¹ m⁻²). We assessed variation in emissions within mounds by calculating a Coefficient of Variation (CV, Table S12), but as our second research question was not focused on intra‐mound variation, we used a mound‐level average in subsequent analyses. Total mound emission was derived by multiplying average mound emission per unit mound surface area by total surface area of the mound (μg CH₄ h⁻¹ mound⁻¹).

2.6. Mound Size and Wall Thickness Measurement

We used photogrammetry to estimate mound surface area and volume. Photogrammetry uses multiple photographs taken from different angles to construct a digital three‐dimensional model of the scanned object (Nauer, Chiri, et al. 2018). To take photogrammetry measurements we first removed debris surrounding the mound and matted down any grass obscuring the mound. Starting at the north‐facing side of the mound, consecutive photos were taken on an Apple iPad Pro (Cupertino, California, United States) as the person scanning walked clockwise around the mound, holding the iPad steady. The top of the mound was also scanned when it was possible; on occasion this was constrained by mound height. Photos were digitally assembled in the Polycam app (Polycam Inc.) to create a three‐dimensional photogrammetric termite mound model.

Individual mound models were exported as .glb files from Polycam app and imported into MeshLab to remove non‐termite mound material using the selection tool. Total mound surface area was calculated in MeshLab using the Compute Geometric Measures function. To calculate total mound volume, cleaned mound models were imported into Blender as. obj files (Blender Online Community 2022). Each mound was first filled with the grid fill command, and volume was calculated using the 3D print toolbox volume function. One limitation of the mound volume estimation method is that we were not directly able to quantify the termite biomass within each mound.

Mound wall thickness was recorded at the final measurement campaign (February 2024) due to the invasive nature of the measurement. After sampling CH₄ emissions, we used a drywall hammer or pickaxe (depending on mound hardness) to break the mound wall at each sampling point. We used digital calipers with 0.01 mm precision to record wall thickness, which was defined as the distance from the external wall to where termite galleries were visible.

2.7. Mound CH₄ Emissions Estimated at the Landscape Scale

We used data from Clement et al. (2021) which reports landscape abundance of occupied termite mounds per hectare for the three study species and at the Station Creek site. We calculated average annual mound CH₄ emission for each of our study species (Tg CH₄ year⁻¹ mound⁻¹) and then multiplied this value by mound density per hectare (Clement et al. 2021) to derive an estimate of landscape‐scale termite mound emissions for northern Australian savanna ecosystems (Tg CH₄ year⁻¹ ha⁻¹). It is estimated that tropical savannas in northern Australia cover approximately 1.9 million km² (Chen et al. 2003), and termite mounds are characteristic features of this landscape, although it is noted that the proportion of dominant species may change for different locations in Australian tropical savannas.

2.8. Mound Microbial Community Sampling and Metagenomic Sequencing

In May 2022 we sampled termite mound material to link microbial communities directly with CH₄ emissions. Following CH₄ emission measurement, the area covered by the chamber was broken to sample internal mound material (where termite galleries were, Chiri et al. 2020) using a sterilized chisel and forceps. Mound substrate was placed into a 5 mL soil vial and subsequently frozen at −20°C.

We selected a subset of samples for microbial community analysis based on their suitability to address two questions: is microbial community variation associated with variation in CH₄ emission (1) between mounds and (2) within mounds? To test drivers of variation between mounds, for each of the three study species we selected the four mounds with the highest average CH₄ emission and the four with the lowest average CH₄ emission. For these mounds (n = 8 per species, n = 24 total mounds), we selected three subsamples within each mound that were closest to the average CH₄ emission for that mound for metagenomic sequencing (n = 72 total subsamples). To test drivers of variation within mounds, we first calculated the CH₄ emission range for each mound and then identified the five mounds with the highest variation for each species, indicated by the greatest CH₄ emission range. For these mounds we selected all subsamples (4 or 5) collected (n = 5 mounds per species, = 15 total mounds, n = 69 total subsamples). In total, we extracted DNA from and sequenced 115 mound samples, and there was overlap in which some samples were considered for both questions (i.e., a highly variable mound could also be a top‐emitting mound).

DNA extractions were carried out with the DNeasy PowerSoil Pro Kit (Qiagen, United States) using the manufacturer's instructions with an input of approximately 250 mg mound material. Mound material was homogenized prior to extraction using a FastPrep‐24 5G Homogenizer MP Biomedicals (MP Biomedicals, United States; settings: sediment—soils/rocks, Speed: 5.5 m s⁻¹, Adapter: quickpro, Time: 30 s, Lysing matrix: E, Quantity: 50 mg, Cycles: 2, Pause time: 300 s). DNA concentration (ng μL⁻¹) and quality (260/280 and 260/230 values) were evaluated with a NanoDrop spectrophotometer (Thermo Fisher Scientific, Massachusetts, United States).

