Our results revealed statistically significant relationships between CTs and BA worldwide (P < 0.05; Figs. 2, 3 and S.1–8), while also identifying regional patterns summarized into a set of six CT domains (CTD; Fig. 3). Most of the CTDs (2, 3, 4, 5 and 6) showed characteristic large-scale CT-fire patterns in specific regions, while CTD 1 had a scattered distribution across global biomes. CTD2 expanded over large areas in southern Australia, South America (Colombia, Equator and Perú), eastern Africa, Ukraine and southeastern Russia; CTD3 in tropical Africa and America; CTD4 in the African savanna; CTD5 in central and South America and eastern Russia; and CTD6 in southern Africa, northern Australia, Cerrado (Brazil), India and the Asian Boreal taiga. During the peak of the fire season (accounting for 84.3 % of the total annual BA; Fig. 1B, C; see supplementary materials for more details), CTs were responsible for modulating 52.9% of the interannual variability in global BA (P < 0.05; Table 1; Fig. S.9). TNA is the leading CT mode related to BA globally, especially in the northern hemisphere, being associated with 25.7% of the total global BA. Changes in convective activity induced by trade winds and latent heat flux anomalies are thought to modulate the SST variability in the region12. TNA dominates variability in terrestrial evaporation and drought in relatively large regions of both hemispheres due to different physical mechanisms explained in the SM7,13. In the Southern hemisphere, SAM is the leading CT mode of climate variability associated with changes in the position and strength of the polar jet around Antarctica14. This variability drives large fluctuations in weather and climate up from the troposphere to the lower stratosphere15, which can result in significant linkages with BA, as shown in our analysis. The SAM was associated with 12.3% of the total global BA (Table 1), due to its significant influence across the African savanna, Australia and the South American Savanna (Cerrado; Fig. 2C and S1.8).
The maps at 0.5°x 0.5° pixel resolution represent the correlations at P < 0.05 (|R| > = 0.34) and P < 0.10 (|R| > = 0.29) between BA and CTs. The six most important CTs in terms of modulated global BA are presented, namely Atlantic Multidecadal Oscillation (AMO (lag 0)), Tropical North Atlantic (TNA (lag 0)), Pacific North American (PNA (lag 3)), Western Pacific (WP (lag 6)), Pacific Decadal Oscillation (PDO (lag 9)) and Southern Annular Mode (SAM (lag 9)). See the supplementary materials for other CT modes and lags.
These domains were derived by applying a hierarchical clustering approach to the correlation coefficients obtained for the different combinations of CTs and lag windows at pixel level. A Spatial distribution of the six CTD. B Characterization of the CTD based on the relationship between each CT and burned area at a given time lag. Colored bars indicate the corresponding CT; color brightness relates (from darker to lighter) to the median, 75th and 95th percentile of Pearson’s R. Correlative bars with the same color are ordered from lag 0 to 9. Dashed lines mark the significance threshold (P < 0.05 (|R| > 0.34) and P < 0.10 (|R| > 0.29)). Atlantic Multidecadal Oscillation (AMO); Arctic Oscillation (AO); East Atlantic (EA); El Niño 3.4 SST index (ENSO); Eastern Pacific (EP); Indian Ocean Dipole (IOD); North Atlantic Oscillation (NAO); Pacific Decadal Oscillation (PDO); Pacific North American (PNA); Southern Annular Mode (SAM); Tropical North Atlantic (TNA); Tropical South Atlantic (TSA); Western Pacific (WP).
Most CTs considered in this analysis featured significant synchronous and lagged signals between CTs and BA. The 0- and 6-month lags were the most important for explaining global BA (Table 1; Figs. 2 and S1.8), as per their influence on weather conditions during the fire season peak (lag 0), and the effect on biomass accumulation and long-term drought leading to dry fuels and flammable landscapes (lag 6). TNA, EA and EP had synchronous and significant correlations with BA, whereas PNA, WP, SAM and NAO were found to predominantly relate to BA through lagged effects. Our study confirms and further expands our knowledge on the prominent role of CTs in modulating global BA through important remote climate feedbacks operating at different spatio-temporal scales, either concurrent or asynchronous, and entailing regional to global impacts. These connections directly influence biomass accumulation and moisture content associated with changing weather conditions.
The spatial extent of influence of each CT varied from sub-continental (WP, EP, EA, AO, IOD) to global scale (AMO, TNA, TSA, NAO, PNA, PDO, ENSO, SAM). Altogether, more than half of the global interannual BA (52.9%) can be explained by CTs. This is slightly higher than previously thought (48.0%) as prior studies did not include the essential effect of SAM8, affecting the major BA hotspots in the southern hemisphere due to changes in the strength and position of the polar jet around Antarctica14. Our analysis also shows that global BA is largely explained by TNA and AMO, rather than by ENSO, as previously thought8. However, although our analysis did not find any significant correlation between TNA and ENSO (Fig. S1), some studies suggest that the low-frequency component of ENSO in the Pacific is associated with TNA anomalies through inter-basin connections16.
