- Email: [email protected]
Soil carbon dioxide fluxes in a mixed floodplain forest in the Czech Republic
European Journal of Soil Biology 82 (2017) 35e42
Contents lists available at ScienceDirect
European Journal of Soil Biology journal homepage: http://www.elsevier.com/locate/ejsobi
Soil carbon dioxide fluxes in a mixed floodplain forest in the Czech Republic Manuel Acosta*, Eva Darenova, Jirí Dusek, Marian Pavelka Global Change Research Institute, Czech Academy of Sciences, B elidla 4a, Brno 603 00, Czech Republic
a r t i c l e i n f o
a b s t r a c t
Article history: Received 19 April 2017 Received in revised form 2 August 2017 Accepted 15 August 2017
Floodplain forests belong among the most productive, dynamic and diverse ecosystems on Earth. Only few studies have focused on the carbon dioxide fluxes of these ecosystems. Therefore, this study investigated the spatial heterogeneity in soil CO2 efflux in a floodplain forest located in the southeast of the Czech Republic. The study also examined which environmental parameters influence soil CO2 efflux. Moreover, using these obtained measurements a soil CO2 efflux model was applied. To achieve the aims of this study, soil CO2 efflux on 30 positions in 16 campaigns was measured from May to November during the growing season 2016. The efflux during the experiment period ranged from 1.59 to 8.54 mmolCO2 m2 s1. The highest soil CO2 effluxes were observed during the summer period while the lowest values were measured during the autumn. A strong relationship between soil CO2 efflux and soil temperature was found (R2 ¼ 0.79). The estimated mean Q10 for the whole 30 positions was of 2.23. We determined that the spatial heterogeneity of soil CO2 efflux was 20% during our study. The cumulative amount of carbon forest floor released from our experimental forest site calculated from our model was 7.4 (±1.1) tC ha1 y1 for 2016. Such data are important for developing our knowledge and understanding about carbon dynamics and to improve carbon models for these ecosystems types. © 2017 Elsevier Masson SAS. All rights reserved.
Handling editor: Yakov Kuzyakov Keywords: CO2 Soil temperature Soil moisture Spatial heterogeneity Q10
1. Introduction Floodplain forests belong among the most productive, dynamic and diverse ecosystems on Earth [1,2]. Due to fluctuations in river flow and subsequent alternation between flooding and drying, floodplain ecosystems are in a state of dynamic equilibrium [3]. Within the floodplains, floods and geomorphic processes interact to create a shifting mosaic of habitat patches [1] where the vegetation patterns are driven by disturbance intensity, inundation regime and by geological and soil properties. With the exception of the most dynamic and frequently disturbed areas, the majority of floodplain vegetation consists of forests of various age and tree species composition. These floodplain forests are highly productive and provide a wide range of ecosystem services such as high biodiversity, flood water retention, a nutrient sink, groundwater recharge, carbon sequestration, timber production, recreational facilities and aesthetic value [4].
* Corresponding author. E-mail addresses: [email protected] (M. Acosta), darenova.e@czechglobe. cz (E. Darenova), [email protected] (J. Dusek), [email protected] (M. Pavelka). http://dx.doi.org/10.1016/j.ejsobi.2017.08.006 1164-5563/© 2017 Elsevier Masson SAS. All rights reserved.
Many studies have investigated floodplain forests from different ecological points of view such as plant composition [5], tree physiology and morphology [6,7], flooding regime and water table variations [8], and soil acidity and nutrient concentrations [9,10]. But few studies have focused on the effect of flood dynamics on the greenhouse gas budgets and carbon storage of riparian ecosystems [11e13]. In large-scale research projects of the last 15 years (CarboEurope IP, Carbomont, Nitroeurope etc.) floodplain forests have been ignored; mostly because they are ecosystems that are not representative of a wider region. Due to their comparatively small areal extension, floodplain forests are rarely captured in monitoring systems along a regular pattern e.g. along transects and grids. Nevertheless, floodplain forests may play an important role in regional carbon cycling and total greenhouse gases (GHG) dynamics. Floodplain forests in temperate areas are highly productive and store carbon (C) in large quantities when they are compared with upland forests. A detailed assessment also showed that fine root biomass as well as above ground plant biomass significantly varied between flooded and diked areas [14]. However, a move from stock data towards a conceptual understanding of ecosystem C dynamics is necessary to assess the role of floodplain forests in C
36
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
sequestration. The reason for the current lack in detailed C flux data is likely related to obstacles linked to the intensive dynamics of the system, such as additional C import and export by flooding, but might also be attributed to the impracticalities of in situ research in periodically flooded areas. The present pilot study was designed to address the general need for basic information about soil CO2 efflux heterogeneity and specifically to gain information about the factors influencing soil CO2 efflux heterogeneity in floodplain forest. Therefore, the aims of our study were i) to quantify soil CO2 efflux (SR) and to estimate the spatial heterogeneity in SR during the growing season 2016 in a floodplain forest located in the southeast of the Czech Republic, ii) to determine the influence of environmental parameters as soil temperature and soil moisture on SR and iii) to create a model of SR based on measured data during the vegetation season 2016. 2. Material and methods 2.1. Site description The experiment site is situated 6.5 km to the north of the confluence of Morava and Dyje rivers (48 40.090 N, 16 56.780 E). It is formed of an alluvial plain. Long-term average annual precipitation is around 550 mm and mean annual temperature is 9.3 C. The experimental site is composed by typical hardwood species. The main tree species composition is English oak (Quercus robur L.), Narrow-leaved ash (Fraxinus angustifolia Vahl), hornbeam (Carpinus betulus L.) and linden (Tilia cordata Mill.). The stand age is 110 years and height 27 m. The herbal layer and understory of the forest site is characterized by Allium ursinum L., Anemone ranunculoides L., Lathyrus vernus (L.) Bernh., Galium odoratum (L.) Scop., Carex sylvatica Huds. and Acer campestre L. The predominating soil types are Eutric Humic Fluvisol, Haplic Fluvisol, and Eutric Fluvisol (according FAO 2014 Classification) with minimal soil depth of 60 cm. The studied floodplain forest, for the past 30 years, has been plagued by hydrological extremes - floods and drought. Particularly drought threatens valuable floodplain forests. The last flood on 2013 was above soil surface only in the lowest parts of the studied site. 2.2. Soil CO2 efflux measurements SR (including vegetation cover d low growing plants covering the ground) was measured in a 150 m long transect on 30 positions planed in 5 m intervals using a closed dynamic chamber system (a portable infrared gas analyzer Li-8100 (Li-Cor, Inc., Lincoln, NE, USA) with a 20 cm survey chamber (Li-8100-103, Li-Cor, USA). At each measured position a PVC collar with 20 cm in diameter was installed into the soil at 5 cm depth. After the chamber closed, a period (dead band) of 15 s was set to allow steady mixing of the air in the chamber. During the following 60 s, CO2 concentration was measured repeatedly at 1s intervals and a linear fitting was used to calculate SR. Each measurement took about 3 min and each position was measured two to three times per day. Measurements were carried out once per month from May till November, 16 measurements campaigns were done during this period. Measurements were not realized in September due to malfunction of the pump of the CO2 analyzer. The 30 measured positions covered the most representative soil surface of the study site, 10 of the measured positions were without vegetation and the other 20 positions with temporary vegetation (Allium ursinum L., Galium odoratum (L.) Scop., between 10 and 70% coverage). The herbal understory is mainly during spring season. During each CO2 efflux measurement, soil temperature at 2 cm (TPD32 penetrate thermometer, Omega, Stamford, CT, USA) and soil moisture in the 0e6 cm profile (ThetaProbe ML2x, Delta-T