Metagenomic sequencing was performed at the Ramaciotti Centre for Genomics (UNSW, Sydney, Australia) on a NovaSeq X Plus 10B (2 × 150 bp; Illumina). Raw reads were first quality controlled using fastp v0.23.4 (Chen et al. 2018), then go reads error correction using the Bayesian‐Hammer as implemented in SPAdes v4.0.0 (Prjibelski et al. 2020). Resultant corrected reads were assembled using Megahit v1.2.9 (Li et al. 2016); only contigs over 1 kb were retained for gene prediction. Parallelized version of Prodigal—pprodigal (Hyatt et al. 2010; Jaenicke 2022) was used for gene prediction; nonredundant genes (protein translations) were generated using CD‐HIT v4.8.1 (Li and Godzik 2006), with parameters setting as ‐c 0.9 ‐M 0 ‐T 0 ‐l 20 ‐G 0 ‐aS 0.9 ‐g 0. KEGG Orthology (KO) and pathway analysis was performed using KofamKOALA (Aramaki et al. 2020 KofamScan v1.3.0 with the latest KOfam database v2024‐08‐29). Resultant tables (reads count per sample) including KO and pathway analysis were generated using CoverM (Aroney et al. 2024).

2.9. Identifying Methanotrophic Microbes and Genes From Metagenomic Sequencing

From the metagenomic analyses described above, we targeted the pmoA gene as it plays a direct role in CH₄ metabolism (Zhou et al. 2015) and has been previously associated with termite mound CH₄ oxidation (Chiri et al. 2020). We removed five samples with low total read counts (< 300 reads) compared to the average read count across the dataset (mean = 18,762,769 reads). To account for variation in sequencing depth, we then resampled reads for the pmoA gene (KO number “K10944”) to standardize total reads to the minimum value of the dataset (6,913,446 reads). We calculated the relative abundance of the pmoA gene to use in subsequent analyses.

We identified methanotrophic bacterial genera from various literature sources and then queried our metagenomic sequencing data outputs for these 16 genera (Table S11). From this list, 10 methanotroph genera were present in our mound samples. Regarding methanotrophic archaea, these genera are less well described in comparison to bacteria, although the Candidatus methanoperedens genus is known to be associated with methanotrophy (Bhattarai et al. 2019). However, we did not detect any methanotrophic members in the archaea metagenomic dataset, and therefore focused our analyses on the bacterial methanotroph community composition at the genus level.

2.10. Analyses

2.10.1. CH₄ Emissions at the Individual Termite and Mound‐Level

We used Analysis of Variance (ANOVA) to evaluate differences in TEFs of the three termite species. We used the R package emmeans (Lenth 2025) to perform pairwise species comparisons with Tukey's HSD adjustment for multiple testing. To evaluate differences among species in termite mound CH₄ emission, we used a linear mixed effect model from the R package lmer (Bates et al. 2015). In the model, CH₄ emission was the response, species was a fixed effect, and mound individual was included as a random effect to account for multiple measurements for each mound. In a post hoc analysis, we compared estimated marginal means among species using Tukey's HSD adjustment for multiple comparisons.

2.10.2. Influence of Species‐Specific Mound Structural Traits and External Environmental Drivers on Mound‐Level CH₄ Emission

We used piecewise Structural Equation Modeling (SEM) to explore the direct and indirect drivers of mound‐level CH₄ emission from termite mounds. Specifically, we modeled how mound volume and mean wall thickness mediated the relationship between termite species and CH₄ flux, while also accounting for temperature and season effects. The SEM was built using the following three equations: (1) Mound volume (V) as a function of species (SP) in a linear model, (2) Mean wall thickness (W) as a function of species in a linear model, and (3) Mean CH₄ emission (E) as a function of all predictors (mean wall thickness (W), volume (V), species (SP), temperature (T), and season (S)) in a random effects model with individual mound as the random effect:

$$\mathrm{V}={\beta }_{0,\mathrm{V}}+{\beta }_{1,\mathrm{V}}\times \mathrm{SP}+{\epsilon }_{\mathrm{V}}$$ (1)

$$\mathrm{W}={\beta }_{0,\mathrm{W}}+{\beta }_{1,\mathrm{W}}\times \mathrm{SP}+{\epsilon }_{\mathrm{W}}$$ (2)

$$\mathrm{E}={\beta }_{0,\mathrm{E}}+\left({\beta }_{1,\mathrm{E}}\times \mathrm{V}\right)+\left({\beta }_{2,\mathrm{E}}\times \mathrm{W}\right)+\left({\beta }_{3,\mathrm{E}}\times \mathrm{SP}\right)+\left({\beta }_{4,\mathrm{E}}\times \mathrm{T}\right)+\left({\beta }_{5,\mathrm{E}}\times \mathrm{S}\right)+{\epsilon }_{\mathrm{E}}$$ (3)

where β _0,x is the model intercept for a given predictor variable (x), β _x is the slope of each predictor variable, and ε _y is the error of the response variable (y). In this analysis, mound volume and mean wall thickness predictors were scaled (mean‐centered and divided by standard deviation).

