Introduction
Tropical forests store about 55% of the global terrestrial Carbon (C), of which nearly a third (∼32%) is sequestered in soils (Pan et al. 2011). Given that soils are so C-rich, the release of C from soils is of much importance and can play critical roles in regulating atmospheric CO2 concentrations and regional and global climate. Soil respiration (Rs) – comprising both autotrophic (root) and heterotrophic (bacterial and mycorrhizal) respiration – is a major contributor to the land-atmosphere C flux, releasing ∼75–98 Pg y-1 of C back to the atmosphere globally (Bond-Lamberty and Thomson Reference Bond-Lamberty and Thomson2010a; Raich and Potter 1995), with tropical forests amongst the largest contributors of these soil-derived atmospheric C fluxes (Raich and Schlesinger Reference Raich and Schlesinger1992; Warner et al. Reference Warner, Bond-Lamberty, Jian, Stell and Vargas2019).
Rates of soil respiration (Rs) are governed by complex interactions between a range of biotic and abiotic drivers, of which climatic factors such as temperature and moisture availability are particularly important (Bond-Lamberty and Thomson Reference Bond-Lamberty and Thomson2010a; Hursh et al. Reference Hursh, Ballantyne, Cooper, Maneta, Kimball and Watts2017; Lei et al. Reference Lei, Guo, Zeng, Zhou, Gao and Yang2021; Raich and Tufekciogul 2000). Future projected changes in both temperature and rainfall patterns are thus likely to have significant impacts on soil-derived land-atmosphere fluxes of CO2, with potential feedbacks on regional and global climates. Although temperature explains much of the variability in Rs across different biomes, other factors such as rainfall, soil C and C inputs to soils (e.g., litterfall) appear to be equally or more important in explaining observed variability in Rs within biomes (Bond-Lamberty and Thomson Reference Bond-Lamberty and Thomson2010a; Huang et al. 2020; Hursh et al. Reference Hursh, Ballantyne, Cooper, Maneta, Kimball and Watts2017; Raich et al. Reference Raich, Potter and Bhagawati2002). For example, in tropical biomes, Rs is often uncorrelated or only weakly correlated with temperature (Davidson et al. Reference Davidson, Verchot, Cattânio, Ackerman and Carvalho2000; Rubio and Detto Reference Rubio and Detto2017), possibly a consequence of the low intra-annual variability in temperatures experienced by these biomes when compared to temperate ecosystems.
On the other hand, soil moisture has been shown to be an important driver of Rs in tropical systems (Adachi et al. Reference Adachi, Ishida, Bunyavejchewin, Okuda and Koizumi2009; Hashimoto et al. Reference Hashimoto, Tanaka, Suzuki, Inoue, Takizawa, Kosaka, Tanaka, Tantasirin and Tangtham2004; Rubio and Detto Reference Rubio and Detto2017; Valentini et al. Reference Valentini, Sanches, de Paula, Vourlitis, de Souza Nogueira, Pinto and de Almeida Lobo2008; Wood et al. Reference Wood, Detto and Silver2013). However, the exact nature of this relationship is unclear and appears variable, with studies reporting positive (Hashimoto et al. Reference Hashimoto, Tanaka, Suzuki, Inoue, Takizawa, Kosaka, Tanaka, Tantasirin and Tangtham2004), negative (Epron et al. Reference Epron, Bosc, Bonal and Freycon2006) and parabolic relationships of Rs with soil moisture (Adachi et al. Reference Adachi, Ishida, Bunyavejchewin, Okuda and Koizumi2009; Gaumont-Guay et al. Reference Gaumont-Guay, Black, Griffis, Barr, Jassal and Nesic2006; Rubio and Detto Reference Rubio and Detto2017; Sotta et al. Reference Sotta, Meir, Malhi, Donato nobre, Hodnett and Grace2004; Takahashi et al. Reference Takahashi, Hirai, Limtong, Leaungvutivirog, Panuthai, Suksawang, Anusontpornperm and Marod2011). Soil moisture can also mediate the effect of temperature on Rs, with low moisture typically constraining the response of Rs to temperature (Deng et al. Reference Deng, Hui, Zhang, Zhou, Liu, Liu, Chu and Li2012; Flanagan and Johnson Reference Flanagan and Johnson2005; Hursh et al. Reference Hursh, Ballantyne, Cooper, Maneta, Kimball and Watts2017; Wen et al. Reference Wen, Yu, Sun, Li, Liu, Zhang, Ren, Fu and Li2006; Wood et al. Reference Wood, Detto and Silver2013) although the opposite – higher temperature sensitivity of Rs at low soil moisture – has also been observed (Guntiñas et al. Reference Guntiñas, Gil-Sotres, Leirós and Trasar-Cepeda2013).