Devices, Cambridge, UK) were measured at the distance 5 cm outside the collar. Moreover, basic microclimatological parameters of the forest stand such as air temperature/humidity at 2 m height (EMS 33, EMS Brno, Czech Republic) and precipitation (rain gauge MetOne 386C, MetOne, USA) were recorded in 30 min intervals. Soil temperature (Pt100, Sensit, Czech Republic) at 2 cm depth was continuously recorded in 30 min intervals at four different plots close to the measured positions during the whole experiment period. 2.3. Data processing Two approaches were used to analyze the measured data. The first was using the Q10 parameter (the proportional change in respiration resulting from 10 C increase in temperature) from van't Hoff [15]. Q10 expresses the dependence of SR on soil temperature. In our analysis we determined a mean Q10 value on the basis of our measurements on all positions at each measurement campaign. The Q10 value was calculated according to equation (1) [16]:
Q10 ¼ e10,b
(1)
where b is the regression coefficient obtained from the natural logarithm of the relationship between CO2 efflux and soil temperature. Then, a reference value of R10 (CO2 efflux normalized to a temperature of 10 C) for each measured position was calculated as:
10T 10
R10 ¼ SR*Q10
(2)
where SR is the measured soil CO2 efflux (mmol CO2 m2 s1) at T soil temperature. Data calculation and equations (1) and (2) were done using Microsoft Office Excel 2007. Moreover, the measured SR data were related to the soil temperature and soil moisture measurements. An exponential regression for soil temperature and logarithmic regression for soil moisture were used in order to determine the dependence of SR on both environmental variables. The second approach was to use parameters Q10 and seasonal averaged R10 (R10ave) for estimating modelled soil CO2 efflux (RM) on the basis of daily mean soil temperature at 2 cm depth in the study site during the whole experiment period (from May till November) using the following equation [17]:
RM ¼
R10ave 10Ts
(3)
Q1010
The spatial heterogeneity of the soil surface CO2 efflux was expressed by the coefficient of variation CV e which is defined as the ratio of the standard deviation to the mean. Moreover, a geostatistical method was applied on SR data. Autocorrelation was studied by semivariograms [18]. Semivariance was estimated by the following expression
gðhÞ ¼
NðhÞ i 1 Xh Zðxi Þ Zðxi þ hÞ2 2NðhÞ i¼1
(4)
where g(h) is the semivariance of pairs of points separated by h distance; N(h) is the number of observations of each pair of points separated by h distance; Z(xi) and Z(xi þ h) are the values of Z variable in points xi and xi þ h. The following mathematical models were then fitted on created semivariograms:
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
37
a) Gaussian model
gðxÞ ¼ r þ s2 , 1 exp
x2 r2
(5)
b) linear model (linear model is special type of power models)
gðxÞ ¼ r þ s2 , 1 exp
3x2 r2
(6)
where r is nugget, s2 is structural variance and r is the range of spatial dependence. Models were selected on the basis of minimization loss function and the coefficient of determination (R2). Nugget effect r represents an intercept of model on the axis of semivariogram. Sill (r þ s2) is a value when the semivariogram curve stabilizes. The proportion of structural variance and sill indicate strong (<25%), moderate (25% < s2/(rþ s2) < 75%) or weak dependence at the sampling scale. The range determines the distance in which the difference of the semivariogram from the sill becomes negligible. The geostatistical analysis of the data was performed in R-project with geoR package create Ribeiro and Diggle [19]. All the measured and obtained data during the measurements campaigns is available at the FTP server: ftp.czechglobe.cz or zeus: zeus.czechglobe.cz upon request. 3. Results 3.1. Environmental conditions The mean air temperature at a height of 2 m in the study site during the experiment period (from May till November 2016) was 16.3 C. The highest air temperature was in summer at the end of June (25.4 C) and the lowest during the autumn at the end of November (4.1 C). The soil temperature and soil moisture during the measurements ranged from 3.1 C to 21.1 C and from 13% to 37%, respectively (Fig. 1). No precipitation events occurred during the measurements campaigns and no flood events occurred during the season 2016 at the study site. 3.2. Soil CO2 efflux, R10 and Q10 The measured SR during the experiment period (MayeNovember) ranged from 1.59 to 8.54 mmolCO2 m2 s1. The averaged SR for the season 2016 was 4.07 mmolCO2 m2 s1. The highest SR were measured during the summer period at the end of June, while the lowest were measured during the autumn in October and
Fig. 2. Monthly measured soil CO2 efflux at the investigated floodplain forest ecosystem during the growing season 2016. Each position (n ¼ 30) was measured 2e3 times per campaign. The boxes show a median and 25th and 75th percentiles, and error bars represent 10th and 90th percentiles.
November (Fig. 2). Summer months showed the highest variation in SR among measured positions (n ¼ 30) compared to late spring and autumn. The lowest variation in SR was observed in May. SR showed an exponential increase with rising soil temperature (Fig. 3a). Measured positions with no vegetation showed lower rates compared to positions with vegetation. Moreover, we observed that position closed to tree showed higher SR compared to those far of them. The determination coefficient (R2) for the relationship between SR and soil temperature for all measured positions was 0.79. While no relationship between SR and soil moisture was found, however, a decreasing trend of SR with increasing soil moisture was observed (Fig. 3b). We determined a mean coefficient of variation (CV) of 20% in spatial heterogeneity of SR during our study. The estimated mean Q10 value for all 30 positions was of 2.23 (±0.4) during the whole experiment period. The averaged seasonal R10 value for all 30 positions was of 2.33 (±0.60) mmolCO2 m2s1. The obtained parameters Q10 and R10 and the mean daily soil temperature at a depth of 2 cm were used to model the SR using Eq. (3) during the period from 1 May to 30 November 2016 (Fig. 4). The cumulative amount of carbon forest floor released from our experimental forest site calculated from the model was 7.4 (±1.1 SD) tC ha1 y1 for 2016.
Fig. 1. Soil temperature at 2 cm depth (Ts2cm, black line) and soil moisture at 5 cm depth (Smois5cm, grey line) were measured continuously in 30-min interval. Soil temperature at 2 cm depth (Tsm2cm, black dots) and soil moisture between 0 and 6 cm (Smois0-6 cm, grey dots) depth were recorded manually during measurements campaigns. All measurements were done during the experiment period 2016.
38
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
Fig. 3. Averaged soil CO2 efflux, soil temperature (a) and soil moisture (b) at the experimental forest per whole measurement period 2016. Soil temperature measured at 2 cm depth and soil moisture between 0 and 6 cm depth. n ¼ 30 positions.
Fig. 4. Averaged measured (dots) and modelled (line) soil CO2 efflux at experimental forest during the vegetation period 2016. n ¼ 30 positions.
