Introduction
Vegetation and soils are important components of the global terrestrial carbon pool, together accounting for ∼2060 Pg of carbon, with about three-quarters of this being stored in soils (Scharlemann et al. Reference Scharlemann, Tanner, Hiederer and Kapos2014, Schlesinger & Andrews Reference Schlesinger and Andrews2000). Soils are also primary mediators of global land-atmosphere carbon fluxes, with soil respiration, which results from root respiration and microbial decomposition of organic matter, releasing 75–100 Pg/year back to the atmosphere (Bond-Lamberty & Thomson Reference Bond-Lamberty and Thomson2010, Raich & Schlesinger Reference Raich and Schlesinger1992). Like most biochemical processes, rates of soil respiration are largely governed by temperature and moisture availability which play major roles in regulating both plant growth and soil microbial activity. Soil respiration rates are typically highest in the tropics where plant growth is luxuriant and conditions for decomposition are optimal and lowest in cold and dry biomes where microbial activity is limited by low temperatures and moisture availability, respectively (Raich & Schlesinger Reference Raich and Schlesinger1992). The tropics currently contribute ∼67% of the global soil CO2 efflux (Bond-Lamberty & Thomson Reference Bond-Lamberty and Thomson2010, Hashimoto et al. Reference Hashimoto, Carvalhais, Ito, Migliavacca, Nishina and Reichstein2015, Schlesinger & Andrews Reference Schlesinger and Andrews2000). Future changes in temperature and precipitation regimes in tropical regions are likely to affect soil respiration rates with potentially significant effects on the global C-cycle and atmospheric CO2 levels (Schlesinger & Andrews Reference Schlesinger and Andrews2000).
Besides climatic variables, changes in land use and land cover also influence global terrestrial-atmosphere CO2 fluxes. Since the 1990s, ∼1.14 Pg of C are estimated to have been released back into the atmosphere annually, largely as a result of the conversion of intact forests to secondary forests, pastures and croplands (Ciais et al. Reference Ciais, Sabine, Bala, Bopp, Brovkin, Canadell, Chhabra, DeFries, Galloway, Heimann, Jones, Stocker, Qin, Plattner, Tignor, Allen, Boschung, Nauels, Xia, Bex and Midgley2013, DeFries et al. Reference DeFries, Field, Fung, Collatz and Bounoua1999, Houghton et al. Reference Houghton, House, Pongratz, Van Der Werf, DeFries, Hansen, Quéré and Ramankutty2012, Pan et al. Reference Pan, Birdsey, Fang, Houghton, Kauppi, Kurz, Phillips, Shvidenko, Lewis, Canadell and Ciais2011). Land-cover changes, besides having significant negative impacts on biodiversity, can also alter ecosystem processes by changing the quantity and quality of carbon inputs to soils, changing soil temperature and moisture regimes and by altering hydrological cycles (Delgado-Balbuena et al. Reference Delgado-Balbuena, Arredondo, Loescher, Huber-Sannwald, Chavez-Aguilar, Luna-Luna and Barretero-Hernandez2013, Parr et al. Reference Parr, Lehmann, Bond, Hoffmann and Andersen2014, Raich & Schlesinger Reference Raich and Schlesinger1992, Veldman Reference Veldman, Overbeck, Negreiros, Mahy, Le Stradic, Fernandes, Durigan, Buisson, Putz and Bond2015). However, the impact of such land-cover changes on ecosystem processes, particularly in tropical regions, has not received sufficient attention. More accurate quantification of the effects of land-cover changes on carbon fluxes to the atmosphere is needed for better implementing policies such as REDD + (Reducing Emissions from Deforestation and Degradation), especially in the tropics which account for a substantial proportion of the emissions due to land-cover changes (Le Quere Reference Le Quéré, Raupach, Canadell, Marland, Bopp, Ciais, Conway, Doney, Feely, Foster and Friedlingstein2009).
In this study, we quantified soil CO2 effluxes over three years across different land-cover types in the climatically sensitive tropical montane forest-grassland mosaics of the Western Ghats biodiversity hotspot in India. These mosaics have undergone extensive land-cover changes since the mid-nineteenth century, primarily through the establishment of non-native plantations in grasslands, especially of acacia (Acacia spp.), eucalyptus (Eucalyptus spp.), pine (Pinus spp.) and tea (Camellia sinensis) (Joshi et al. Reference Joshi, Sankaran and Ratnam2018, Sukumar et al. Reference Sukumar, Suresh and Ramesh1995). Specifically, we looked at how land-cover changes influenced both total and temporal patterns of soil respiration and how such changes were mediated by alterations to moisture and temperature regimes.