The SEM was built using the psem function from the piecewiseSEM R package (Lefcheck 2016). We assessed model fit using the Fisher's C statistic, and directed separation tests were used to verify that there were no important missing paths. Individual component models in the SEM were evaluated using standardized coefficients, confidence intervals, and p‐values.

2.11. Mound Methanotroph Communities, Genes, and Influence on CH₄ Emission: Variation Between Mounds

For the subset of metagenomic data that captured variation between mounds, we explored patterns in the composition of methanotroph genera within termite mounds using a Principal Coordinates Analysis (PCoA). The analysis was based on a Bray–Curtis dissimilarity matrix of the mound methanotroph community and used the wcmdscale function from the vegan R package (Oksanen et al. 2025) with 2 dimensions (k = 2). We calculated weighted genera scores, and the resulting PCoA1 and PCoA2 scores were used to explore relationships between methanotroph community structure, termite species, and variation in CH₄ emission.

We then tested if methanotroph communities were predicted by termite species or by CH₄ emission using permutational multivariate analysis of variance (PERMANOVA) with the adonis2 function (Oksanen et al. 2025), including 999 permutations to assess significance. Then, we used a linear mixed effect model to determine the relationship between methanotroph communities (summarized by PCoA1 and PCoA2 scores) and the relative abundance of the pmoA gene. In the model, PCoA1, PCoA2, and their interaction were predictors, and pmoA relative abundance was the response. Termite mound individual was included as a random effect.

Finally, we used a linear mixed effect model to determine if methanotroph community and gene predictors influenced CH₄ emission. Model predictors included the relative abundance of the pmoA gene, termite mound species, the interaction between PCoA1 and PCoA2 scores, and the sum of all methanotroph genera relative abundances. Mound individual was included as a random effect. To explore this same relationship for each individual methanotroph genus, we ran separate models for each genus where the relative abundance of an individual methanotroph genus and termite mound species were predictors, CH₄ emission was the response, and mound individual was a random effect.

2.12. Mound Methanotroph Communities, Genes, and Influence on CH₄ Emission: Variation Within Mounds

For the subset of metagenomic data that captured variation within mounds, we carried out the same PCoA as described above to test PCoA1 and PCoA2 scores (including their interaction) as well as relative abundance of the pmoA gene, termite mound species, and the relative abundance of the sum of all methanotroph genera as predictors of CH₄ emission. Mound individual was included as a random effect. All analyses were conducted in R version 4.3.2.

3. Results

3.1. Scaling CH₄ Emissions From Individual Termite‐ to Mound‐ to Landscape‐Levels

TEF values for individual termites (μg CH₄ g⁻¹ termite h⁻¹) varied significantly by species (F _(2,12) = 29.41, p < 0.001) with C. acinaciformis (mean TEF = 1.07) having a higher TEF than A. laurensis (mean TEF = 0.73), which had a higher TEF than N. magnus (mean TEF = 0.37). All pairwise comparisons were significant (Figure 2a, Table S3). Average CH₄ emissions from termite mounds (per unit surface area) also significantly varied by species (F _(2,125) = 16.74, p < 0.001) where N. magnus (3426 μg CH₄ h⁻¹ m⁻²) had significantly greater emissions than both A. laurensis (1270 μg CH₄ h⁻¹ m⁻²) and C. acinaciformis (1323 μg CH₄ h⁻¹ m⁻²) but the latter two species did not significantly differ from one another (Figure 2b, Table S4).

FIGURE 2.

(a) Termite emission factors (μg CH₄ h⁻¹ g termite⁻¹) for the study species (purple = A. laurensis, green =  C. acinaciformis , yellow =  N. magnus ). (b) Average mound‐level CH₄ emission by mound surface area (μg CH₄ h⁻¹ m⁻²) for the three study species. Boxplot center lines indicate the median and whiskers show ±1.5 IQR, and asterisks indicate significantly different pairwise contrasts. (c) Bar plot of landscape‐scale mound CH₄ emissions (Tg CH₄ year⁻¹ ha⁻¹) by species. Bar plot error bars indicate one standard deviation.

At the landscape scale, we estimated a combined annual emission of 2 × 10⁻⁹ Tg CH₄ ha⁻¹ y⁻¹ from termite mounds of our three study species (Table S5). Amitermes laurensis mounds contributed the most CH₄ on a per hectare basis (Figure 2c, mean = 1.04 × 10⁻⁹ Tg CH₄ ha⁻¹ year⁻¹), N. magnus was the second highest CH₄ emitter (Figure 2c, mean = 8.17 × 10⁻¹⁰ Tg CH₄ ha⁻¹ year⁻¹), while C. acinaciformis (Figure 2c, mean = 1.69 × 10⁻¹⁰ Tg CH₄ ha⁻¹ year⁻¹) contributed the least to landscape mound CH₄ emissions.