3.3. Geostatistical analysis Concerning our geostatistical analysis, all sampling dates showed a positive nugget, which can be explained either by sampling error, shorter-range variability, or random and inherent variability (Fig. 5). The highest nugget was detected on the campaign of June when it exceeded 1.0. On the other hand, the lowest nugget was observed on the campaign of May. In the first four campaigns we can observe reaching the sill within the used scale of graphs, while on the campaign of October and November the values of sill were very high. The value of s2/(rþ s2) was always higher than 0.75 indicating spatial weak dependence at the sampling scale used. The variograms for each measurements campaign are shown in Fig. 5. The statistical models fitted and the relevant parameters are summarised in Table 1. 4. Discussion Our obtained values of SR at the floodplain forest are within the range reported in the literature from other types of forest [20e23]. From 818 studies reported in a global database of SR [23], only three studies were carried out in floodplain forest ecosystems. Nevertheless, there is still a scarcity in SR data in floodplain forest ecosystems, therefore deep comparisons are not possible. In a study at different forested watershed areas in USA, Gomez et al. [24] demonstrated that these kind of ecosystems are hot spots of greenhouse gases production (in CO2 eq) and had high CO2 fluxes relative to other ecosystems. In several studies, soil temperature
explains most of the SR [17,20,23,25], in our study at floodplain forest we also found soil temperature as a main factor influencing SR as confirmed by the high R2 of the relationship between SR and soil temperature. Simcic et al. [26] compared the respiratory rate (as CO2 efflux) of soils and sediments from five floodplain habitats (channel, gravel, islands, riparian forest and grassland) exposed to experimental inundation and at three incubation temperatures (4 C, 12 C and 20 C). They found that respiration rate increased with increasing temperature in samples from all habitats and substantially higher respiration rates in soils from riparian forest and grassland compared to other habitats. In our study, SR exhibited a seasonal pattern (Fig. 4) that was related to soil temperature. Similar seasonal pattern between SR and soil temperature was also found by Dilustro et al. [27] in south-eastern mixed pine forests (USA) and other studies [21,23,28]. In general, all land-cover ecosystems types showed a seasonality in SR, in most case following the annual temperature cycle [29]. Furthermore, in our study, we did not find any evidence of soil moisture influencing CO2 efflux. A possible reason for this may be connected with the fact that in our conditions soil moisture was not a limiting factor to SR during the measuring campaigns. However, the soil moisture during our measurement campaigns presented a very narrow range (between 22% and 27%), apart from the last campaign in autumn when soil moisture ranged between 37 and 38%. On the other hand, no flood event occurred in the studied forest during our investigation period. Dilustro et al. [27] pointed out that soil CO2 efflux was significantly related to soil moisture only in sandy sites when soil water content was above the wilting
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
39
Fig. 5. Semivariograms of the normalized CO2 efflux (R10) for individual measurement campaigns at the experimental forest in 2016. The horizontal line shows the sill of the semivariograms. Where letters stand for the date of measurement campaigns: a e 10 May, b e 22 June, c e 21 July, d e 30 August, e e 10 October and f e 23 November.
40
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
Table 1 The models used for the semivariograms and the fitted parameters of soil CO2 efflux for individual measurement campaigns at the experimental forest. Date
Model
Range (r)
Sill (rþ s2)
Structural variance (s2)
Nugget (r)
R2
s2/(rþ s2)
10 May 22 Jun 21 Jul 30 Aug 10 Oct sNov
exp gauss exp exp exp gauss
2.57 123.72 17.62 7.29 140174.9 3480.0
0.35 2.79 1.42 0.96 17.1 1644.4
0.26 1.48 0.91 0.48 16.9 1644.1
0.09 1.31 0.51 0.47 0.19 0.30
0.02 0.71 0.69 0.06 0.49 0.86
1.36 1.88 1.57 1.98 1.01 1.00
point threshold. A significant positive relationship between soil CO2 efflux and soil moisture was also found by Billing et al. [30] in boreal floodplain forest study. However, they reported that soil temperature was a more influential parameter on soil CO2 efflux than soil moisture. CO2 efflux from soil surface is usually highly heterogeneous. This corresponds with the heterogeneity (20e60%) observed in other studies for forest soils [23,31e33]. In our conditions, CV of spatial heterogeneity laid in the lower range, it means that spatial heterogeneity of SR was smaller compared to other types of forest ecosystems. Factors driving spatial heterogeneity of SR vary among different ecosystems. Stoyan et al. [18] pointed out that spatial distribution of soil respiration is defined by the overlapping distributions of substrates, soil physical conditions, soil organisms, and temperature and moisture conditions. In their study, they concluded that the main cause of soil heterogeneity at small scale (2 m2) are likely to be at micro scales controlled in part by plant root and plant residue patterns. In our study site some measured positions were without and with vegetation (herbal understory) that influences the total measured soil CO2 efflux. Positions with no vegetation showed lower rates compared to positions with vegetation. On the other hand, Tang and Baldocchi [34] reported that the mayor reason for the high spatial variation in soil CO2 efflux could be explained by the different contribution of functionally different components such as rhizosphere respiration (including root autotrophic respiration and associated mycorrhizae respiration) and microbial heterotrophic respiration in vegetation-covered land. However, we consider that the main factors influencing the spatial variation of SR in our study are vegetation composition of forest [31], our experimental forest is composed by different tree species, mainly by oak, ash, hornbeam, linden and herbal understory; and root biomass distribution [35]. In a study of root biomass distribution from soil cores (45 cm depth) carried at our experimental forest, was found that the highest root (0e2 mm in diameter) biomass distribution were found at the Ah and M1 horizons (unpublished data). The soil surface in the investigated site is flat and homogeneous and no soil skelet (stones) is present due to the alluvial origin of the soil. The main factors influencing the spatial variation of soil CO2 efflux are vegetation in the below canopy layer and root biomass distribution [20,35]. Apart from these factors, amount of litter, soil microbial biomass and soil pH are soil descriptors which are supposed to be related to the spatial variation of soil respiration [36]. Moreover, soil quality parameter as soil A layer, soil organic layer mass and soil texture has been identified to be significantly related to general pattern of soil CO2 efflux [27]. The parameter Q10 was estimated for temperature measured at a depth of 2 cm. Q10 is commonly used for normalizing measured CO2 efflux to a reference temperature (Eq. (2)) in order to investigate factors other than temperature [37], or it is used in carbon models to simulate soil or ecosystem CO2 fluxes (Eq. (3)) [23,25]. Our obtained Q10 value (2.23) laid between the reported range for different types of forest in other studies [23,38,39]. Several studies at field and in laboratory conditions have shown a number of biotic