The six fire-CT domains identified in this paper allow for better comprehension of regional to global fire activity, which is useful to anticipate BA, adverse fire seasons and forecast which areas of the world are likely to burn simultaneously in the future8. As such, the CT associations with BA highlighted in our analysis provide valuable information for land managers and emergency services worldwide, particularly in regions where long-range fire risk forecasts are not readily available. The percentage of BA modulated by CTs is high, given that fire is not just controlled by global climate but also regionally by land management, fire suppression capabilities and practices.
There is increasing evidence that some CTs are non-stationary, strengthening or weakening their signal on multi-decadal timescales; behavior which may further influence climate and BA6. The source of these low-frequency changes is still unclear but may arise due to internal variability of the climate system or anthropogenic climate change17. In fact, the trends of some CT indices is a well-established phenomenon and has already been linked to climate change17. For instance, previous research identified an intensification of TNA variability under climate warming. This enhanced variability may strengthen the ENSO-forced PNA pattern and tropospheric temperature anomalies due to an eastward shift of ENSO-induced equatorial Pacific convection and of enhanced ENSO variability, which may reinforce the associated wind and moist convection anomalies18. Also, a robust positive trend in the SAM is projected for the end of the 21st century under different climate change scenarios due to its response to stratospheric ozone depletion over Antarctica14, which could result in rainfall decreases in the midlatitudes and increases in the high latitudes of the southern hemisphere, leading to potential and dramatic changes in BA at sub-continental scales15. Also, CTs may show asymmetric behavior in magnitude, duration or seasonal timing in relation to their positive and negative phases, resulting in climate responses that may be non-linear or sensitive only to a specific CT phase19. In light of the uncertain global changes in the forthcoming decades, it is wise to recognize that the observed relationships reported in this study may change in strength and nature in the future.
Below, we highlight the main findings by continent and biome to facilitate interpretation considering the underlying weather patterns behind the CT-fire associations at subcontinental scales. Extended insights into global CT-fire relationships and potential mechanisms explaining these patterns are shown in the Supplementary Materials due to space constraints.
Africa
Africa accounts for about 70% of BA11 and 50% of fire carbon emissions worldwide20. Correlations between weather patterns and some CTs have been recently investigated13, but few studies have addressed the linkage between CTs and wildfires in Africa4. Most fires occurred in areas occupied by savannas (CTD 4). However, other biomes such as the tropical moist forest of the Congo Basin (CTD 3), one of the most biodiverse areas on Earth, also deserve special attention given the increasing anthropogenic pressure it is experiencing.
Savanna fires responded mostly to +TNA (lag 0 and 3), +AMO, +SAM (lag 9), +EA (lag 9) and +ENSO in the northern hemisphere (CTD 4); +AMO, +TNA (lag 0, 6 and 9), -SAM (lag 9), +PNA (lag 0), –PNA (lag 3) and -NAO (lag 6 and 9) in the southern hemisphere (CTD 4); both +IOD and +ENSO conditions in East Africa (CTD 2), related to above-average rainfall, were negatively correlated with BA21, specifically in Kenyan and Ethiopian subtropical grass-savanna biome (lag 0, 3 and 6). Our analysis revealed significant lagged CT-fire relationships, in accordance with fire dynamics in this biome, driven by a rapid fuel build-up during the wet season and by a warm season capable of drying fuels out. +TNA (lag 0 and 3) and +AMO were correlated with increased BA likely due to changes in atmospheric circulation driving lower precipitation22. The +SAM (lag 9) was strongly correlated with increased BA in the CTD 4 (northern savanna; lag 9) and decreased BA in CTD 4 (southern savannas) likely due to altered precipitation and vegetation moisture15. Also, in the CTD 4 in the southern hemisphere, +NAO (lag 6 and 9) was associated with reduced BA, which is consistent with the fact that it promotes wetter conditions prior to the fire season peak23. +PNA (lag 3) was associated with decreasing BA as it is potentially linked to above average convective precipitation over central Africa during the monsoon season.
