Global termite methane emissions have been affected by climate and land-use changes

Termites with symbiotic methanogens are a known source of atmospheric methane (CH4), but large uncertainties remain regarding the flux magnitude. This study estimated global termite CH4 emissions using a framework similar to previous studies but with contemporary datasets and a biogeochemical model. The global termite emission in 2020 was estimated as 14.8 ± 6.7 Tg CH4 year−1, mainly from tropical and subtropical ecosystems, indicating a major natural source from upland regions. Uncertainties associated with estimation methods were assessed. The emission during the historical period 1901–2021 was estimated to have increased gradually (+ 0.7 Tg CH4 year−1) as a result of combined influences of elevated CO2 (via vegetation productivity), climatic warming, and land-use change. Future projections using climate and land-use scenarios (shared socioeconomic pathways [ssp] 126 and 585) also showed increasing trends (+ 0.5 to 5.9 Tg CH4 year−1 by 2100). These results suggest the importance of termite emissions in the global CH4 budget and, thus, in climatic prediction and mitigation.

www.nature.com/scientificreports/CH 4 emissions, but the high heterogeneity of the emissions makes it difficult to conduct spatially representative measurements.Therefore, there remain large uncertainties in present estimations of global termite CH 4 emissions (Supplementary Table S1).Moreover, little is known about temporal changes in termite emissions, which may be affected by global and local environmental changes.Revisiting global termite CH 4 emissions and assessing their temporal variation would improve our understanding of the global CH 4 cycle and, eventually, help in making climatic projections that include biogeochemical feedback.
The objectives of this study were thus to (1) revisit the estimation of global termite CH 4 emissions by using modern data and methods, and (2) investigate temporal changes and their driving factors by conducting simulations for historical and future periods.Additionally, this study explored the range of the estimation uncertainty by analyzing estimates derived by using different emission factors and methods.

Global total emissions
Termite CH 4 emissions were estimated globally using a framework similar to previous studies (i.e., emission = termite density × emission factor; see "Methods and data") and contemporary datasets.This study took account of influential factors such as climate, land-use, and vegetation photosynthetic productivity (Fig. 1a) during the historical (1901-2020) and future (2021-2100) periods.Climate and land-use were derived from existing datasets (Supplementary Fig. S1), and vegetation productivity was obtained from a simulation of terrestrial carbon cycle with a process-based biogeochemical model (VISIT).
The land area of potential termite habitat was estimated using an empirical temperature threshold (see "Methods and data") to be 92.9 × 10 6 km 2 in 2020, mostly in Africa, Australia, and South America, as well as large parts of Asia, Europe, and North America.Vast regions without termite habitat were seen in northern Eurasia and North America.The actual habitat area, including a restriction to take account of land use for agriculture, was estimated as 79.8 × 10 6 km 2 .By estimating empirically using model-simulated GPP in tropical ecosystems and assuming termite densities per habitat area in non-tropical ecosystems (see "Methods and data"), total termite biomass was estimated.This estimate, 122.3 Tg dry weight, is comparable to that of Rosenberg et al. 18 , who used thousands of measurement data to derive a total of 300 Tg (uncertainty range, 100-500 Tg) for underground (soil) and aboveground arthropods, of which 40%, or 120 (40-200) Tg, are Isoptera (termites).The results of Rosenberg et al. and this study imply that total termite biomass is larger than that of Formicidae (ants) and comparable to that of humans (100 Tg dry weight) 19 .www.nature.com/scientificreports/By using emission factors derived from a recently published dataset by Zhou et al. 20 and termite biomass density mentioned above, the global total termite CH 4 emission was estimated as 14.8 ± 6.7 Tg CH 4 year −1 (mean ± standard deviation of random-sampling ensembles; see Statistical analysis of uncertainty in "Methods and data" section).This value is intermediate among those reported by previous studies (Table S1) and close to a relatively new independent estimate by van Asperen et al. 21of 14. 96 Tg CH 4 year −1 , based on observed tropical emissions data from Amazonia.It is slightly higher than the estimate adopted in the synthesis of the global CH 4 budget of the Global Carbon Project: 9 [3-15] Tg CH 4 year −1 (Ref. 5).These differences among studies are discussed later.At this point, it is sufficient to note that we need to be careful about how emissions and related parameters are defined.As reported by Nauer et al. 22 , about half (20-80%) of the CH 4 produced by termites may be oxidized within the mounds without reaching the atmosphere; therefore, the use of emission factors obtained from isolated termites (e.g., in a cuvette) is likely to cause overestimation of emissions to the atmosphere.Note that the present study did not select data by observation method.Nonetheless, clearly, the role of termite emissions in the CH 4 budget is remarkable, especially those from upland regions.Because of methanotrophic oxidation, aerobic soils are a sink of atmospheric CH 4 and estimated with a process-based model (VISIT, see "Methods and data") to be 31.6Tg CH 4 year −1 from the termite-inhabited area in 2020 (Fig. 1b).Termite emissions likely offset about 47% of the absorption flux at landscape or larger scales.Note that the amount of offset varies among locations and through time, as described later.