and physical characteristics, which differ among forest stands, can influence soil CO2 efflux and it is strongly related to soil temperature as a classic Q10 relationship between biological activity and temperature [40]. Sim ci c et al. [26] in their study of experimental inundation and incubation temperatures, pointed out that the temperature sensitivity of inundated soils and sediments was higher of that previously measured in non-flooded terrestrial and floodplain soil [20,41], indicating that the thermal sensitivity of soil respiration most likely increases during inundation. Other studies reported different range of Q10 values of soil temperature at different depths ranging from 1 to 4 [23,29,42e44]. Nevertheless, it is necessary to point out that our Q10 value at the floodplain forest ecosystem was obtained during a season without flood event. Therefore, this Q10 should be taken as a value during low water availability at the study floodplain forest. Numerous models have been developed to express temperature sensitivity of soil CO2 efflux [23,29,40,45,46]. However, models make many simplifying assumptions and these assumptions can have a large impact on model predictions. Regardless of the simplicity of our model, it is built on data set that is underrepresent ecosystem, it follows the logic that the main abiotic driver that determines soil CO2 efflux is soil temperature. When comparing the SR model and measured data, our model shows a good agreement with our measured chamber data during the summer months (JuneeAugust) while underestimation of SR is observed during the spring and autumn months (Fig. 4). This observed underestimation was likely due to the lack of large diurnal variation of soil temperature during the measurement campaigns and it could lead to inaccurate Q10 calculation. Another possible reason could be the small variation of R10 during the year which might have been due to low activity of root growth or photosynthesis during spring and autumn. Darenova et al. [17], pointed out that SR measurements only during day time could underestimate determination of seasonal sums of CO2 released from soil. In our study, all the campaigns were done during the day time. Moreover, our model was based on a monthly measurement campaigns, leading to missing the effect of episodic rain events that sometimes results huge SR pulse through stimulation of microorganisms [47]. Nevertheless, the used model presents an overview of the soil CO2 efflux dynamic over the investigated growing season. Concerning our geostatistical analysis on our measured soil CO2 efflux positions (Fig. 5), it showed that during the campaign in May there was no high differences in soil CO2 efflux rates between sampling positions that were close together and those that were far apart. This may be connected to the hypothesis that when the favourable soil conditions decrease, the variability of soil CO2 efflux was reduced. Stoyan et al. [18] pointed out that the more structured variability observed in the deeper soil supports the hypothesis that the heterogeneity is due either to the non-uniform addition of organic material to the surface or the variations in moisture content. The soil moisture and organic material varied within the soil, resulting in smoother shifts between high and low soil respiration areas. The summer months (June, July and August, Fig. 2) showed
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
higher differences in soil CO2 efflux rates among sampling positions regarding distance compared to late spring (May). This could be explained by the activity of decomposers working on leaf litter from previous fall in a moisture environment and higher soil temperature in summer resulted in soil activity in our floodplain forest that was higher in summer than in late spring. November showed low differences in soil CO2 efflux rates between sampling positions that were close together but higher differences in rates for those that were far apart. We consider that a possible explanation of this could be connected with the vicinity of trees to individual measuring position, we observed that position closed to tree showed higher soil CO2 efflux compared to those far of them. The same situation was regarding the influence of herbal understory in positions where vegetation was presented compared to positions without vegetation. In a study of spatial variability of soil CO2 efflux, Brito et al. [48], pointed out that depending on the topographic position of the experiment, soil CO2 emission could present differences in range values, which represents the changes in scale of each property. However, it is worth to point out that geostatistical analyses provide powerful analytical tools to capture the horizontal variability of a property and have received increasing interest by soil biologists in recent years. In its simplest form geostatistics define the degree of autocorrelation among the measured data points and interpolates values between measured points based on the degree of autocorrelation encountered [49], but it is tricky to define the reason of this autocorrelation. As was stated before, the present pilot study was designed to address the general need for basic information about soil CO2 efflux heterogeneity in floodplain forest and to gain information about the factors influencing this heterogeneity. Our results represent a novel information about soil CO2 efflux in this kind of ecosystem that has been missing concerning carbon fluxes studies. The magnitude of soil CO2 effluxes and its heterogeneity in floodplain forest is mainly caused by the rich biodiversity of this kind of ecosystem and it is significant compared to upland forest ecosystems. Our results also provide a conceptual understanding of soil CO2 efflux dynamics in floodplain forests and the main environmental factors influencing these dynamics in order to improve the knowledge of carbon forest floor emission in this kind of terrestrial ecosystem. Our pilot study also demonstrated the importance of floodplain forest in terms of carbon issues and the need for longer and deeper investigations. 5. Conclusions Our study provides evidence of the influence of soil temperature, as a main factor, on SR and the seasonal variability of SR at floodplain forest ecosystem. In our floodplain forest ecosystem, soil moisture was not a limiting factor of SR. The spatial heterogeneity of SR at our investigated forest was 20%. The cumulative amount of carbon forest floor released from our experimental forest site was 7.4 (±1.1) tC ha1 y1 for the 2016. The majority of carbon cycling studies including modelling are mainly carried out in typical forest ecosystem and do not take into account the specific conditions of floodplain forest ecosystem. Even though, our study was carried out in a season without floods, the results of our study help to improve our knowledge and models concerning carbon issues in this kind of ecosystem during period of low water availability. Such data are important for evaluating the effect of climate change or other possible influences on carbon cycling and carbon sequestration in lowland floodplain forest ecosystems. Acknowledgements This work was supported by the Ministry of Education, Youth
41