Methods
Study area
Our study was conducted in the high elevation forest-grasslands mosaics of the upper Nilgiri landscape (11.27° N 76.55° E, elevation: ∼2300 m) in the southern Western Ghats, India (Figure S1). The upper reaches (1200–2650 m) of India’s Western Ghats biodiversity hotspot (8–21°N, 73–78°E) are characterised by stunted evergreen tree forest patches, locally known as sholas, embedded within grasslands, with abrupt transitions between them (Figure S1).
These bi-phasic mosaics are Pleistocene relics that have existed for more than 20,000 years (Sukumar et al. Reference Sukumar, Ramesh, Pant and Rajagopalan1993,Reference Sukumar, Suresh and Ramesh1995) and support a diverse array of plant and animal species, many of which are endemic (see Karunakaran et al. Reference Karunakaran, Uniyal and Rawat1998, Mohandass & Davidar Reference Mohandass and Davidar2009, Robin & Sukumar Reference Robin and Sukumar2002, Schaller Reference Schaller1970, Thomas & Palmer Reference Thomas and Palmer2007). Over 80 species of trees, lianas and shrubs have been documented from the shola forests in the region (Mohandass & Davidar Reference Mohandass and Davidar2009). Rubiaceae, Lauraceae and Myrtaceae are the dominant families in these forests and comprise ∼ 60% of the total stem density (Mohandass & Davidar Reference Mohandass and Davidar2009). Montane grasslands at the site support around 40 species of C4 grasses (Joshi et al. in prep).
Large-scale plantations of alien tree species, chiefly eucalyptus (Eucalyptus globulus), pine (Pinus patula) and wattle (Acacia mearnsii), were established in grasslands of this landscape in the latter half of the nineteenth century (Joshi et al. Reference Joshi, Sankaran and Ratnam2018, Prabhakar Reference Prabhakar1994, Sukumar et al. Reference Sukumar, Suresh and Ramesh1995). Among these, wattle eventually became invasive and now occupies large tracts of former grassland. The landscape is currently characterised by a mix of native and exotic vegetation. The prominent land-cover types include native shola forests (∼452 km2), grasslands (∼139 km2), plantations of eucalyptus and pine and invasive wattle patches (∼277 km2) and commercial tea (∼369 km2) (Arasumani et al. Reference Arasumani, Khan, Vishnudas, Muthukumar, Bunyan and Robin2019). We selected four land-cover types for this study that were in close proximity to each other – native shola forests, grasslands, invasive wattle patches and pine plantations.
Average monthly temperatures in the study area range from a mean minimum of 5°C in January to a mean maximum of 24°C in April (Figure 1) (Caner et al. Reference Caner, Seen, Gunnell, Ramesh and Bourgeon2007). Nocturnal frosts are common in grasslands during the winter (November to March) when temperatures routinely drop down below 0°C, but they are rare occurrences within shola forests. Annual precipitation is spatially and temporally variable across the landscape, ranging from 2500 to 5000 mm, with most of the rainfall received between May and November from the south-west monsoon. (Figure 1). Soils of the area are derived from parent rocks which are gneiss, charnockites and schists (Caner et al. Reference Caner, Seen, Gunnell, Ramesh and Bourgeon2007). However, C and N contents of shola forest soils are significantly greater than that of adjoining grasslands. The C content of shola forest soils and grassland soils in the top 15 cm average 12.9 and 8.5%, respectively, while soil N values average 0.9 and 0.5%, respectively (Raghurama Reference Raghurama2013).
Figure 1. Temperature and precipitation regimes in the study area. (a) Mean minimum and mean maximum monthly temperatures in the study region. (b) Mean monthly precipitation received in the study region. (Error bars in both figures represent standard errors of the mean; source: Indian Meteorological Department (IMD) data for Ooty from 1969 to 2005).