As climate predictions for the future include changes in both temperature and rainfall patterns (Intergovernmental Panel on Climate Change (IPCC) 2023), our ability to predict responses of the C cycle hinges upon a better understanding of the factors that influence soil C fluxes across different biomes. In particular, the limited numbers of empirical observations from tropical forests, which have amongst the highest predicted Rs rates, preclude good estimates of the variance in these rates, and have been identified as a major factor limiting robust global estimates of Rs, and for calibrating and reducing uncertainty in earth system models (Hursh et al. Reference Hursh, Ballantyne, Cooper, Maneta, Kimball and Watts2017). The tropical forests of South Asia, in particular, are severely under-represented in most global syntheses of soil respiration (Akande et al. Reference Akande, Ma, Huang, He and Chang2023; Bond-Lamberty and Thomson Reference Bond-Lamberty and Thomson2010a, Reference Bond-Lamberty and Thomson2010b; Hursh et al. Reference Hursh, Ballantyne, Cooper, Maneta, Kimball and Watts2017; Lei et al. Reference Lei, Guo, Zeng, Zhou, Gao and Yang2021).
In this study, we characterised the response of soil respiration to changes in moisture availability and temperature in two seasonal wet evergreen forest plots in the Western Ghats, India. These forests differ from other seasonal tropical wet forests in the length of the dry season (> 6 months) and in the intensity of rainfall they receive during the wet season (>2500 mm). Specifically, our objectives were to (1) quantify and characterise inter-annual and seasonal variation of soil respiration and (2) examine the effects of temperature and moisture availability on soil respiration rates at annual, seasonal and instantaneous temporal scales. Based on the published literature, we expected soil moisture availability to exert stronger controls over Rs than soil temperature in this seasonal tropical wet forest. Further, we expected Rs to be weakly positively correlated with temperature, but show a parabolic relationship with soil moisture.
Materials and methods
Site description
We conducted the study at two long-term wet evergreen forest monitoring sites – Hosagadde (14.478° N, 74.760° E) and Mulagunda (14.471°N, 74.693°E) – located in the Uttara Kannada district in the state of Karnataka in the central Western Ghats, India. Both sites are part of the Long-term Ecosystem Monitoring Network – India (LEMoN – India). Both sites have similar vegetation with Tetrapilus dioicus (Roxb.) L.A.S.Johnson, Hopea ponga (Dennst.) Mabb., Psydrax dicoccos Gaertn., Holigarna arnottiana Hook.f., Knema attenuata (Wall. ex Hook.f. & Thomson) Warb. and Syzygium densiflorum Wall. ex Wight & Arn. being the dominant tree species in Hosagadde, and Tetrapilus dioicus, Garcinia gummi-gutta (L.) N.Robson, Syzygium densiflorum and Knema attenuata being the dominant tree species in Mulagunda. Both sites are at an elevation of ∼550 m and have a long-term mean annual temperature of 24 °C. The mean annual rainfall during the study period was 3730 mm and 4228 mm at Hosagadde and Mulagunda, respectively (University of East Anglia Climatic Research Unit; Harris et al. Reference Harris, Osborn, Jones and Lister2020). We classified months with more than 100 mm rainfall as the wet season (June – October) and months with <100 mm rainfall as the dry season (November – May) (Figure 1A). Soils in the region have been characterised as a mixture of Eutric Nitosols and Acrisols under the FAO system (FAO-UNESCO 1988; Krishnaswamy et al. Reference Krishnaswamy, Chappel, Bhalla, Sankaran, Vaidyanathan, Badiger and Varghese2017; Rome, FAO-UN Reference Rome1998). Values for soil texture, water holding capacity, soil C and soil N in the top 20cm of soils from both plots have been presented in Table 1.