Spatial distributions
Spatial distributions of termite density estimated in this study (Supplementary Fig. S2) appear to be qualitatively comparable to those of termite diversity 16,17 .Termite CH 4 emission intensity varied spatially from < 0.05 g CH 4 m −2 year −1 in deserts to > 0.2 g CH 4 m −2 year −1 in tropical forests in Africa, Southeast Asia, and South America (Fig. 1b), where high termite density was assumed (Fig. S2).The global pattern estimated in this study is roughly comparable to that obtained by Fung et al. 14 (Supplementary Fig. S4).Regionally, Africa and South America accounted for about 55% of the total emission (4.5 and 3.5 Tg CH 4 year −1 , respectively) (Supplementary Fig. S3).Observations by field studies have shown similarly high termite CH 4 emissions from tropical biomes.For example, Martius et al. 23 conducted observations at wood-feeding termite nests in Amazonia and obtained comparable fluxes (~ 0.18 g CH 4 m −2 year −1 ) 24 .Brümmer et al. 25 reported termite emission in the savanna of West Africa to be about 0.25 g CH 4 m −2 year −1 ; in their study, croplands were a net CH 4 sink because soil uptake was larger than termite emission there.
Upland ecosystems such as grasslands and deserts can absorb atmospheric CH 4 due to soil oxidation by methanotrophs.This study indicated that termite emissions are a substantial source and influence the upland CH 4 budget, although other sources such as biomass burning are in a comparable magnitude and sometimes influential 26 .As a result of the heterogeneous distribution of these fluxes, it was shown that upland ecosystems can be both a net sink and a net source, depending on relative intensities of soil oxidation and termite emission (Fig. 1c).Such spatial heterogeneity is, although its area-based intensity is weaker than wetlands, important when interpreting and evaluating the upland CH 4 budget especially using atmospheric observation data.

Historical variability
During the historical period, global termite emissions were estimated to have gradually increased from 13.1 ± 0.1 Tg CH 4 year −1 in the 1900s (1901-1910) to 14.8 ± 0.2 Tg CH 4 year −1 in the 2010s (2011-2020) (mean ± standard deviation of interannual variability).The increase was associated with land-use, climate, and atmospheric changes, and could be attributable to changes in the habitat area and termite biomass.Potential (temperaturelimited, black dotted line in Fig. 2a) termite habitat expanded in temperate to boreal regions as a result of climatic warming (~ 2.9 × 10 6 km 2 ).In contrast, actual habitat (also impacted by land use, red line in Fig. 2a) was estimated to decrease by 2.8 × 10 6 km 2 , mainly because of conversion from natural vegetation to croplands.Termite biomass was estimated to decrease when considering only land use and habitat loss (black line in Fig. 2b).When including the effects of atmospheric CO 2 increase and resultant fertilization on vegetation productivity (on average by 29%, estimated by the VISIT model), termite biomass was estimated to increase by 31 Tg dry weight (yellow line in Fig. 2b).Eventually, these factors explain the increase of termite CH 4 emissions as shown by a difference between black and yellow lines of Fig. 2c.
The overwhelming increase in anthropogenic CH 4 emissions during the historical period (> 100 Tg CH 4 year −1 ) 27 can make it difficult to detect the impact of the change of termite emissions on the atmospheric CH 4 concentration.Nevertheless, the increased termite emissions may substantially influence the CH 4 budget of upland areas, which cover a vast land area.Note, however, that soil CH 4 oxidation was estimated to increase even more rapidly than termite emissions because of the elevated atmospheric CH 4 concentration (nearly doubling from the 1910s to 2010s, estimated by the VISIT model).