3.2. Relative Importance of Structural Traits and External Environmental Drivers in Mound‐Level CH₄ Emission

There was a good overall fit to the data (marginal R ² = 0.30, conditional R ² = 0.68) in the SEM analysis, indicating no significant missing pathways (Fisher's C = 8.372, df = 10, p = 0.593). Higher mound surface temperatures were positively associated with increased emissions (β = 2808.2, p = 0.038, Figure 3, Table S6). Season significantly influenced flux (p < 0.001, Figure 3), with CH₄ emissions peaking during the wet‐to‐dry transition season (β = 4443.2, p < 0.001, Table S6). Termite species had a marginally significant effect on mean CH₄ emissions (p = 0.064, Figure 3), and mounds constructed by C. acinaciformis (β = 2747.3, p < 0.01, Table S6) and N. magnus (β = 5027.0, p < 0.001, Table S6) exhibited significantly higher CH₄ emissions than those built by A. laurensis.

FIGURE 3.

(a) Structural equation modeling diagram showing significant (solid line) and nonsignificant (dashed line) predictors of mound CH₄ emission. Values on each line indicate β values from the SEM, and β values for each level of categorical predictors are included. Conditional plots for (b) temperature (°C) and (c) season visualized with visreg (Breheny and Burchett 2017). Lines represent model predicted values (for each season) and mean mound CH₄ emission, while holding all other predictors constant (from Equation 3). Full SEM results can be found in Table S6.

In contrast, mound volume and mean wall thickness, while hypothesized mediators of mound structure effects on CH₄ emission, were not significant predictors (volume: p = 0.72; wall thickness: p = 0.55, Figure 3). Neither volume nor wall thickness differed significantly among species (Table S6).

3.3. Between Mound Variation in CH₄ Emissions: Bacterial Methanotroph Communities and the pmoA Gene

Methanotrophic bacteria represented a small proportion of the bacterial community (0.25% of the total reads) and were dominated by Methylocystis sp. (Figure S2). Methylomicrobium, Methylobacter, Methylocaldum, Methylococcus, and Methylomonas were associated with high PCoA1 and PCoA2 values, which explained 36.9% and 5.2% of variation in the data, respectively. Methylocapsa was associated with low PCoA1 and high PCoA2 values, while the opposite was true for Methylacidimicrobium, Methylosinus, and Methylocystis genera. Methylocella was the only genus associated with low PCoA1 and PCoA2 values (Figure 4a).

FIGURE 4.

(a) PCoA plot based on Bray–Curtis dissimilarity for genera‐level community composition of methanotrophs in mound material of the three termite study species (denoted by colored points and ellipses). Arrows indicate weighted genera vectors. (b) A significant interaction between PCoA1 and PCoA2 predicting the relative abundance of the pmoA gene (yellow line, PCoA2 = 0.06; green line, PCoA2 = 0.1; purple line, PCoA2 = −0.04).

Overall, there were poor links between bacterial methanotrophs and mound CH₄ emissions. Termite species explained only 1.5% of the variation in methanotroph communities (at the genus level, R ² = 0.015, F = 0.48, p = 0.73) and CH₄ flux explained only 1.4% (R ² = 0.014, F = 0.90, p = 0.37). There was minimal association between CH₄ emission and methanotroph community structure in terms of PCoA1 and PCoA2 (Figure 4a).

We found that pmoA gene relative abundance was negatively related to PCoA1 (β = −7.99, SE = 2.65, df = 55.11, t = −3.01, p = 0.004, Figure 4b) and not significantly related to PCoA2 scores (p = 0.65). However, there was a significant interaction between PCoA1 and PCoA2 (β = −343.33, SE = 154.88, df = 56.00, t = −2.22, p = 0.03, Figure 4b). At high values of PCoA2, there was a negative relationship between PCoA1 and pmoA gene relative abundance, while at low values of PCoA2, there was no significant relationship between PCoA1 and pmoA gene relative abundance.

Finally, CH₄ emissions were not significantly predicted by relative abundance of pmoA, the sum of all methanotroph relative abundances, termite species, or PCoA1 and PCoA2 scores and their interaction (Table S7). Additionally, the relative abundance of any single methanotroph bacterial genus was not significantly related to CH₄ emission (Table S8).

3.4. Within Mound Variation in Methane Emissions: Bacterial Methanotroph Communities and the pmoA Gene

For those mounds with high intra‐mound variation in CH₄ emissions, N. magnus mounds emitted marginally more CH₄ than A. laurensis mounds (β = 3420.91, SE = 1533.97, df = 11.32, t = 2.23, p = 0.047), but no other predictors (relative abundance of pmoA, sum of all methanotroph relative abundances, PCoA1 and PCoA2 and their interaction) were significant (Table S9).