Monthly variation in means and standard errors of A) monthly rainfall, B) soil moisture, C) soil temperature and D) soil respiration (Mg C ha-1 month-1). Months within dashed lines (Jun – Oct) represent the wet season.
Soil texture (shown as % sand and clay), water holding capacity (WHC), soil C and soil N from the top 20cm soil layer at Hosagadde and Mulagunda. Values reported for soil C and N include the mean and standard error based on 10 soil cores
Field measurements
We established permanent vegetation monitoring plots of 1 hectare area each at Hosagadde and Mulagunda in 2013 and 2014, respectively, based on the RAINFOR-GEM (Marthews et al. Reference Marthews, Riutta, Oliveras Menor, Urrutia, Moore, Metcalfe, Malhi, Phillips, Huaraca Huasco and Ruiz Jaén2014) and CTFS protocols (Condit Reference Condit and Condit1998). All trees >10 cm girth at breast height (GBH) in plots were identified, spatially mapped, tagged and their GBH measured. Plots were re-censused annually. We additionally quantified soil respiration in plots following the protocols outlined in Marthews et al. (Reference Marthews, Riutta, Oliveras Menor, Urrutia, Moore, Metcalfe, Malhi, Phillips, Huaraca Huasco and Ruiz Jaén2014). In each plot, we established 25 polyvinyl carbonate (PVC) soil collars (10 cm height and 10.6 cm diameter) by inserting them 5–6 cm into the soil. We used portable infra-red gas analysers (IRGA) EGM-4 and EGM5 (PP Systems, USA) to measure instantaneous Rs (g CO2 m-2.h-1). Soil respiration was measured fortnightly from January 2014 to August 2020 in Hosagadde, and from January 2015 till August 2020 in Mulagunda.
Along with instantaneous collar-level Rs readings, we measured soil temperature using a digital thermometer with a 125mm stainless steel penetration probe (Hanna instruments). We also quantified instantaneous volumetric soil moisture at 30 cm depth during each sampling occasion using a FieldScout TDR 200 soil moisture metre (Spectrum Technologies, USA) at 3 points beside each soil collar, which were then averaged. For ambient air temperature measurements, we used data from Thermochron i-button loggers (Maxim Integrated) placed at the centre and four corners of the 1 ha plot which recorded temperatures at 30 min intervals. Daily rainfall measurements were obtained from a tipping bucket rain gauge installed adjacent to the plot (Krishnaswamy et al., unpublished data).
Soil Respiration measurement and calculations
We calculated Rs following Marthews et al. (Reference Marthews, Riutta, Oliveras Menor, Urrutia, Moore, Metcalfe, Malhi, Phillips, Huaraca Huasco and Ruiz Jaén2014) as follows:
$${R_{uc}} = \;\left( {{{{C_{10}} - {C_1}}}\over{{{t_{10}} - {t_1}}}} \right)\left( {{P}\over{{{T_a}}}} \right)\left( {{{{V_d}}}\over{A}} \right)\left( {{{44.01*0.36}}\over{{{R_u}}}} \right)$$
$${R_s} = {R_{uc}}\left( {{{{V_d} + {V_{added}}}}\over{{{V_d}}}} \right)$$
where R uc is the uncorrected soil respiration (g CO2 m-2 h-1), estimated as the difference in CO2 concentrations C 1 and C 10 , measured between times t 1 and t 10 , where t 10 was the last reading and t 1 was typically 40 seconds before t10, P is the ambient pressure (mb) at time t 10 as recorded by the IRGA, T a is air temperature at time t 10 in Kelvin, V d the volume of the soil respiration chamber (0.001229 m3), A is the ground area of the soil collar over which Ruc was measured, and R u , the Universal Gas Constant (8.314 J K-1 mol-1). R s is the soil respiration corrected for the additional volume of the soil collar V added (in m3) based on the height (h cm) of the PVC collar protruding above the soil surface.