Study design
In October 2014, we identified three replicate sites in each of four different land-cover types: shola forests, grasslands, wattle patches and pine plantations for the study. Replicate sites within each land-cover type were separated by a minimum distance of 100 m. At each site, we established five soil respiration collars spaced at least 1 m apart from each other in randomly selected locations that were representative of the land-cover type. In all cases, collars were established on gentle south facing slopes and >10 m from habitat edges. Respiration collars were 10 cm long open-ended PVC pipes with an inner diameter of 10 cm, inserted 3 cm into the soil with 7 cm exposed above the soil surface. Collars remained permanently installed in soils throughout the study, with a few exceptions when collars were lost (due to animal activity or burnt following a fire event) and had to be replaced. In such cases, collars were replaced immediately, and at least a week before the next measurement. In total, we had 60 respiration collars across the different land-cover categories.
Soil respiration was measured every 2 weeks on average at each of the 60 collars using an EGM-4 Environmental Gas Monitor (EGM-4, PP Systems, USA). All measurements were made between 10 am and 4 pm. All plant material, primarily leaf litter, within collars was carefully removed before respiration measurements were taken and placed back after the measurements. CO2 efflux rates were calculated following the procedure outlined in Marthews et al. (Reference Marthews, Metcalfe, Malhi, Phillips, HuaracaHuasco, Riutta, Ruiz Jaén, Girardin, Urrutia, Butt, Cain and OliverasMenor2014) as follows:
$${R_{UC}} = \;{{{C_{10}} - {C_1}} \over {{t_{10}} - {t_1}}} \times {P \over {{T_a} + \;273.15}}\; \times \;{{{V_d}} \over A}\; \times \;{{44.01\; \times \;0.36} \over {{R_U}}}$$
$${R_C} = {R_{UC}} \times \;{{{V_{d\;}} + \;{V_{{\rm{added}}}}} \over {{V_d}}}$$
Here, Ruc is the soil CO2 efflux calculated without correcting for the added volume of the respiration collar. C10-C1 represents the difference in CO2 efflux between the last 10 readings of each measurement cycle (each cycle normally lasting 124 seconds with 4–5 seconds per reading), and t10-t1 is the time difference, in seconds, between these measurements. In cases where measurements had less than 10 readings, we used the first and last flux values for the calculations. P represents the ambient atmospheric pressure averaged over the period of measurement, Ta the atmospheric temperature in Kelvin, Vd the volume within the soil respiration chamber, A the soil area over which CO2 efflux was measured, and Ru the Universal Gas Constant (8.314 J K–1 mol–1). Equation (2) was used to estimate Rc – the CO2 efflux corrected for the added volume of the respiration collar at the time of measurement. Further details of the methods and calculations of CO2 efflux can be found in Marthews et al. (Reference Marthews, Metcalfe, Malhi, Phillips, HuaracaHuasco, Riutta, Ruiz Jaén, Girardin, Urrutia, Butt, Cain and OliverasMenor2014).
Alongside CO2 efflux measurements, ambient atmospheric temperature, soil temperature and soil moisture were also quantified. Soil temperature and moisture were measured at 0–12 cm depth at three locations around each collar using a handheld long-stem thermometer (HI145-00 and HI145-01, Hanna Instruments, USA) and a FieldScout TDR 200 soil moisture metre (Spectrum Technologies, USA), respectively.
For our analyses, we excluded measurements (58 out of 3355) which showed a decline in CO2 accumulation within collars over the time period of measurement and had a negative slope, that is, they were indicative of measurement errors in the field. All data were then averaged by site (i.e., across the five collars) for the analysis. We analysed the effects of land-cover type, soil temperature and soil moisture on soil respiration rates using a linear mixed-effects model framework with the interaction of land-cover type, soil temperature and soil moisture as fixed effects and site as a random effect. Soil respiration values were log-transformed before analyses to meet the assumptions of homogeneity and normality of residuals. We also calculated annual estimates of soil respiration (g CO2 per m2 per yr) for each land-cover type by scaling up the mean hourly CO2 efflux rates (g CO2 per m2 per hr) estimated for each site for the corresponding year (average hourly efflux rates × 24 × 365).
The R package nlme with lme (Pinheiro et al. Reference Pinheiro, Bates, DebRoy and Sarkar2018) was used to conduct the mixed-effects model analysis. All analyses were conducted using R version 4.0.3 (R Core Team, Reference Team2020).