4. Discussion

In this study we sought to understand how the internal and external termite mound environment shaped CH₄ emission by measuring the influence of species‐level differences for various spatial scales, mound structural traits, seasons, temperatures, and methanotroph communities. We showed that the termite species contributing the most to CH₄ emissions differed at the individual, mound, and landscape‐scale. Differences at the landscape‐scale are likely due to mound abundance, and while not directly tested, differences at the mound‐level could be driven by differences in termite biomass. Given that most CH₄ is likely emitted by termites inside mounds, accounting for the internal mound environment should be a critical consideration in calculating net CH₄ emission. Factors controlling termite mound emissions are therefore an important consideration for accurately predicting termite contributions to the CH₄ budget. From our work, mound CH₄ emissions were most strongly influenced by termite species inhabiting the mounds, temperature, and season. Furthermore, while microbial methanotrophy within the mound has been shown to be an important filter controlling net CH₄ release, we did not find direct associations between either the bacterial methanotroph communities or the pmoA gene abundances and CH₄ release. However, the relationship between methanotroph communities and the pmoA gene suggests that expression, rather than relative abundance, could reveal the interplay between termite methanogenesis, bacterial methanotrophy, and net mound CH₄ emission. Below we further explore the findings of this field study and discuss how incorporating scale, species‐level differences, external environmental conditions, and methanotrophic microbes into CH₄ modeling can contribute to accurately estimating termites' role in global CH₄ budgets.

4.1. Species Contributions to CH₄ Emissions Differ Depending on the Scale of Measurement

The scale of measurement mattered for determining the relative role of different termite species in contributing to CH₄ emission. Coptotermes acinaciformis individuals had the highest TEF values, N. magnus had the highest mound‐level emissions, and when landscape‐level mound density was factored into CH₄ emission calculations, A. laurensis mounds released the most CH₄. At the individual scale, variation between species was most likely underpinned by differences in termite diet and the gut microbial community that each species harbors (Clement 2022). While we were not able to directly test the effect of diet on CH₄ emission, there are known differences in how CH₄ varies by feeding group (Sanderson 1996; Zhou et al. 2022), which is ultimately controlled by termite gut microbial composition, specifically the presence and abundance of methanogenic microbes (Rouland et al. 1993). Interestingly, Coptotermes acinaciformis, which had the highest TEF in our study (contrasting Sanderson 1996 findings on wood‐feeders), has been shown to have a greater abundance of an archaeal methanogenic gut microbe (Methanobrevibacter sp.) compared to N. magnus and A. laurensis (from data collected at the same study site, Clement 2022). While we observed high CH₄ production for C. acinaciformis individuals, consistent with a greater abundance of methanogenic gut microbes (Clement 2022), we found no evidence of microbial methanotrophy reducing CH₄ emissions at the mound‐level. To explain such discrepancies between individual and mound‐level emissions, further investigation of the balance of methanogenesis within termites and methanotrophy occurring in the termite mound is necessary.

Differences in top emitting species at the mound and landscape scale were underpinned by tradeoffs between high average mound emissions and high termite mound densities. Nasutitermes magnus mounds were larger (mean surface area = 4.5 m²) and had an average mound emission (3426 μg CH₄ h⁻¹ m⁻²) that was greater than A. laurensis (1270 μg CH₄ h⁻¹ m⁻²). Greater emissions from N. magnus mounds could result from more termite biomass housed in larger mounds, although mound size is an imperfect indicator of colony size (Josens and Soki 2010). However, A. laurensis mounds, although they were smaller (mean surface area = 0.7 m²) were approximately 22 times more abundant compared to N. magnus mounds per hectare, leading to a greater overall contribution by A. laurenesis when CH₄ was estimated annually per hectare (1.04 × 10⁻⁹ > 8.17 × 10⁻¹⁰ Tg CH₄ ha⁻¹ year⁻¹). We propose that in tropical savannas and other ecosystems with high mound abundance, combining entire mound CH₄ measurements with landscape‐scale mound density to calculate a “mound emission factor” per area can better represent ecosystem CH₄ emission, rather than scaling up with a TEF which is common practice (Saunois et al. 2020; Ito 2023).

4.2. Temperature and Season Impact CH₄ Emission More Than Mound Structure

The differences in CH₄ emissions that we found at various scales of measurement (individual, mound, landscape) also emphasize the importance of the “colony context”. Methane emissions from termites are largely released from workers inside the mound. Thus, net CH₄ emissions are not only shaped by production from gut methanogens within termites (Brune and Dietrich 2015), but also by mound structure, microbial methanotrophy, and the external environmental conditions (i.e., temperature and seasonality, Korb 2003; Singh et al. 2019; Nauer, Hutley, and Arndt 2018; Jamali, Livesley, Dawes, Hutley, and Arndt 2011). In our ecosystem, temperature and season emerged as strong predictors of CH₄ emission, contributing to the growing body of literature suggesting termites and termite‐driven decomposition are highly sensitive to external environmental conditions (Zanne et al. 2022; Law, Flores‐Moreno, et al. 2024).