In order to maintain data quality during field measurements, we compared the CO2 concentration at the end of each measurement to that at the start, and only included measurements where we observed a linear increase of CO2 between t1 and t10 (Marthews et al. Reference Marthews, Riutta, Oliveras Menor, Urrutia, Moore, Metcalfe, Malhi, Phillips, Huaraca Huasco and Ruiz Jaén2014). We also repeated measurements where the end CO2 value was lower than the start value. If the end measurement was equal to the start concentration, i.e., no respiration, we repeated the measurement for a total of three times before recording zero respiration. Additionally, during data processing, we excluded eight negative Rs values (4 each in Hosagadde and Mulagunda) that were missed during the field data check. Annual estimates of Rs at the plot level were obtained by spatially averaging hourly flux rates (gCO2m-2h-1) across all 25 collars in plots and scaling these up for the corresponding year (average hourly flux rates × 24 × 365). Seasonal fluxes were similarly obtained by first estimating average hourly flux rates for each season and then scaling them up for the duration of the season.
Statistical methods
We assessed the effects of temperature and moisture availability on Rs at multiple temporal scales – annual, seasonal and instantaneous. At the annual scale, we used linear models to test the relationship between annual Rs and (a) annual rainfall, (b) mean annual temperature, (c) number of rainy days, and (d) number of days with heavy rainfall (>50 mm). At the seasonal scale, we used a linear mixed effects model (LMMs) to test whether Rs values differed between wet and dry seasons, with season (wet, dry) included as a fixed factor and soil collar number nested within year within site as a random factor to account for inherent differences between soil collars within each site, and non-independence of measurements over time at each collar. We had 11–24 data points/collar/year/site for a total of 6886 observations. Finally, we used linear mixed effects models to test whether spatially averaged temperature and moisture levels could predict instantaneous Rs. For this, we used spatially averaged Rs (averaged over 25 collars) within each sampling period for each site as the response variable and spatially averaged temperature (as recorded by the IRGA) and a quadratic term with spatially averaged soil moisture and site ID as fixed factors. We used the calendar year as a random factor to account for inter-annual differences in soil respiration rates. We tested two candidate models: the first included an interaction term between temperature and soil moisture and the second did not have an interaction included. In all, we had 13–24 sampling times/year/site and a total 276 spatially averaged Rs values. Rs values were square root transformed to meet model assumptions of linearity and homoscedasticity.
All analyses were carried out using R version 4.3.0 (R Core Team 2023). We used the R packages glmmTMB for all models, DHARMa for model diagnostics, MASS to inform us about suitable transformations to meet model assumptions, ggeffects, ggpubr and ggplot2 for data visualisation and to plot model predictions, and tidyverse for data cleaning and organisation (Brooks et al. Reference Brooks, Kristensen, Benthem, Magnusson, Berg, Nielsen, Skaug, Maechler and Bolker2017; Hartig Reference Hartig2022; Kassambara Reference Kassambara2023; Lüdecke Reference Lüdecke2018; Venables and Ripley Reference Venables and Ripley2002; Wickham Reference Wickham2016; Wickham et al. Reference Wickham, Averick, Bryan, Chang, McGowan, François, Grolemund, Hayes, Henry, Hester, Kuhn, Pedersen, Miller, Bache, Müller, Ooms, Robinson, Seidel, Spinu, Takahashi, Vaughan, Wilke, Woo and Yutani2019).
Results
Rainfall, soil moisture and temperature seasonality
During the study period, annual rainfall varied between 2800 and 5300 mm in Hosagadde and between 3700 and 7600 mm in Mulagunda. Rainfall was very seasonal, occurring primarily during the south-west monsoon from June to October. Hosagadde received very heavy rains ( >1000mm month-1) in July and August and Mulagunda between June and August. Outside of the monsoon, both sites received <100 mm rainfall month-1 (Figure 1A). Similar to rainfall variations, soil moisture levels were also strongly seasonal (Figure 1B). In both sites, they were lowest from February to April (average volumetric water contents of 13.9 cm3cm-3 and 9.1 cm3cm-3 in Hosagadde and Mulagunda, respectively), and highest between July and October (35.6 cm3cm-3 and 30.2 cm3cm-3 in Hosagadde and Mulagunda, respectively). Mean annual ambient air temperatures during the study period at both sites were ∼22°C. Soil temperatures (at 125mm depth) ranged between 17–26°C in Hosagadde and 19 to 30°C in Mulagunda (Figure 1C).