Projected emissions
The projections of termite CH 4 emissions indicated that they will increase, with the pattern of increase dependent on the future scenario (Fig. 3).By the 2090s, under a mitigation-oriented scenario (ssp126), termite emissions were estimated to increase by 0.5 Tg CH 4 year −1 (0.2-0.7 Tg CH 4 year −1 , depending on the climate scenarios), whereas under an adaptation-oriented scenario (ssp585), the estimated increase was 5.9 (4.8-7.0)Tg CH 4 year −1 .In the ssp226-based estimation, termite emissions showed a broad peak around the 2050s and then decreased gradually.This overshoot pattern is apparently comparable to the pattern of the atmospheric CO 2 concentration in the scenario, which leads to a corresponding variation in vegetation productivity 3 .In contrast, the estimated termite CH 4 emissions under the ssp585 scenario showed steady increases, again in parallel with the atmospheric CO 2 level and associated climatic change.The differences among the climate projections by the five climate models were small in both scenarios compared with the difference between the ssp126 and ssp585 scenarios.
In the ssp126-based estimation, termite CH 4 emissions increased mainly in northern temperate to boreal regions, where termite habitat is currently limited by cold temperatures (Fig. 3b), and emissions in tropical and subtropical areas were relatively unaffected.In the ssp585-based estimation, termite CH 4 emissions were  www.nature.com/scientificreports/estimated to increase not only in the temperate to boreal but also in tropical to subtropical regions.The increases in temperate to boreal regions of Northern Europe, Eurasia, and North America were associated with the northward expansion of termite habitat, whereas the increases in the tropical to subtropical regions of Africa, Southeast Asia, and South America were associated with increases in vegetation productivity.Globally, these increases completely offset the decreased emissions in subtropical areas caused by land-use conversion from natural vegetation to croplands (e.g., in savanna regions of Africa and South America).
The projected climate change will affect termite distribution and activities, leading to various (i.e., both positive and negative) indirect impacts on ecological processes such as carbon and nutrient cycling.To date, few studies have attempted to predict termite activities including CH 4 emissions, although future emissions from other natural sources and their climatic feedback have been explored 28 .Zanne et al. 29 estimated future changes in termite-induced woody decay by using climate scenarios similar to those used in the present study.Their conclusion that termite functions in the terrestrial carbon cycle will be enhanced in the future is consistent with the findings of the present study, although they put more focus on the sensitivity of emissions to the temperature change (> 6.8 times per 10 °C warming).If termite feeding activities are as sensitive to temperature as implied by Zanne et al. 29 , then the future increase of termite CH 4 emissions estimated in this study, 40% in the ssp585based estimation for the 2090s, could be much larger.By contrast, Buczkowski and Bertelsmeier 30 , using a species distribution model, suggested that habitat expansion of invasive termites may occur, for example, in Europe.Further studies should examine future termite CH 4 emissions both by considering physiological mechanisms and by conducting continuous, extensive field observations.