and Sports of the Czech Republic within the National Sustainability Program I (NPU I), grant number LO1415 and grant number LD15040. We would like to thank to Mr. Ryan McGloin for the linguistic revision. References camps, The ecology of interfaces: riparian zones, Annu. Rev. [1] R.J. Naiman, H. De Ecol. Syst. 28 (1997) 621e658. [2] K. Tockner, J.A. Stanford, Riverine flood plains: present state and future trends, Environ. Conserv. 29 (2002) 308e330. [3] K. Tockner, F. Malard, J.V. Ward, An extension of the flood pulse concept, Hydrol. Process. 14 (2000) 2861e2883. [4] K. Tockner, S.E. Bunn, C. Gordon, R.J. Naiman, G.P. Quinn, J.A. Standord, Flood plains: critically threatened ecosystems, in: N.V.C. Polunin (Ed.), Aquatic Ecosystems: Trends and Global Prospects, Cambridge University Press, Cambridge, 2008, pp. 45e61. [5] D.P. Turner, S.V. Ollinger, J.S. Kimball, Integrating remote sensing and ecosystem process models for landscape- to regional-scale analysis of the carbon cycle, BioScience 54 (2004) 573e584. [6] J. Cerm ak, Leaf distribution in large trees and stands of the floodplain forest in southern Moravia, Tree Physiol. 18 (1998) 727e737. [7] M. Kazda, J. Salzer, I. Reiter, Photosynthetic capacity in relation to nitrogen in the canopy of Quercus robur, Fraxinus angustifolia and Tilia cordata flood plain forest, Tree Physiol. 20 (2000) 1029e1037. [8] J. Glaser, M. Wulf, Effects of water regime and habitat continuity on the plant species composition of floodplain forests, J. Veg. Sci. 20 (2009) 37e48. [9] T.T. Kozlowski, Responses of woody plants to flooding and salinity, Tree Physiol. 1 (1997) 1e29. [10] M. Southwell, M. Thoms, Patterns of nutrient concentrations across multiple floodplain surface in a large dryland river system, Geogr. Res. 49 (2011) 431e443. [11] T.J. Battin, S. Luyssaert, L.A. Kaplan, A.K. Aufdenkampe, A. Richter, L.J. Tranvik, The boundless carbon cycle, Nat. Geosci. 2 (2009) 598e600. [12] D. Baldwin, G.N. Rees, J.S. Wilson, M.J. Colloff, K.L. Whitwoth, T.L. Pitman, T.A. Wallace, Provisioning of bioavailable carbon between the wet and dry phases in a semi-arid floodplain, Oecologia 172 (2013) 539e550. [13] M. Reichstein, M. Bahn, P. Ciais, D. Frank, M.D. Mahecha, S.I. Seneviratne, J. Zscheischler, C. Beer, N. Buchmann, D.C. Frank, D. Papale, A. Rammig, P. Smith, K. Thonicke, M. van der Velde, S. Vicca, A. Walz, M. Wattenbach, Climate extremes and the carbon cycle, Nature (2013) 287e295. [14] I. Rieger, F. Lang, B. Kleinschmit, I. Kowarik, A. Cierjacks, Fine root and aboveground carbon stocks in riparian forests: the roles of diking and environmental gradients, Plant Soil 370 (2013) 497e509. [15] J.H. Van’ t Hoff, Lectures on theoretical and physical chemistry, in: R.A. Lehfeldt (Ed.), Part I Chemical Dynamics, Edward Arnold, London, 1898, pp. 224e229. [16] S. Linder, E. Troeng, The seasonal variation in stem and coarse root respiration of a 20-year-old Scots pine (Pinus sylvestris L.), in: W. Tranquillini (Ed.), €ume, Mitt. Forstl., vol. 142, Bundesversuchsanst Wien, Dickenwachstum der Ba 1981, pp. pp125e140. [17] E. Darenova, M. Pavelka, M. Acosta, Diurnal deviations in the relationship between CO2 efflux and temperature: a case study, Catena 123 (2014) 263e269. € hm, G.P. Robertson, E.A. Paul, Spatial heterogeneity [18] H. Stoyan, H. De-Polli, S. Bo of soil respiration and related properties at the plant scale, Plant soil 222 (2000) 203e214. [19] J.R. Ribeiro, P.J. Diggle, geoR: a package for geostatistical analysis, R-News 1 (2) (2001). ISSN 1609e3631. [20] N. Buchmann, Biotic and abiotic factors controlling soil respiration rates in Picea abies stands, Soil Biol. Biochem. 32 (2000) 1625e1635. [21] D. Gaumont-Guay, T.A. Black, H. Mccaughey, A.G. Barr, P. Krishnan, R.S. Jassal, Z. Nesic, Soil CO2 efflux in contrasting boreal deciduous and coniferous stands and its contribution to the ecosystem carbon balance, Glob. Change Biol. 15 (2009) 1302e1319. [22] A. Shvaleva, R. Lobo-do-Vale, C. Cruz, S. Castaldi, A.P. Rosa, M.M. Chaves, J.S. Pereira, Soil-atmosphere greenhouse gases (CO2, CH4 and N2O) exchange in evergreen oak woodland in southern Portugal, Plant Soil Environ. 57 (10) (2011) 471e477. [23] B. Bond-Lamberty, A. Thomson, A global database of soil respiration data, Biogeosciences 7.6 (2010) 1915e1926. [24] J. Gomez, P. Vidon, J. Gross, C. Beier, J. Caputo, M. Mitchell, Estimating greenhouse gas emissions at the soileatmosphere interface in forested watersheds of the US Northeast, Environ. Monit. Asses 188 (5) (2016) 1e16. [25] M. Khomik, M.A. Arain, J.H. Mccaughey, Temporal and spatial variability of soil respiration in a boreal mixewood forest, Agric. For. Meteorol. 140 (2006) 244e256. [26] T. Sim ci c, H. Mori, C. Hossli, C.T. Robinson, M. Doering, The response in floodplain respiration of an alpine river to experimental inundation under different temperature regimes, Hydrol. Process. 29 (2015) 5438e5450. [27] J.J. Dilustro, B. Collins, L. Duncan, C. Crawford, Moisture and soil texture effects on soil CO2 efflux components in southeastern mixed pine forests, For. Ecol. Manag. 204 (1) (2005) 87e97.
42
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
[28] J. Chen, Q. Wang, M. Li, F. Liu, W. Li, L. Yin, Effects of deer disturbance on soil respiration in a subtropical floodplain wetland of the Yangtze River, Eur. J. Soil Biol. 56 (2013) 65e71. [29] J.W. Raich, C.S. Potter, D. Bhagawati, Interannual variability in global soil respiration, 1980e94, Glob. Change Biol. 8 (8) (2002) 800e812. [30] S.A. Billings, D.D. Richter, J. Yarie, Soil carbon dioxide fluxes and profile concentrations in two boreal forests, Can. J. For. Res. 28 (12) (1998) 1773e1783. [31] B.E. Law, F.M. Kelliher, D.D. Baldocchi, P.M. Anthoni, J. Irvine, D. Moore, S. Van Tuyl, Spatial and temporal variation in respiration in a young ponderosa pine forest during a summer drought, Agric. For. Meteorol. 10 (2001) 27e43. [32] Y. Kosugi, T. Mitani, M. Ltoh, S. Noguchi, M. Tani, N. Matsuo, S. Takanashi, S. Ohkubo, A.R. Nik, Spatial and temporal variation in soil respiration in a Southeast Asian tropical rainforest, Agric. For. Meteorol. 147 (2007) 35e47. [33] E. Darenova, M. Pavelka, L. Macalkova, Spatial heterogeneity of CO2 efflux and optimization of the number of measurement positions, Eur. J. Soil Biol. 75 (2016) 123e134. [34] J. Tang, D.D. Baldocchi, Spatialetemporal variation in soil respiration in an oakegrass savanna ecosystem in California and its partitioning into autotrophic and heterotrophic components, Biogeochemistry 73 (1) (2005) 183e207. [35] F.E. Moyano, L.K. Werner, C. Rebmann, Soil respiration fluxes in relation to photosynthetic activity in broad-leaf and needle-leaf forest stands, Agric. For. Meteorol. 148 (2008) 135e143. [36] P.J. Hanson, N.T. Edwards, C.T. Garten, J.A. Andrews, Separating root and soil microbial contributions to soil respiration: a review of methods and observations, Biogeochemistry 48 (2000) 115e146. [37] R.S. Jassal, T.A. Black, M.D. Novak, D. Gaumont-Guay, Z. Nesic, Effect of soil water stress on soil respiration and its temperature sensitivity in an 18-yearold temperate Douglas-fir stand, Glob. Change Biol. 14 (2008) 1305e1318. [38] W. Borken, Y.J. Xu, E.A. Davidson, F. Beese, Site and temporal variation of soil respiration in European beech, Norway spruce, and Scots pine forests, Glob. Change Biol. 8 (2002) 1205e1216. [39] G. Saiz, K. Black, B. Reidy, S. Lopez, E.P. Farrell, Assessment of soil CO2 efflux and its components using a process-based model in a young temperate forest site, Geoderma 139 (2007) 79e89.