Soil Respiration
Annual Rs ranged from 10.9 – 20.2 Mg C ha-1y-1 in Hosagadde (mean = 14.58, SD = 3.38 cv = 0.23), and 8.44 – 12.05 Mg ha-1 y-1 in Mulagunda (mean = 9.85, SD = 1.6, cv = 0.16) (Figure S1), and was negatively related to total annual rainfall at both sites (p < 0.01) (Figure 2A). Annual Rs was not correlated with annual mean temperature (p = 0.8) or number of rainy days in a year (p = 0.23) (Table S1). However, it was significantly negatively correlated with the number of days where the rainfall was higher than 50 mm (p = 0.002) (Figure 2B). Soil respiration in both sites was significantly lower during the wet season than during the dry season (p < 0.001) (Figure 2C). In both sites, Rs was strongly seasonal, reaching high levels pre- and post-monsoon, and was lowest during the monsoon in July and August, months which received >1000mm rainfall (Figure 1D). At the instantaneous scale, the model without the interaction term between temperature and soil moisture had better support (lower AIC, Table S3) than the one with the interaction term included. Spatially averaged instantaneous Rs showed a parabolic relationship with mean soil moisture at both sites (significant quadratic term; p < 0.001) with respiration peaking at ∼21 cm3cm-3 soil moisture (Figure 2D). For any given soil moisture level, respiration rates increased with increasing soil temperatures (Figure 2D).
Soil respiration was negatively correlated with rainfall at annual time scales (A) and was lower in years with a greater number of very wet (>50mm rainfall) days (B). Soil respiration was lower in the wet season when compared to the dry season (C) at both sites. Soil respiration was also correlated with soil moisture and temperature at the instantaneous timescale(D). In A, B &D, lines and bands show estimates and confidence intervals from models and points show data. In C, points and bars show means and standard errors of instantaneous soil respiration within each season.
Discussion
In this highly seasonal rainforest in the Western Ghats, rainfall was the dominant driver of soil respiration (Rs), influencing Rs at annual, seasonal and instantaneous time scales. Rs was lowest during periods of high soil moisture availability, i.e., during the wet season and in wetter years, and highest when soil moisture was low, i.e., during the dry season and in drier years. In contrast, temperature was only of secondary importance, with warmer temperatures leading to higher rates of instantaneous soil respiration for any given soil moisture level.
Instantaneous soil respiration showed a parabolic relationship with soil moisture availability, peaking at ∼21 cm3cm-3 VWC and declining with both increases and decreases in soil moisture. Dry conditions can reduce Rs because of limited water availability for both microbial and root respiration, while high moisture availability can reduce rates of CO2 diffusion in waterlogged soils while also reducing microbial activity and root respiration as a result of soils becoming anoxic (Adachi et al. Reference Adachi, Ishida, Bunyavejchewin, Okuda and Koizumi2009; Cusack et al. Reference Cusack, Ashdown, Dietterich, Neupane, Ciochina and Turner2019; Gaumont-Guay et al. Reference Gaumont-Guay, Black, Griffis, Barr, Jassal and Nesic2006; Rubio and Detto Reference Rubio and Detto2017; Valentini et al. Reference Valentini, Sanches, de Paula, Vourlitis, de Souza Nogueira, Pinto and de Almeida Lobo2008; Wood et al. Reference Wood, Detto and Silver2013). Similar parabolic relationships of Rs with soil moisture have also been reported from other tropical forests across the globe (Adachi et al. Reference Adachi, Ishida, Bunyavejchewin, Okuda and Koizumi2009; Cusack et al. Reference Cusack, Ashdown, Dietterich, Neupane, Ciochina and Turner2019; Koehler et al. Reference Koehler, Corre, Veldkamp