Results
Temporal dynamics of soil respiration across land-cover
Soil respiration rates across land-cover types varied both seasonally and annually and ranged from 0.02 to 2.94 g CO2 per m2 per hr across the study in the different land-cover types. Soil CO2 efflux typically peaked during the south-west monsoon season (June–October) and declined through the winter (November–January) to lowest values in summer (February to May; Figure 2a). Highest instantaneous respiration rates were recorded in grasslands (average 0.54 g CO2 per m2per hr across the study) and lowest rates from pine soils (average 0.41 g CO2 per m2 per hr1, Table 1; Figure 2a). However, annual soil CO2 effluxes did not differ significantly between land-cover categories (one-way ANOVAs, followed by pairwise comparisons using Tukey’s Honestly Significant Difference (HSD) test, P > 0.15 for all pairwise comparisons for all three years). Across all land-cover categories, annual fluxes were greater in 2016–17 compared to the previous two years (Figure 3).
Figure 2. Patterns of change in (a) soil respiration, (b) soil temperatures 12 cm below the surface and (c) soil moisture in the top 12 cm over time (n = 3). Error bars represent standard errors of the mean (gaps in 2b and 2c are due to lack of measurements during the corresponding period because of instrument failure).
Table 1. Mean soil temperature, soil moisture and soil respiration across the land-cover categories. Range presented in brackets
Figure 3. Annual estimates of soil CO2 effluxes across different land-cover types based on three years of monitoring. Bar graphs on the top show monthly precipitation (starting from October to September next year) measured at study sites during the corresponding year, with the total annual precipitation reported within brackets at the top.
Seasonal soil CO2 effluxes mirrored patterns in soil moisture. Over the course of the study, soil moisture levels ranged from 1% to 85% (volumetric water content), and not unexpectedly, peaked during the monsoon and were lowest in summer (Figure 2b). Soil moisture levels were higher in shola forests (average 40%) compared to other land-cover categories which did not differ from one another (average: 28%; Table 1). Soil temperature was similarly highly variable and ranged from 9.1 to 31.8°C over the course of the study. As expected, soil temperature also showed a distinct seasonal pattern, being highest in summer (April) and lowest during the winter (December) (Figure 2c). Temperatures were consistently higher by 3–5°C, but also more variable, in grasslands and wattle patches compared to closed canopy shola forests and pine plantations (Table 1, Figure 2c).
Interactive effects of land-cover type, soil temperature and moisture on soil respiration
The linear mixed model analysis revealed a marginally significant interactive effect of land-cover type, soil moisture and soil temperature on respiration, collectively explaining 60% of the variation in the data (F = 2.30, df = 3, P = 0.07; Table S1 and Table S2). To visualise this interaction, we modelled soil respiration as a function of temperature for equally spaced soil moisture levels across the different land-cover types (Figure 4). Soil moisture values were fixed at the quantiles of measured instantaneous soil temperature measurements over the study (Figure 4). In both grassland and wattle patches, soil respiration was not affected by soil temperatures when moisture was low but increased with increasing soil temperature at high moisture levels (Figure 4). In pine plantations, respiration was similarly largely unaffected by temperature at low moisture levels but decreased with increasing temperatures when soil moisture was high (Figure 4). In contrast, soil respiration did not seem to be significantly impacted by changes in either soil temperature or moisture in shola forests (Figure 4).
Figure 4. Effect plots depicting the interactive effect of soil moisture and soil temperature on soil CO2 effluxes across the different land-cover types (each column represent soil moisture in %).
Discussion
Our results highlight that soil respiration in all focal land-cover categories was primarily governed by precipitation (Davidson et al. Reference Davidson, Verchot, Cattanio, Ackerman and Carvalho2000, Fan et al. Reference Fan, Yang and Han2015, Sheng et al. Reference Sheng, Yang, Yang, Chen, Xie, Guo and Zou2010), peaking during the monsoon when root growth, microbial activity and litter decomposition are typically high (Singh & Gupta Reference Singh and Gupta1977) and dropping to their lowest levels during the summer. However, intra-annual patterns of precipitation distribution also potentially play a role in regulating total annual soil CO2 effluxes in this system. In our study, the highest annual soil respiration levels were recorded during 2016–17 (Figure 3). These high levels were not attributable to greater total rainfall or higher temperatures during this year compared to earlier years. Although the reasons underlying these high respiration levels are unclear, one plausible explanation could be the more even distribution of precipitation recorded during 2016–17, which potentially lengthened the growing season, resulting in the greater overall fluxes recorded (Figure 3).