Estimation uncertainty
Previous estimates of the global termite CH 4 emissions range from 1 to 152 Tg CH 4 year −1 (summarized in Supplementary Table S1).This wide disparity, especially in the early studies, is apparently associated with biases stemming from the use of a limited number of observations that could not adequately represent the vast area of termite habitats.In addition, the mechanistic understanding of the factors that determine the spatial and temporal variation of the emissions was insufficient.
To address the estimation uncertainty, the results of several supplementary estimations were compared (see emission maps in Supplementary Fig. S5).The estimate described in the previous sections, using land use-and vegetation productivity-based termite density and emission factors from Zhou et al. 20 (Fig. 1b and Fig. S5d), was referred to hereafter as the control.First, when emission factors specific to each land-cover type, derived from Sanderson were used instead of random sampling from the dataset, the total global emission in 2020 was estimated as 17.3 ± 2.6 Tg CH 4 year −1 , that is, 18% higher than the control estimate of this study.This higher estimate is attributable to the high emission factor (5.9 µg CH 4 g −1 termite h −1 ) obtained from Sanderson and applied to all tropical forests (see Fig. S5 for the spatial pattern).Second, when termite density was determined by land use only (i.e., no effect of vegetation productivity), the global total emission in 2020 was estimated as 15.6 ± 7.1 Tg CH 4 year −1 , that is, 6% higher than the control estimate of this study.In this case, the uniform termite densities (8 g dry weight m −2 in tropical deciduous forests and 11 g dry weight m −2 in tropical evergreen forests, after Sanderson, 1996 31 ) applied to tropical ecosystems, resulted in a higher global total value.Third, combining the first and second cases, the global termite emission was estimated as 19.2 ± 3.0 Tg CH 4 year −1 , that is, 30% higher than the control estimate of this study.However, this third estimate is close to those of Fung et al. 14 and Sanderson 31 , who used land use-specific termite density and emission factors.Thus, the selection of the emission factor dataset and of the termite-density mapping method can explain a large part of the disparity among the previous studies, excepting the extremely low or high ones certainly attributable to the use of biased data.Remarkably, the estimation procedures also affected the temporal trend of termite emissions.When only climate and land-use effects were included (i.e., the effect of vegetation productivity was ignored), global termite biomass and CH 4 emissions were estimated to gradually decrease through time because of deforestation in tropical areas and resultant habitat loss (Fig. 2).In the future, the estimation uncertainty is expected to be reduced through the accumulation of additional field and laboratory measurement data, improved upscaling that takes account of the spatial representativeness of data and the determining mechanisms, and verification using independent evidence.