Interestingly, we found that CH₄ emissions in the wet‐to‐dry transition season were significantly greater than the mid‐wet and dry‐to‐wet seasons. Our results revealed far less seasonal variation in mound CH₄ emission compared to previous studies, as wet‐to‐dry transition season emissions were only 28%–33% higher than the dry‐to‐wet and mid‐wet seasons. In contrast, one study in northern Australia reported up to 26‐fold higher emissions in the wet compared to the dry season for Microcerotermes nervosus mounds (Jamali, Livesley, Dawes, Cook, et al. 2011), and another study in Africa showed that wet season emissions were 64% lower than the dry season (Räsänen et al. 2023). Our wet season CH₄ measurements followed Cyclone Jasper, which flooded parts of the site; it is possible that this anomalous inundation altered seasonal termite population dynamics and concealed some of the extreme wet season dynamics in CH₄ emission observed by Jamali, Livesley, Dawes, Hutley, and Arndt (2011). This study and others showed that seasonality can influence CH₄ emission, but across the studies, there is not strong consensus on which season has the greatest emissions. These differences merit a closer investigation of the underlying factors driving seasonal shifts. Field studies investigating fluctuations in the termite mound population with alterations to water and food resources (Jamali, Livesley, Dawes, Hutley, and Arndt 2011), as well as interactions between temperature and moisture availability on the metabolic functioning of both termites and microbes, are suggested avenues of future research.

As temperatures increased so too did mound CH₄ emission, pointing to climatic regulation of termite metabolic activity (Jamali, Livesley, Dawes, Hutley, and Arndt 2011; Zanne et al. 2022), as well as the methanogenic microbes in termite guts. The relationship that we define between temperature and CH₄ emissions can help to parameterize biogeochemical cycling models that could be used to understand how termite mound CH₄ emissions, at savanna and global scales, shift with future climates (Zanne et al. 2022). Our results are especially relevant as studies project that termite‐mediated decomposition rates respond positively to increased temperatures (Zanne et al. 2022), and termite range expansion is another predicted consequence of climate warming (Zanne et al. 2022; Ito 2023). Ito (2023) estimated that from elevated CO₂ (via vegetation productivity), climatic warming, and land‐use change, termite CH₄ emissions could increase between 0.5–5.9 Tg year⁻¹. Given that our study provides species‐specific, field‐based measurements to define a temperature‐emission relationship, this could serve as a foundation to evaluate the validity of such claims.

Additionally, when predicting if emissions will increase due to higher temperatures it is necessary to consider which termite species will expand in range and what nesting behaviors they have (Law, Allison, et al. 2024). Increases in subterranean‐dwelling termites are unlikely to affect net CH₄ emission (Eggleton et al. 1999), but mound‐building species are better adapted to temperature change as the mound structure can maintain internal homeostasis (Korb 2011; Wijas et al. 2022). As such, significant additional CH₄ emissions from mound building termites could result from the combination of warmer ambient temperatures and range expansion. Modifying existing ecosystem models to include parameters for termites, including termite CH₄ generation, range expansion, and microbial oxidation occurring within the mound would be a step forward in accounting for such shifts (Law, Allison, et al. 2024). For one northern Australian termite species (Microcerotermes sp.), Law, Allison, et al. (2024) showed that CH₄ production decreased above 35°C, suggesting that there may be an upper limit CH₄ emission, yet our models showed that CH₄ emission continued to increase above this temperature. It is therefore important to quantify how termites and termite‐associated microbial functions respond more broadly to temperature increases and if a thermal maximum limits CH₄ emissions at the mound level. Given that we measured three species of termites, we provide a small insight as to how termite CH₄ emissions can respond to changing temperatures. Notably these are the dominant species at our site and are therefore the main contributors to carbon dynamics in our study ecosystem. Moving forward, relationships between termite mound CH₄ emission and temperature should be tested in other ecosystems and across a range of species, which would be useful for informing parameterization of the global CH₄ budget.

4.3. The Influence of Mound Methanotroph Communities on Functional Genes, but Not CH₄ Emission

In our analysis of metagenomic data from termite mound material we found that specific combinations of the bacterial methanotroph community (represented by PCoA1 and PCoA2 scores) jointly determined the response of CH₄ emission. Interestingly, methanotrophs were rare members of the mound bacterial community (0.25% of all genera), which is in accordance with results reported by Chiri et al. (2020). Of the 16 methanotrophic bacterial genera that we identified, 10 genera were found in the termite mound material we sampled, suggesting that termite mounds harbor a diverse methanotroph community. While we showed no species‐level differences in methanotrophic members in mound material, future work to characterize methanotrophs found in the mounds of other species in other places would help to determine if the methanotroph communities assembling in termite mounds are broadly consistent.