and Sueta2009; Matson et al. Reference Matson, Corre, Langs and Veldkamp2017; Schwendenmann et al. Reference Schwendenmann, Veldkamp, Brenes, O’Brien and Mackensen2003; Sotta et al. Reference Sotta, Meir, Malhi, Donato nobre, Hodnett and Grace2004). The soil moisture at peak respiration at our sites is comparable with reported values from Southeast Asian forests (∼21 cm3cm-3) (Adachi et al. Reference Adachi, Ishida, Bunyavejchewin, Okuda and Koizumi2009; Takahashi et al. Reference Takahashi, Hirai, Limtong, Leaungvutivirog, Panuthai, Suksawang, Anusontpornperm and Marod2011), but lower than those from the Neotropics (30–40 cm3cm-3) (Gaumont-Guay et al. Reference Gaumont-Guay, Black, Griffis, Barr, Jassal and Nesic2006; Rubio and Detto Reference Rubio and Detto2017; Sotta et al. Reference Sotta, Meir, Malhi, Donato nobre, Hodnett and Grace2004; Wood et al. Reference Wood, Detto and Silver2013), potentially a consequence of differences in soil texture across these sites. Soil water content for optimal Rs has previously been shown to be correlated with soil texture and clay content (Balogh et al. Reference Balogh, Pintér, Sz, Cserhalmi, Papp and Nagy2011; Cable et al. Reference Cable, Ogle, Williams, Weltzin and Huxman2008).
Patterns of seasonal Rs in our sites are consistent with earlier results from neotropical forests that report lower Rs in wet seasons relative to dry seasons in high rainfall sites ( >4000 mm MAP). In contrast, studies from drier sites report higher wet season Rs relative to dry season Rs (Cusack et al. Reference Cusack, Ashdown, Dietterich, Neupane, Ciochina and Turner2019; Fernandes et al. Reference Fernandes, Bernoux, Cerri, Feigl and Piccolo2002; Hashimoto et al. Reference Hashimoto, Tanaka, Suzuki, Inoue, Takizawa, Kosaka, Tanaka, Tantasirin and Tangtham2004; Koehler et al. Reference Koehler, Corre, Veldkamp and Sueta2009; Matson et al. Reference Matson, Corre, Langs and Veldkamp2017; Rubio and Detto Reference Rubio and Detto2017). Our results are also in accordance with experimental dry-down studies that report increases in Rs following rainfall reductions in wet sites and decreases in Rs with drought in drier sites (Cleveland et al. Reference Cleveland, Wieder, Reed and Townsend2010; van Straaten et al. Reference van Straaten, Veldkamp and Corre2011; Wood et al. Reference Wood, Detto and Silver2013). Thus, at a global scale, the nature of the relationship between seasonal Rs and rainfall is likely determined by where sites lie along the dry-wet gradient.
In both our study sites, annual soil respiration was negatively correlated with the frequency of high rainfall events ( >50mm day-1), but was not correlated with the frequency of rainy days, indicating that rainfall intensity is a key factor in limiting soil respiration in these forests. Not only do these sites receive extremely high rainfall but most of this rainfall occurs during the five monsoon months, with the wettest months receiving >1000mm rainfall month-1, while the rest of the year is mostly dry. Inhibition of soil respiration following extreme rain events has also been previously reported from other Asian monsoonal forests (Yu et al. Reference Yu, Chiang, Lai, Tsai and Wang2021) and suggests that future increases in the frequency and intensity of extreme rain events are likely to significantly impact annual soil respiration rates in these forests. Temperature, on the other hand, did not appear to be a strong driver of annual soil respiration rates in our study site, as has also been reported from other tropical forests (Deng et al. Reference Deng, Hui, Zhang, Zhou, Liu, Liu, Chu and Li2012, Reference Deng, Zhang, Han, Chu, Zhang and Hui2018; Rubio and Detto Reference Rubio and Detto2017), possibly because it was largely invariant, with a range of less than 10 degrees in both of our sites.