Global CH 4 budget and termite emissions
The estimated global total emission in 2020, 14.8 ± 6.7 Tg CH 4 year −1 , confirms that termite emissions constitute a substantial component of the global CH 4 budget: about 2% of the global total (natural + anthropogenic) emissions and 4% of natural emissions 5,32 .The total termite emission is larger than anthropogenic emissions from most countries (except China, India, United States, Brazil, Russia, and Indonesia) and comparable to emissions from paddy fields in East Asia 33 .
Termite emission is one of the major emission sources in upland areas (other sources: wildfires, wild animals, and geological processes).Indeed, termite emissions can turn many uplands into net CH 4 sources, even after uptake by soil methanotrophic oxidation is subtracted (Fig. 1c).However, the emissions were generally weak in their intensity (on the order of 0.1 g CH 4 m −2 year −1 ; Fig. 1b) and distributed over a vast area of uplands; as a result, it is difficult to detect and quantify the signal using atmospheric observations made from, for example, tall towers and satellites.Also, because the emissions are produced by common microbe taxa, the stable carbon isotope ratio of termite-emitted CH 4 (δ 13 C-CH 4 , − 63.4 ± 6.4‰) is not distinguishable from that of CH 4 emitted from wetlands and enteric fermentation 34 .These considerations suggest that a bottom-up approach is needed, but they also indicate the importance of taking termite emissions into account when evaluating national and regional CH 4 budgets (e.g., Ito et al. 33 ).
The results of this study imply that, in the future, termite emissions will increase globally (by 0.5-5.9Tg CH 4 year −1 by the end of this century), as a result of rising atmospheric CO 2 and climate change.The poleward expansion of potential termite habitat is projected to result in additional CH 4 emissions from temperate to boreal regions, even under the mitigation-oriented ssp126 scenario (Fig. 3).Although the projected magnitude of the change in termite emissions is smaller than the projected magnitude of the change in wetland emissions (+ 20 to 150 Tg CH 4 year −1 ) 28 , the increase of termite emissions may have significance for regional and global CH 4 budgets and climatic change.Based on the 20-or 100-year horizon Global Warming Potential values (79.7 and 27.0, respectively; after IPCC, 2021 3 ), the increase of termite emissions corresponds to CO 2 emissions of 4-129 Tg C year −1 .The increase in termite emissions can, thus, substantially influence efforts to mitigate climatic change through emission reduction, especially under the Global Methane Pledge, which calls for country-level CH 4 emissions to be reduced by 30% by 2030.
Clearly, when considering ecosystem management and climatic mitigation, we should note that the impacts of land use and climatic changes are complicated and interconnected.Land-use conversion for food and bioenergy production, especially in tropical regions, should suppress the emission increase to some extent.This may not be a main factor driving land-use decisions, because land-use conversion has stronger impacts on the CO 2 budget and, possibly, biodiversity.Although not explored in this study, extreme climate events associated with climatic warming may affect termite activities and perhaps ecosystem integrity.The results presented in this study have implications for ecosystem management that considers the overlooked effects of decomposers and the non-CO 2 greenhouse gas budget.

Limitations and future perspectives
This study revisited global termite CH 4 emissions, and it provides the first estimation of the temporal changes, but it has several limitations.First, up-to-date datasets were used, the spatial distributions of termite density and emission factors were not spatially resolved with high reliability.A new dataset compiled by Zhou et al. 20 was used to capture the frequency distribution of termite emission factors, but the dataset does not differentiate among regional and phylogenetic groups 17,35,36 .Similarly, the present study did not treat soil-feeding (humivorous) and wood-feeding termites separately, although the former is reported to release a larger amount of CH 4 23 .Further accumulation of observational data and analyses is required to fully characterize the spatial patterns of termites' functional attributes.Second, the time lag in termite migration was not considered; instead, was implicitly assumed that termites are sufficiently mobile (through dispersion by flight or marching) 11 to keep up with the habitat expansion caused by climatic warming.Several genetic and conservation studies have reported historical biogeographic aspects of termites 37,38 , but no direct observations of temporal changes in termite density in primary and secondary ecosystems is available (Fig. 2b).It is still uncertain whether termites can adapt to future climate change, which is predicted to proceed at unprecedented rates; therefore, the results of the present study (Fig. 3) likely show only the potential response.Third, actual ecological interactions are likely to be much more complicated than those included in this study.Termites can affect ecosystem structure and functions by altering carbon and nutrient cycles, while at the same time being themselves influenced by changes in vegetation and natural enemies.For example, Ashton et al. 39 reported that termite abundance increases during droughts and that termites in tropical forests show higher drought resistance because of accelerated decomposition and altered soil properties.In addition, da Cunha et al. 40 showed host plant differences influence the geographic distribution of wood-feeding termites.Furthermore, several ant species are termite predators, and their abundance thus affects termite density 41 .These ecological interactions might affect the termite habitat, diversity, and functions under changing environments, and detailed studies are needed to elucidate these mechanisms.To date, no terrestrial carbon cycle model or dynamic vegetation model, especially among those embedded in Earth system models, explicitly includes termite-driven processes 42 .Considering the extensive distribution, biomass, and dynamic flows of termites, it would be meaningful for these models to include termite-related factors.Their inclusion would surely result in improved reliability and ability to capture biogeochemical feedbacks.