We found that BA in the tropical moist forest biome (CTD 3) was mainly associated with -TNA (lag 0) and PNA (lag 0, 3, and 9), which promoted wetter conditions while BA diminished, and +SAM (lag 9) which promoted drought conditions thereby possibly boosting BA24. In southern Africa (CTD 1, 2 and 6; Fig. 1D), we found significant effects of TNA, AMO, ENSO (lag 3, 6 and 9), PNA (lag 0), NAO (lag 9), AO (lag 9), EA (lag 0 and 9), EP (lag 0), IOD (lag 0 and 6), and SAM (lag 9) on BA. The +SAM (lag 9) promoted BA due to reduced rainfall in the austral summer in Angola (CTD 6). PNA (lag 3) was positively correlated with BA because of increased precipitation promoting biomass before the fire season. NAO and AO influenced precipitation with a marked dipole in CTD 2 and 6. In CTD 6, precipitation during austral winter (lag 6 and 9) increases with +NAO, whereas the opposite occurs in CTD 2, influencing BA accordingly23.
Americas
Wildfires in the Americas affected a diverse range of biomes each displaying unique fire regimes and peak season timing. Most of the fires occurred in the tropical savannas (Colombian Llanos, CTD 3, and Cerrado CTD 6), where teleconnections such as +PDO and +ENSO, linked to anomalously warm and dry conditions25, concurred in correlation with increased BA, while +IOD and +SAM, which are known to foster precipitation, were both associated with reduced BA25. +TNA and +AMO drive increases in BA in CTD 6 probably due to significant increases in terrestrial evaporation during and before the fire season peak7.
Amazon fires (CTD 2) threaten one of the most megadiverse regions in the world. CTs associated with increases in BA included the +AMO, +TNA, +NAO and the +SAM, particularly at 9-month lag25. Drought is mediated by +TNA and +AMO associated with a sustained northward position of the ITCZ26. +PNA is accompanied by positive precipitation anomalies, diminishing fire activity, in the austral summer27. In the southern end of the continent, Patagonia and the Pampas (CTD 2 and 6), within the temperate grassland and shrubland biome, +SAM is known to boost wildfire activity by promoting significantly warmer and drier conditions during the austral summer25,28.
Wildfires within the North American taiga biome are a dominant disturbance shaping landscape diversity. The positive phases of PDO, ENSO, and IOD correlate with anomalously warmer and drier conditions across all of the Pacific Northwest and Canada, and are likely responsible for the observed increase in BA at the 9-month lag29,30. The +SAM is known to drive cooler temperatures across northern latitudes in North America, which is reflected in the decrease in BA at the 9-month lag27. Also, EA, AO, NAO, EP and WP correlate with BA at different time lags due to changes in terrestrial evapotranspiration as found by previous research7. The southeastern United States hosts a number of primarily temperate ecoregions. The positive phases of NAO, EA and ENSO were all positively correlated with BA through producing cooler and wetter conditions as identified by previous research31. CT-fire relationships in western United States were generally weak, mainly related to +AMO, +PDO and +ENSO, which may induce drought conditions and increased BA. BA in Southern Mexico and Central America (CTD 5) is correlated with the positive phases of AMO, TNA, ENSO and PDO, all of them associated with drought conditions32.
Asia
Asia’s major BA hotspots are located in the temperate steppe (central Asia), the tropical and subtropical dry forests of southeastern Asia (India, Thailand) and the temperate broadleaf forests of China. BA in southern Asia (CTD 5 and 6), mainly correlated with -ENSO (lags 0, 3 and 6), -PDO, -IOD (lag 3), AMO, TNA, +SAM (lags 0 and 3), TSA (lags 6 and 9), +EA (lag 6), -EP (lag 3) and +WP (lag 6 and 9), with a fire season extending from January to May varying latitudinally across the continent (Fig. 1). BA across mainland SE Asia correlated with +ENSO (lag 0, 3) +PDO, +IOD (lag 3, 6 and 9), +EP (lag 9), +TNA (lag 0, 3), +AMO and +EA (lag 3). +PDO influences the East Asian Summer Monsoon, modulating drought across the region33, +ENSO amplifies frequency, duration and intensity of heat waves and drought in much of Indochina, therefore likely contributing to increases in BA34. +IOD restricts rainfall in June-August, particularly in Cambodia and central Laos, where we witness the greatest BA increases. Wildfire activity in the Maritime region of SE Asia, often related to peatland fires, was predominantly associated with drought mediated by the positive phases of ENSO (lag 0, 3 and 6), TNA (lag 0), IOD (lag 0), PNA (lag 6), WP (lag 9).