[40] J. Lloyd, J.A. Taylor, On the temperature dependence of soil respiration, Funct. Ecol. 8 (1994) 315e323. [41] E. Samaritani, J. Shrestha, B. Fournier, E. Frossard, F. Gillet, C. Guenat, P.A. Niklaus, K. Tockner, E.A.D. Mitchell, Heterogeneity of soil carbon pools and fluxes in a channelized and a restored floodplain section (Thur River, Switzerland), Hydrol. Earth Syst. Sci. 15 (2011) 1757e1769. [42] E. Davidson, I. Jansens, Y. Luo, On the variability of respiration in terrestrial ecosystems: moving beyond Q10, Glob. Change Biol. 12 (2) (2006) 154e164. [43] P.J. Hanson, N.T. Edwards, C.T. Garten, J.A. Andrews, 2000. Separating root and soil microbial contributions to soil respiration: a review of methods and observations, Biogeochemistry 48 (2000) 115e146. [44] M. Acosta, M. Pavelka, L. Montagnani, W. Kutsch, A. Lindroth, R. Juszczak, D. Janous, Soil surface CO2 efflux measurements in Norway spruce forests: comparison between four different sites across Europedfrom boreal to alpine forest, Geoderma 192 (2013) 295e303. [45] M. Reichstein, A. Rey, A. Freibauer, J. Tenhunen, R. Valentini, J. Banza, P. Casals, Y. Cheng, J.M. Grünzweig, R. Israel, J. Irvine, R. Joffre, B.E. Law, D. Loustau, F. Miglietta, W. Oechel, J.M. Ourcival, J.S. Pereira, A. Peressotti, F. Ponti, Y. Qi, S. Rambal, M. Rayment, J. Romanya, F. Rossi, V. Tedeschi, G. Tirone, M. Xu, D. Yakir, Modeling temporal and large-scale spatial variability of soil respiration from soil water availability, temperature and vegetation productivity indices, Glob. Biogeochem. Cycles 17 (2003) 1104. [46] M. Tuomi, P. Vanhala, K. Karhu, H. Fritze, J. Liski, Heterotrophic soil respiration -comparison of different models describing its temperature dependence, Ecol. Model. 211 (2008) 182e190. [47] X. Liu, S. Wan, B. Su, D. Hui, Y. Luo, Response of soil CO2 efflux to water manipulation in a tallgrass prairie ecosystem, Plant Soil 240 (2) (2002) 213e223. [48] L.D.F. Brito, J. Marques Júnior, G.T. Pereira, N. La Scala Junior, Spatial variability of soil CO2 emission in different topographic positions, Bragantia 69 (2010) 19e27. [49] G.P. Robertson, Geostatistics in ecology: interpolating with known variance, Ecology 68 (3) (1987) 744e748.
Contents lists available at ScienceDirect
European Journal of Soil Biology journal homepage: http://www.elsevier.com/locate/ejsobi
Soil carbon dioxide fluxes in a mixed floodplain forest in the Czech Republic Manuel Acosta*, Eva Darenova, Jirí Dusek, Marian Pavelka Global Change Research Institute, Czech Academy of Sciences, B elidla 4a, Brno 603 00, Czech Republic
a r t i c l e i n f o
a b s t r a c t
Article history: Received 19 April 2017 Received in revised form 2 August 2017 Accepted 15 August 2017
Floodplain forests belong among the most productive, dynamic and diverse ecosystems on Earth. Only few studies have focused on the carbon dioxide fluxes of these ecosystems. Therefore, this study investigated the spatial heterogeneity in soil CO2 efflux in a floodplain forest located in the southeast of the Czech Republic. The study also examined which environmental parameters influence soil CO2 efflux. Moreover, using these obtained measurements a soil CO2 efflux model was applied. To achieve the aims of this study, soil CO2 efflux on 30 positions in 16 campaigns was measured from May to November during the growing season 2016. The efflux during the experiment period ranged from 1.59 to 8.54 mmolCO2 m2 s1. The highest soil CO2 effluxes were observed during the summer period while the lowest values were measured during the autumn. A strong relationship between soil CO2 efflux and soil temperature was found (R2 ¼ 0.79). The estimated mean Q10 for the whole 30 positions was of 2.23. We determined that the spatial heterogeneity of soil CO2 efflux was 20% during our study. The cumulative amount of carbon forest floor released from our experimental forest site calculated from our model was 7.4 (±1.1) tC ha1 y1 for 2016. Such data are important for developing our knowledge and understanding about carbon dynamics and to improve carbon models for these ecosystems types. © 2017 Elsevier Masson SAS. All rights reserved.
Handling editor: Yakov Kuzyakov Keywords: CO2 Soil temperature Soil moisture Spatial heterogeneity Q10
1. Introduction Floodplain forests belong among the most productive, dynamic and diverse ecosystems on Earth [1,2]. Due to fluctuations in river flow and subsequent alternation between flooding and drying, floodplain ecosystems are in a state of dynamic equilibrium [3]. Within the floodplains, floods and geomorphic processes interact to create a shifting mosaic of habitat patches [1] where the vegetation patterns are driven by disturbance intensity, inundation regime and by geological and soil properties. With the exception of the most dynamic and frequently disturbed areas, the majority of floodplain vegetation consists of forests of various age and tree species composition. These floodplain forests are highly productive and provide a wide range of ecosystem services such as high biodiversity, flood water retention, a nutrient sink, groundwater recharge, carbon sequestration, timber production, recreational facilities and aesthetic value [4].
* Corresponding author. E-mail addresses: [email protected] (M. Acosta), darenova.e@czechglobe. cz (E. Darenova), [email protected] (J. Dusek), [email protected] (M. Pavelka). http://dx.doi.org/10.1016/j.ejsobi.2017.08.006 1164-5563/© 2017 Elsevier Masson SAS. All rights reserved.