Surprisingly, we were unable to directly link local methanotroph communities or pmoA gene abundance to lower CH₄ emissions. In contrast, Nauer, Hutley, and Arndt (2018) showed that 50% of termite‐produced CH₄ was oxidized on average prior to leaving the mound for three North Australian termite species (Microcerotermes nervosus, Macrognathotermes sunteri, and Tumulitermes pastinator), suggesting that mound methanotrophs play an important role in net CH₄ release. It appears that the missing link between methanotrophy potential, as indicated by relative abundances, and the result of CH₄ oxidation in termite mounds is an understanding of pmoA expression. Metatranscriptomics could be used to better identify the direct function of bacterial methanotrophy and its subsequent impacts on CH₄ emission. Future studies should pursue a more integrated approach with field and laboratory based CH₄ measurements and microbial ecology, using both metagenomics and metatranscriptomics to determine environmental and community‐based methanotrophic controls on net CH₄ emission from termite mounds. For example, the controlled environment of mound mesocosm experiments could be used to study how mound methanotrophs are activated over time, by season or with temperature, which could subsequently be related to CH₄ emissions from mounds and individual termites (Law, Allison, et al. 2024). Furthermore, there is significant opportunity to holistically describe the termite mound microbiome, yet such sampling may be difficult to pair with emission measurements in the field, as it is highly destructive (Chiri et al. 2020). Lastly, mapping methanotroph‐associated gene expression across a range of external environmental conditions in the field, particularly temperature, would provide insight as to the temporal and climate‐mediated shifts in methanotrophy within the mound.

4.4. What Is the Most Effective Way to Upscale CH₄ Emissions From Termite Mounds?

To complement the TEF and biomass approach, mound‐level CH₄ emission factors may have greater utility when applied in mound‐dominated landscapes to quantify CH₄ emissions. Emissions at the mound‐level have been recorded for a growing number of species across tropical ecosystems (Table S10), and while not as robustly characterized as TEFs (Zhou et al. 2022), collecting CH₄ emission data across the diversity of mound‐building termites is a key research objective for improving global estimates (Law, Allison, et al. 2024). It will be important for future studies to more accurately map termite biomass by nesting strategy at a global scale, so that the mound‐level emission factor approach can be applied where it is best suited. More generally, we challenge researchers to consider how the environment surrounding termites may alter the amount of CH₄ that reaches the atmosphere. Additionally, given the variability in termite mound sizes, it is useful to pair mound‐level emission factors with average mound sizes. Photogrammetry is an accessible tool for this, as it is a portable method utilizing a simple smartphone app and a small amount of post‐processing that can be done in open‐source programs (i.e., Polycam, MeshLab, Blender).

Even with a broad database of mound‐level emission factors, there remains the challenge of determining landscape‐level mound abundance (Law, Allison, et al. 2024). Termite mounds range in spatial distribution, demonstrating both aggregated and more evenly distributed patterns (Davies et al. 2014; Levick et al. 2010), making field surveys challenging. Technologies such as airborne Lidar allow for epigeal mounds to be detected and mapped over large areas with greater ease (Levick et al. 2010; D'hont et al. 2021) and could be used to quantify mound dimensions (i.e., height and volume, D'hont et al. 2021). However, determining if remotely‐sensed mounds are active and occupied by termites remains an important limitation, but can be resolved with field data ground truthing the proportion of mounds active in an ecosystem (Clement et al. 2021). Paired with broader characterization of mound‐level CH₄ emission across the termite diversity of a given region, remotely sensed termite mound abundance data will substantially reduce uncertainty in CH₄ accounting efforts.

5. Conclusion

Our findings highlight both the complexity and the opportunity in refining global termite CH₄ emissions by incorporating the context and dynamics of the internal and external termite mound environment. Termite species, temperature, and season emerge as important factors that should be used to improve estimates of global termite CH₄ emissions. Future research directly linking methanotrophy‐associated gene expression in microbes within mounds to CH₄ oxidation could help define a microbial oxidation factor applicable to mound CH₄ emissions on a larger scale. Additionally, measuring mound‐level emission factors for the global diversity of mound‐building species can incorporate microbial oxidation and relieve errors inherent in calculations relying only on TEFs and termite biomass. Our understanding of the factors influencing termite mound methane in Australian savannas can be used to refine termite contributions to the global CH₄ budget.

Author Contributions

Abbey R. Yatsko: conceptualization, methodology, investigation, writing – original draft, formal analysis. Caleb Jones: investigation, writing – review and editing. Paul Eggleton: formal analysis, writing – review and editing. Baptiste Wijas: investigation, formal analysis, writing – review and editing. Amy E. Zanne: conceptualization, writing – review and editing, supervision. Jeff R. Powell: formal analysis, writing – review and editing. Marcos Pérez‐Losada: formal analysis. Ignacio Ramos‐Tapia: formal analysis.