Eastern Asia (China) holds the world’s largest subtropical forests, and falls mostly within the CTDs 1, 2, 3 and 6, with a fire season peak timing varying latitudinally (Fig. 1). We found significant effects of +ENSO on BA (lag 6 and 9), probably amplified by +PDO35. Reduction in BA was found under +AMO and +TNA (lag 0), which have been reported to enhance precipitation over the region36. Also, +EA (lag 0), +EP (lag 0) and +WP (lag 9) correlated with increased BA, all these driving weather conditions of limited precipitation, high temperature and dry fuels during fire seasons37,38.
The Asian Boreal taiga is a vast landscape dominated by coniferous forests. The western taiga across the Ural Mountains (CTD 6) experienced heightened BA during the positive phase of AO (lag 0, 3 and 9), NAO (lag 3 and 9), EA (lag 9), WP (lag 6 and 9) and SAM (lag 6). Reductions in BA correlated with the positive phases of IOD (lag 0, 3 and 6) and AMO. Eastern Siberia (CTD 5) appeared to be susceptible to increases in BA during +NAO, +AO, +PNA (lag 3, 6) and +IOD (lag 0) events. We also observed similar impacts of AO and NAO (both highly correlated, partly due to shared storm tracks) between the northeastern Atlantic and Arctic regions during the wintertime39. NAO was found to be positively correlated with BA at 9-month time lag, which may indicate fuel build-up connecting to fire activity within the Siberian taiga.
The Eurasian Steppe (or Great Steppe) is among the largest grasslands in the world, extending from Hungary to China. Around 90% of BA in Central Asia was located within Kazakhstan (CTD 2 and 6)11. AMO, TNA, PNA and EA were the most prominent teleconnections related to wildfire activity in this region. Spring EA is associated with decreased summertime precipitation and reduced aboveground net primary productivity40. It has also been observed that a significantly higher sea level pressure and geopotential height in the summer following a warm winter AMO could lead to adiabatic warming and elevated risk of wildfires.
Europe
Approximately, half a million hectares are burned across Europe every year. More than 85% of the BA is located in southern Europe, a Mediterranean fire-prone area11. CT-fire relationships in Europe were weaker compared to other continents, because a considerable part of the continent is under relatively low-fire prone climate regimes in a fragmented landscape, which contributes to fading these relationships.
NAO is among the most notable CTs with significant influence within Europe5. +NAO is associated with northward shifts in the location and intensity of the North Atlantic jet stream responsible for driving moisture transport across Europe. +NAO correlated with reduced BA in temperate continental and Scandinavian Europe (lag 3, Fig. S4), likely owing to increased westerly winds, which bring moist air into Europe producing mild winters accompanied with frequent rain6 prior to the fire season peak. +EA, often in its connection with NAO, correlated with increasing BA across central and western Europe (lag 9) likely due to warm temperature and low precipitation anomalies in late spring41. In contrast, +EA was associated with decreased BA in eastern Europe. Fires in the Mediterranean and temperate regions were mostly related to +PDO, +IOD and +ENSO, which are linked to anomalies in precipitation, promoting vegetation build-up before the fire season peak5,42. TSA positively correlated with BA in the Spanish Mediterranean coast (lag 9), whereas +AMO and +TNA (lag 3, 6 and 9) were predominantly associated with increased BA in Greece and Turkey.
Oceania
The 2019–2020 extraordinary Australian wildfires showed how climatic events can cause unprecedented large-scale impacts through the combination of sustained record-breaking high temperatures with low precipitation10. The equatorial Pacific SST variability is the main driver of observed precipitation patterns although the Indic and Atlantic Oceans also influence Australian precipitation and temperature by altering the tropical Pacific climate through the trans-basin variability mechanism43,44. Two CTDs (2, and 6; Fig. 3) encompassed the main CT-fire relationships in this continent. In northern Australia (CTD 6), BA is enhanced by +AMO, +TNA (lag and 9), +IOD (lag 0, 3 and 6) and -PDO. In the CTD 2, in addition to the effects of the aforementioned CTs, we found an effect of +ENSO in southeastern Australia that brings increased surface air temperature and drought conditions45, potentially coupled with +IOD46. +IOD brings easterly wind anomalies across the Indian Ocean that diminish rainfall over Australia46. Precipitation is enhanced over northern Australia by +PDO in contrast to the eastern and western ends where it is weakened40. TSA positively correlated with BA in southeastern Australia probably due to increased surface air temperature during the Austral summer. As expected, the effects of TNA and AMO in this region attained opposite signs. Also, +EP (lag 3), +WP (lag 6), PNA (lag 3 and 6), +EA (lag 3 and 6) and +NAO (lag 6) appear to relate to increases in BA in the CTD 2 as found by Shi and Touge (2022)47. +SAM (lag 6) was found to be negatively correlated with BA, likely due to concurring cool and wet conditions over southeastern Australia28.
Source: Ecology - nature.com