Methods and data
In this study, global termite CH 4 emissions were estimated using empirical approaches adopted in previous studies and updated data of climate, land-use and land-cover, and termite distribution and emission factors.Also, the use of a process-based biogeochemical model (VISIT: Vegetation Integrated SImulator for Trace gases 33 ) that simulates vegetation productivity and soil carbon cycle, allowed including environmental responses in a mechanistic manner.

Geographic distribution of termites
Termite habitat area was assumed to be limited by temperature, while rainfall may affect termite diversity within tropical habitats 43,44 .To find a termite threshold of potential termite habitat, two global datasets of field-observed termite colonies were examined.
For each dataset, records that included latitude and longitude values were used.Additionally, several studies in the literature were included (yellow stars in Fig. 4): Pullan 45 for Africa, Palin et al. 46 for the Amazon-Andes area, Jamali et al. 47 for Eucalyptus forests in Australia, and Sheffrahn et al. 48for global highland observations.Based on the observed termite distribution, a temperature threshold below which termites cannot survive over winter was examined.A global gridded dataset of historical climate conditions produced by the Climate Research Unit (CRU) TS4.05 49 was used to derive mean monthly temperatures for 2001-2020.The climatic envelope of termite habitat, which encompasses most of the observed termite colonies and captures well their distribution boundaries, was examined.On the basis of many trials, a minimum monthly temperature higher than − 8 °C (Fig. 4) was selected as the temperature threshold explaining the observed termite distribution.Using this threshold, potential termite habitat was estimated annually during the study period.

Termite CH 4 emission
Termite CH 4 emission at an arbitrary point (µg CH 4 m −2 h −1 ) is calculated as follows: Biomass density of termites (g termite m −2 ) within the climatic envelope, as described in the previous section, was first estimated on the basis of land-cover type, as in previous studies 14 .This study referred to Sanderson 31 for the mean termite density of each land-cover type; this value ranged from 0 g m −2 in tundra and polar desert to 11 g m −2 in tropical evergreen forest (Table 1).For cropland, the termite density and emission factor indicated in Table 1 were used in all cases.The global distribution of natural vegetation types was derived from Ramankutty and Foley 50

Figure 1 .
Figure 1.Upland CH 4 budget including termite emissions.(a) Schematic diagram of the upland CH 4 budget, including termite emissions.Numbers in square brackets indicate the global total CH 4 flux estimated by this study for the year 2020 (Tg CH 4 yrar -1 ).Distributions of (b) termite CH 4 emissions and (c) the net CH 4 flux, including soil oxidation uptake estimated by a process-based model (maps generated by Panoply 5.2.9, https:// www.giss.nasa.gov/ tools/ panop ly/).

Figure 2 .
Figure 2. Temporal change in the estimated global termite CH 4 emissions.(a) Potential and actual termite habitat areas, (b) total termite biomass in dry weight, and (c) total termite CH 4 emissions.The shading shows standard deviation ranges obtained from 1000 ensemble calculations using randomly sampled emission factors.

Figure 3 .
Figure 3. Projected global termite CH 4 emissions.(a) Interannual variability under the ssp126 and ssp585 scenarios using five climate projections.Thin lines show individual climate model results, and thick lines show their mean.Distribution of the estimated changes for (b) ssp126 and (c) ssp585 from the 2010s to the 2090s (maps generated by Panoply 5.2.9, https:// www.giss.nasa.gov/ tools/ panop ly/).
, and the historical change in cropland was derived from the Land Use Harmonization version (1) Emission = Termite biomass density × Emission factor.

Figure 4 .
Figure 4. Distribution of observed termite colonies and potential habitat.Small red dots, University of Florida Termite Collection dataset; large blue dots, iNaturalist dataset; yellow stars, literature data.Areas with a minimum monthly mean temperature higher than − 8 °C (climatological mean) are colored by cropland fraction (Supplementary Fig. S1b) (Map generated by QGIS 3.28.7,https:// qgis.org/ en/ site/).

Table 1 .
31nd-cover/use type-specific emission factors and termite biomass density after Sanderson31.