Funding

This work was supported by the University of Miami Biology Department and a National Science Foundation Graduate Research Fellowship (Award #1938060) to ARY. Funding was also acquired from the US National Science Foundation, Ecosystem Studies Cluster, under awards DEB‐1655759 and DEB‐2149151 to AEZ as well as a UK NERC grant NE/K01613X/1 to PE.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Diagram of chamber‐based sampling for larger mounds (five measurement points, for C. acinaciformis and N. magnus ). (a) A N. magnus mound in the field with sampling chambers fixed in place. (b) The semi‐closed sampling chamber system uses an external ring of potting clay to maintain airtight conditions. Tubing coming out from the chamber connects to the LGR UGGA. (c) Schematic for sampling larger mounds at five locations on each mound: north‐facing (N), east‐facing (E), south‐facing (S), west‐facing (W), and top (T). Note that in the case of A. laurensis mounds, the top (T) measurement was omitted.

Figure S2: Relative abundance of the ten methanotroph genera identified in the mound material metagenomic dataset for evaluating variation between mounds.

Figure S3: Example of how data was recorded during the measurement of individual termite CH₄ emissions (expressed as ppm CH₄). Red indicates where the measurement chamber is inactive and open, therefore only recording concentration of CH₄ in ambient air. Yellow indicates the calibration period, where the chamber is closed but measurements are not yet recorded. Green indicates the measurement period, where the chamber is closed and measurements are being made of changes to CH₄ concentration. This measurement came from 2.17 g of N. magnus individuals, which had a TEF of 0.33 μg CH4 h⁻¹ g termite⁻¹, which was calculated from the slope of the line in the measurement period.

Table S1: Environmental conditions and mound CH₄ emission sample sizes for the four resampling campaigns based on season. Monthly average temperatures for May 2022, November 2022, and August 2023 were derived from a weather station at the field site. The average temperature in February 2024 was sourced from the NASA POWER dataset using the R package “nasapower” (Sparks 2018) due to the fact that in December 2023 Cyclone Jasper destroyed the field site weather station. Precipitation data for each season is calculated as the 1 month average precipitation prior to field sampling and was sourced from the Australian Bureau of Meteorology (BOM 2024).

Table S2: Description of mounds re‐measured across the four sampling campaigns. Bolded mound IDs indicate mounds that were successfully measured in each of the four campaigns. If a mound was destroyed or died, it is indicated when this occurred and what mound served as the replacement.

Table S3: Summary statistics for post hoc pairwise comparisons to determine significant species‐level differences in TEF values. Bolded p‐values indicate significant pairwise contrasts.

Table S4: Summary statistics for post hoc pairwise comparisons to determine significant differences between mound‐level CH₄ emissions for the three study species. Bolded p‐values indicate significant pairwise contrasts.

Table S5: Results summarized across all levels of CH₄ emission: termite individual (TEF), average mound‐level (per mound), and landscape scale (per hectare). Data for the mounds occupied per ha come from Clement et al. (2021). Average TEF units: μg CH₄ g⁻¹ termite h⁻¹. Average mound emission units: Tg CH₄ mound⁻¹ year⁻¹. Landscape‐scale mound emission units: Tg CH₄ ha⁻¹ year⁻¹. Asterisks (*) in the average TEF, average mound‐level emission, and landscape‐scale mound emission columns indicate the top‐emitting species at each level of inference (individual, mound, landscape).

Table S6: Summary of structural equation modeling (SEM) output.

Table S7: Model summary of linear mixed effects model testing all relevant predictors of CH₄ emission to understand variation across mounds. For termite species‐level comparisons, A. laurensis is the reference species.

Table S8: Model results for testing individual methanotroph genera relative abundance as a predictor of CH₄ emission. Bolded values indicate significance.

Table S9: Model summary of linear mixed effects model testing all relevant predictors of CH₄ emission to understand variation within mounds. For termite species‐level comparisons, A. laurensis is the reference species.

Table S10: Summary of published CH₄ emissions for termite mounds at the species level. Note that the sample sizes from this study include mounds that were a part of the initial, larger sampling campaign, hence the greater sample size for each species.

Table S11: List of bacterial genera associated with methanotrophy. These genera were used to query the metagenomics dataset for methanotroph candidates to include in our analysis on variation in bacterial methanotroph community composition within and between mounds.

Table S12: Coefficient of Variation (CV) in CH₄ emissions for individual mounds (calculated from 4 to 5 emission subsamples) which were remeasured in all four seasons.

Table S13: Mean mound size (as surface area, m² and as volume, m³) and mound wall thickness (mm) for the three study species.

Data Availability Statement

Data and code have been uploaded to Zenodo at https://doi.org/10.5281/zenodo.19141664 (Yatsko 2026). The metagenomic data supporting this study are available under Bioproject PRJNA1234298 at https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1234298.