Many studies have investigated floodplain forests from different ecological points of view such as plant composition [5], tree physiology and morphology [6,7], flooding regime and water table variations [8], and soil acidity and nutrient concentrations [9,10]. But few studies have focused on the effect of flood dynamics on the greenhouse gas budgets and carbon storage of riparian ecosystems [11e13]. In large-scale research projects of the last 15 years (CarboEurope IP, Carbomont, Nitroeurope etc.) floodplain forests have been ignored; mostly because they are ecosystems that are not representative of a wider region. Due to their comparatively small areal extension, floodplain forests are rarely captured in monitoring systems along a regular pattern e.g. along transects and grids. Nevertheless, floodplain forests may play an important role in regional carbon cycling and total greenhouse gases (GHG) dynamics. Floodplain forests in temperate areas are highly productive and store carbon (C) in large quantities when they are compared with upland forests. A detailed assessment also showed that fine root biomass as well as above ground plant biomass significantly varied between flooded and diked areas [14]. However, a move from stock data towards a conceptual understanding of ecosystem C dynamics is necessary to assess the role of floodplain forests in C
36
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
sequestration. The reason for the current lack in detailed C flux data is likely related to obstacles linked to the intensive dynamics of the system, such as additional C import and export by flooding, but might also be attributed to the impracticalities of in situ research in periodically flooded areas. The present pilot study was designed to address the general need for basic information about soil CO2 efflux heterogeneity and specifically to gain information about the factors influencing soil CO2 efflux heterogeneity in floodplain forest. Therefore, the aims of our study were i) to quantify soil CO2 efflux (SR) and to estimate the spatial heterogeneity in SR during the growing season 2016 in a floodplain forest located in the southeast of the Czech Republic, ii) to determine the influence of environmental parameters as soil temperature and soil moisture on SR and iii) to create a model of SR based on measured data during the vegetation season 2016. 2. Material and methods 2.1. Site description The experiment site is situated 6.5 km to the north of the confluence of Morava and Dyje rivers (48 40.090 N, 16 56.780 E). It is formed of an alluvial plain. Long-term average annual precipitation is around 550 mm and mean annual temperature is 9.3 C. The experimental site is composed by typical hardwood species. The main tree species composition is English oak (Quercus robur L.), Narrow-leaved ash (Fraxinus angustifolia Vahl), hornbeam (Carpinus betulus L.) and linden (Tilia cordata Mill.). The stand age is 110 years and height 27 m. The herbal layer and understory of the forest site is characterized by Allium ursinum L., Anemone ranunculoides L., Lathyrus vernus (L.) Bernh., Galium odoratum (L.) Scop., Carex sylvatica Huds. and Acer campestre L. The predominating soil types are Eutric Humic Fluvisol, Haplic Fluvisol, and Eutric Fluvisol (according FAO 2014 Classification) with minimal soil depth of 60 cm. The studied floodplain forest, for the past 30 years, has been plagued by hydrological extremes - floods and drought. Particularly drought threatens valuable floodplain forests. The last flood on 2013 was above soil surface only in the lowest parts of the studied site. 2.2. Soil CO2 efflux measurements SR (including vegetation cover d low growing plants covering the ground) was measured in a 150 m long transect on 30 positions planed in 5 m intervals using a closed dynamic chamber system (a portable infrared gas analyzer Li-8100 (Li-Cor, Inc., Lincoln, NE, USA) with a 20 cm survey chamber (Li-8100-103, Li-Cor, USA). At each measured position a PVC collar with 20 cm in diameter was installed into the soil at 5 cm depth. After the chamber closed, a period (dead band) of 15 s was set to allow steady mixing of the air in the chamber. During the following 60 s, CO2 concentration was measured repeatedly at 1s intervals and a linear fitting was used to calculate SR. Each measurement took about 3 min and each position was measured two to three times per day. Measurements were carried out once per month from May till November, 16 measurements campaigns were done during this period. Measurements were not realized in September due to malfunction of the pump of the CO2 analyzer. The 30 measured positions covered the most representative soil surface of the study site, 10 of the measured positions were without vegetation and the other 20 positions with temporary vegetation (Allium ursinum L., Galium odoratum (L.) Scop., between 10 and 70% coverage). The herbal understory is mainly during spring season. During each CO2 efflux measurement, soil temperature at 2 cm (TPD32 penetrate thermometer, Omega, Stamford, CT, USA) and soil moisture in the 0e6 cm profile (ThetaProbe ML2x, Delta-T
Devices, Cambridge, UK) were measured at the distance 5 cm outside the collar. Moreover, basic microclimatological parameters of the forest stand such as air temperature/humidity at 2 m height (EMS 33, EMS Brno, Czech Republic) and precipitation (rain gauge MetOne 386C, MetOne, USA) were recorded in 30 min intervals. Soil temperature (Pt100, Sensit, Czech Republic) at 2 cm depth was continuously recorded in 30 min intervals at four different plots close to the measured positions during the whole experiment period. 2.3. Data processing Two approaches were used to analyze the measured data. The first was using the Q10 parameter (the proportional change in respiration resulting from 10 C increase in temperature) from van't Hoff [15]. Q10 expresses the dependence of SR on soil temperature. In our analysis we determined a mean Q10 value on the basis of our measurements on all positions at each measurement campaign. The Q10 value was calculated according to equation (1) [16]:
Q10 ¼ e10,b
(1)
where b is the regression coefficient obtained from the natural logarithm of the relationship between CO2 efflux and soil temperature. Then, a reference value of R10 (CO2 efflux normalized to a temperature of 10 C) for each measured position was calculated as:
10T 10
R10 ¼ SR*Q10
(2)
where SR is the measured soil CO2 efflux (mmol CO2 m2 s1) at T soil temperature. Data calculation and equations (1) and (2) were done using Microsoft Office Excel 2007. Moreover, the measured SR data were related to the soil temperature and soil moisture measurements. An exponential regression for soil temperature and logarithmic regression for soil moisture were used in order to determine the dependence of SR on both environmental variables. The second approach was to use parameters Q10 and seasonal averaged R10 (R10ave) for estimating modelled soil CO2 efflux (RM) on the basis of daily mean soil temperature at 2 cm depth in the study site during the whole experiment period (from May till November) using the following equation [17]:
RM ¼
R10ave 10Ts
(3)
Q1010
The spatial heterogeneity of the soil surface CO2 efflux was expressed by the coefficient of variation CV e which is defined as the ratio of the standard deviation to the mean. Moreover, a geostatistical method was applied on SR data. Autocorrelation was studied by semivariograms [18]. Semivariance was estimated by the following expression
gðhÞ ¼
NðhÞ i 1 Xh Zðxi Þ Zðxi þ hÞ2 2NðhÞ i¼1
(4)
where g(h) is the semivariance of pairs of points separated by h distance; N(h) is the number of observations of each pair of points separated by h distance; Z(xi) and Z(xi þ h) are the values of Z variable in points xi and xi þ h. The following mathematical models were then fitted on created semivariograms:
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
37
a) Gaussian model
gðxÞ ¼ r þ s2 , 1 exp
x2 r2
(5)
b) linear model (linear model is special type of power models)
gðxÞ ¼ r þ s2 , 1 exp
3x2 r2
(6)
where r is nugget, s2 is structural variance and r is the range of spatial dependence. Models were selected on the basis of minimization loss function and the coefficient of determination (R2). Nugget effect r represents an intercept of model on the axis of semivariogram. Sill (r þ s2) is a value when the semivariogram curve stabilizes. The proportion of structural variance and sill indicate strong (<25%), moderate (25% < s2/(rþ s2) < 75%) or weak dependence at the sampling scale. The range determines the distance in which the difference of the semivariogram from the sill becomes negligible. The geostatistical analysis of the data was performed in R-project with geoR package create Ribeiro and Diggle [19]. All the measured and obtained data during the measurements campaigns is available at the FTP server: ftp.czechglobe.cz or zeus: zeus.czechglobe.cz upon request. 3. Results 3.1. Environmental conditions The mean air temperature at a height of 2 m in the study site during the experiment period (from May till November 2016) was 16.3 C. The highest air temperature was in summer at the end of June (25.4 C) and the lowest during the autumn at the end of November (4.1 C). The soil temperature and soil moisture during the measurements ranged from 3.1 C to 21.1 C and from 13% to 37%, respectively (Fig. 1). No precipitation events occurred during the measurements campaigns and no flood events occurred during the season 2016 at the study site. 3.2. Soil CO2 efflux, R10 and Q10 The measured SR during the experiment period (MayeNovember) ranged from 1.59 to 8.54 mmolCO2 m2 s1. The averaged SR for the season 2016 was 4.07 mmolCO2 m2 s1. The highest SR were measured during the summer period at the end of June, while the lowest were measured during the autumn in October and
Fig. 2. Monthly measured soil CO2 efflux at the investigated floodplain forest ecosystem during the growing season 2016. Each position (n ¼ 30) was measured 2e3 times per campaign. The boxes show a median and 25th and 75th percentiles, and error bars represent 10th and 90th percentiles.