Across the different land-cover types, lowest mean instantaneous respiration rates were recorded in pine plantations (Table 1, Figure 2), which is consistent with previous large-scale syntheses that report ∼10% lower respiration in coniferous stands compared to adjacent broad-leaved forests (Raich & Tufekciogul Reference Raich and Tufekciogul2000, Singh & Gupta Reference Singh and Gupta1977). The low rates of soil respiration recorded in pine plantations are potentially attributable to the poor quality and low nutrient content of pine needles and lower soil temperatures, resulting in slower litter decomposition rates and lower soil respiration rates in pine plantations (Balsar & Wixon Reference Balser and Wixon2009; Gao et al. Reference Gao, Huang, Ye, Yue and Chen2018, Singh & Gupta Reference Singh and Gupta1977). Instantaneous soil respiration rates, however, did not differ much between grasslands, shola forests and wattle plantations (Table 1, Figure 2).
Although seasonal patterns of soil respiration were largely governed by soil moisture, instantaneous soil CO2 efflux rates were additionally regulated by soil temperature in ways that differed across land-cover types. In most land-cover types, respiration was largely unaffected, or only marginally increased with soil temperature when moisture levels were low. Low water content is known to limit CO2 production in soils, such that temperature effects are typically manifest only when there is sufficient water to allow root and microbial CO2 production (Schlesinger & Andrews Reference Schlesinger and Andrews2000). However, the nature of temperature–respiration relationships at high moisture levels differed between land-cover types (Figure 3). Respiration increased with temperature when moisture was high in both grassland and wattle patches, decreased in the case of pine plantations and remained largely unchanged in shola forests (Figure 3). The reasons underlying these differences are presently unclear but are potentially related to differences in microbial communities, which play significant roles in soil respiration in these systems (Tiruvaimozhi & Sankaran Reference Tiruvaimozhi and Sankaran2019), and their temperature optima in the different land-cover types. A contemporary study in the same grassland has shown that the contribution of microbes to soil respiration is significantly higher than that of roots and arbuscular mycorrhizal fungi (Tiruvaimozhi & Sankaran Reference Tiruvaimozhi and Sankaran2019). Overall, our results suggest that changes in vegetation type can affect soil C cycling not only by changing above and below-ground C stocks and altering microclimate but also by altering the temperature and moisture sensitivity of soil respiration, potentially through changes in soil microbial communities, even in the absence of any underlying differences in soil type.
Conclusions
Our results indicate that different vegetation types growing adjacent to one another under the same climatic and environmental conditions can have different soil respiration rates and different sensitivities of soil respiration to temperature and moisture. Soil respiration is a combination of both autotrophic (root) and heterotrophic activity (microbes and soil fauna) (Hanson et al. Reference Hanson, Ducklow and Field2000) and is influenced by several factors, including substrate quality and quantity, microbial activity, composition and biomass, and soil temperature and moisture (Bowden et al. Reference Bowden, Davidson, Savage, Arabia and Steudler2004 and references therein). All of these likely differ between land-cover types, but their relative contributions to the observed differences in soil respiration between land-cover types in our study remains unknown. Continued invasions and changes in land-cover, coupled with changes in temperatures and rainfall climatology, are likely to further alter soil respiration in this system in the future. Given the importance of soil respiration for net ecosystem C balance and global carbon budgets, a better understanding of the factors controlling soil respiration in different land-cover types is critical in order to predict the long-term responses of C-cycles in this ecosystem.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0266467424000142
Acknowledgements
We thank Ministry of Earth Sciences, India (grant number: MoES/NERC/16/02/10 PC-11) and National Centre for Biological Sciences, Bangalore, for financial support for conducting this study. We are grateful to Tamil Nadu Forest Department for permits. We thank Dr Jagdish Krishnaswamy (ATREE) and Dr Ravi Bhalla (FERAL) for sharing precipitation data, and Dr Kavita Isvaran, Nachiket Kelkar, Dr Sandip Pulla and Dr Varun Varma for help with the analysis. We also thank Amol, Kavita, Manaswi, Selvakumar, Susilan, Yadugiri, the FERAL field team in the Nilgiris, residents of Emerald village and the frontline staff of the Nilgiri South Forest Division of Tamil Nadu Forest Department for all their help during field data collection.
Author contributions
Atul Arvind Joshi: conceptualisation, methodology, formal analysis, investigation and writing – original draft. Jayashree Ratnam: methodology and writing – review and editing. Harinandanan Paramjyothi: investigation and writing – review and editing. Mahesh Sankaran: conceptualisation, methodology, formal analysis, funding acquisition, supervision and writing – review and editing.