Despite our two study sites being in close proximity to one another, soil respiration rates differed substantially between the two sites. Mean annual Rs ranged from 8.44 – 12.05 Mg.ha-1.yr-1 (mean = 9.85 Mg.ha-1.yr-1) at Mulagunda and from 10.9 – 20.2 Mg.ha-1.yr-1 (mean = 14.58 Mg.ha-1.yr-1) at Hosagadde, as compared to the global mean of ∼13.1 Mg.ha-1.yr-1 reported for tropical broadleaved forests (Raich et al. Reference Raich, Potter and Bhagawati2002; Warner et al. Reference Warner, Bond-Lamberty, Jian, Stell and Vargas2019). The higher rainfall levels experienced in Mulagunda (Figure 1A) are a potential contributor to the lower fluxes at this site. However, even for the same level of soil moisture, instantaneous Rs tended to be lower in Mulagunda compared to Hosagadde (Figure 2D). Additionally, soil C levels are known to be positively correlated with heterotrophic soil respiration, as soils with higher organic carbon have more substrates for microbial respiration (Wang et al. Reference Wang, He, Wang and Liu2013; Xu and Shang Reference Xu and Shang2016). It is likely that in addition to high rainfall inhibiting soil respiration at Mulagunda, soil C levels, which are much higher in Hosagadde, enable higher soil respiration rates than Mulagunda. Finally, other factors including edaphic properties – which have been shown to exert significant controls on soil respiration rates, sometimes more so than climatic factors (Haaf et al. Reference Haaf, Six and Doetterl2021), also potentially contribute to the observed differences (Davidson et al. Reference Davidson, Savage, Bolstad, Clark, Curtis, Ellsworth, Hanson, Law, Luo, Pregitzer, Randolph and Zak2002; Yao et al. Reference Yao, Ciais, Viovy, Li, Cresto-Aleina, Yang, Joetzjer and Bond-Lamberty2021).
Our study, the first to detail soil respiration at multiple temporal scales from a highly seasonal rainforest in the Western Ghats of India contributes to the growing body of literature on controls of soil respiration across the tropics. The combination of very high rainfall and extreme seasonality in these sites places them in a unique climatic niche globally and adds valuable insights into our broader understanding of the potential relationships between rainfall regimes and soil respiration. Our key finding that both high annual rainfall and extremely rainy days suppress soil respiration indicates that future changes in the climatology of the monsoon in this region are likely to significantly influence C effluxes from this system in the future. Experimental studies that manipulate the intensity and frequency of heavy rainfall events can help to better quantify these patterns and thereby better parameterise earth system models. Further, the separation of autotrophic and heterotrophic components of respiration can allow us to quantify the sensitivity of each component and better understand the biology underlying carbon fluxes in this ecosystem.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S0266467425100126
Acknowledgements
We would like to thank Mahesha from Amminhalli for collecting data throughout the study with HVR. Thanks to Jagdish Krishnaswamy and FERAL for providing us with rainfall data, Balachandra Hedge for maintaining the weather station at Hosagadde, Sandeep Pulla, Anand MO and Deepak Barua for discussions during analysis, and BEER Lab folks at NCBS, especially Manaswi Raghurama and Mayank Kohli for their detailed suggestions on the manuscript. We would also like to thank Abhirami Ravichandran, Anish Paul and Arpitha Jayanth for their help with soil analysis. We would like to thank the LEMoN-India network for making this study possible. Finally, we would like to thank the Karnataka Forest Department for granting permission to conduct the study.
Competing interests
The authors declare no competing interests.
Financial support
We thank and acknowledge funding provided by the Changing Water Cycle programme, jointly administered by the Ministry of Earth Sciences (Grant Ref: MoES/NERC/16/02/10 PC-11), Government of India and Natural Environment Research Council, United Kingdom (Grant Ref: NE/I022450/1); The Science and Engineering Research Board, Department of Science and Technology, Government of India (Grant No: EMR/2016/003722); and the National Centre for Biological Sciences; Department of Atomic Energy, Government of India, Grant/Award Number: RTI 4006. MS also acknowledges funding provided by Science and Engineering Research Board, Department of Science and Technology, Government of India (Grant No: DIA/2018/000038) and the R. M. Tulpule Charitable Trust.