November (Fig. 2). Summer months showed the highest variation in SR among measured positions (n ¼ 30) compared to late spring and autumn. The lowest variation in SR was observed in May. SR showed an exponential increase with rising soil temperature (Fig. 3a). Measured positions with no vegetation showed lower rates compared to positions with vegetation. Moreover, we observed that position closed to tree showed higher SR compared to those far of them. The determination coefficient (R2) for the relationship between SR and soil temperature for all measured positions was 0.79. While no relationship between SR and soil moisture was found, however, a decreasing trend of SR with increasing soil moisture was observed (Fig. 3b). We determined a mean coefficient of variation (CV) of 20% in spatial heterogeneity of SR during our study. The estimated mean Q10 value for all 30 positions was of 2.23 (±0.4) during the whole experiment period. The averaged seasonal R10 value for all 30 positions was of 2.33 (±0.60) mmolCO2 m2s1. The obtained parameters Q10 and R10 and the mean daily soil temperature at a depth of 2 cm were used to model the SR using Eq. (3) during the period from 1 May to 30 November 2016 (Fig. 4). The cumulative amount of carbon forest floor released from our experimental forest site calculated from the model was 7.4 (±1.1 SD) tC ha1 y1 for 2016.
Fig. 1. Soil temperature at 2 cm depth (Ts2cm, black line) and soil moisture at 5 cm depth (Smois5cm, grey line) were measured continuously in 30-min interval. Soil temperature at 2 cm depth (Tsm2cm, black dots) and soil moisture between 0 and 6 cm (Smois0-6 cm, grey dots) depth were recorded manually during measurements campaigns. All measurements were done during the experiment period 2016.
38
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
Fig. 3. Averaged soil CO2 efflux, soil temperature (a) and soil moisture (b) at the experimental forest per whole measurement period 2016. Soil temperature measured at 2 cm depth and soil moisture between 0 and 6 cm depth. n ¼ 30 positions.
Fig. 4. Averaged measured (dots) and modelled (line) soil CO2 efflux at experimental forest during the vegetation period 2016. n ¼ 30 positions.
3.3. Geostatistical analysis Concerning our geostatistical analysis, all sampling dates showed a positive nugget, which can be explained either by sampling error, shorter-range variability, or random and inherent variability (Fig. 5). The highest nugget was detected on the campaign of June when it exceeded 1.0. On the other hand, the lowest nugget was observed on the campaign of May. In the first four campaigns we can observe reaching the sill within the used scale of graphs, while on the campaign of October and November the values of sill were very high. The value of s2/(rþ s2) was always higher than 0.75 indicating spatial weak dependence at the sampling scale used. The variograms for each measurements campaign are shown in Fig. 5. The statistical models fitted and the relevant parameters are summarised in Table 1. 4. Discussion Our obtained values of SR at the floodplain forest are within the range reported in the literature from other types of forest [20e23]. From 818 studies reported in a global database of SR [23], only three studies were carried out in floodplain forest ecosystems. Nevertheless, there is still a scarcity in SR data in floodplain forest ecosystems, therefore deep comparisons are not possible. In a study at different forested watershed areas in USA, Gomez et al. [24] demonstrated that these kind of ecosystems are hot spots of greenhouse gases production (in CO2 eq) and had high CO2 fluxes relative to other ecosystems. In several studies, soil temperature
explains most of the SR [17,20,23,25], in our study at floodplain forest we also found soil temperature as a main factor influencing SR as confirmed by the high R2 of the relationship between SR and soil temperature. Simcic et al. [26] compared the respiratory rate (as CO2 efflux) of soils and sediments from five floodplain habitats (channel, gravel, islands, riparian forest and grassland) exposed to experimental inundation and at three incubation temperatures (4 C, 12 C and 20 C). They found that respiration rate increased with increasing temperature in samples from all habitats and substantially higher respiration rates in soils from riparian forest and grassland compared to other habitats. In our study, SR exhibited a seasonal pattern (Fig. 4) that was related to soil temperature. Similar seasonal pattern between SR and soil temperature was also found by Dilustro et al. [27] in south-eastern mixed pine forests (USA) and other studies [21,23,28]. In general, all land-cover ecosystems types showed a seasonality in SR, in most case following the annual temperature cycle [29]. Furthermore, in our study, we did not find any evidence of soil moisture influencing CO2 efflux. A possible reason for this may be connected with the fact that in our conditions soil moisture was not a limiting factor to SR during the measuring campaigns. However, the soil moisture during our measurement campaigns presented a very narrow range (between 22% and 27%), apart from the last campaign in autumn when soil moisture ranged between 37 and 38%. On the other hand, no flood event occurred in the studied forest during our investigation period. Dilustro et al. [27] pointed out that soil CO2 efflux was significantly related to soil moisture only in sandy sites when soil water content was above the wilting
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
39
Fig. 5. Semivariograms of the normalized CO2 efflux (R10) for individual measurement campaigns at the experimental forest in 2016. The horizontal line shows the sill of the semivariograms. Where letters stand for the date of measurement campaigns: a e 10 May, b e 22 June, c e 21 July, d e 30 August, e e 10 October and f e 23 November.
40
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
Table 1 The models used for the semivariograms and the fitted parameters of soil CO2 efflux for individual measurement campaigns at the experimental forest. Date
Model
Range (r)
Sill (rþ s2)
Structural variance (s2)
Nugget (r)
R2
s2/(rþ s2)
10 May 22 Jun 21 Jul 30 Aug 10 Oct sNov
exp gauss exp exp exp gauss
2.57 123.72 17.62 7.29 140174.9 3480.0
0.35 2.79 1.42 0.96 17.1 1644.4
0.26 1.48 0.91 0.48 16.9 1644.1
0.09 1.31 0.51 0.47 0.19 0.30
0.02 0.71 0.69 0.06 0.49 0.86
1.36 1.88 1.57 1.98 1.01 1.00
point threshold. A significant positive relationship between soil CO2 efflux and soil moisture was also found by Billing et al. [30] in boreal floodplain forest study. However, they reported that soil temperature was a more influential parameter on soil CO2 efflux than soil moisture. CO2 efflux from soil surface is usually highly heterogeneous. This corresponds with the heterogeneity (20e60%) observed in other studies for forest soils [23,31e33]. In our conditions, CV of spatial heterogeneity laid in the lower range, it means that spatial heterogeneity of SR was smaller compared to other types of forest ecosystems. Factors driving spatial heterogeneity of SR vary among different ecosystems. Stoyan et al. [18] pointed out that spatial distribution of soil respiration is defined by the overlapping distributions of substrates, soil physical conditions, soil organisms, and temperature and moisture conditions. In their study, they concluded that the main cause of soil heterogeneity at small scale (2 m2) are likely to be at micro scales controlled in part by plant root and plant residue patterns. In our study site some measured positions were without and with vegetation (herbal understory) that influences the total measured soil CO2 efflux. Positions with no vegetation showed lower rates compared to positions with vegetation. On the other hand, Tang and Baldocchi [34] reported that the mayor reason for the high spatial variation in soil CO2 efflux could be explained by the different contribution of functionally different components such as rhizosphere respiration (including root autotrophic respiration and associated mycorrhizae respiration) and microbial heterotrophic respiration in vegetation-covered land. However, we consider that the main factors influencing the spatial variation of SR in our study are vegetation composition of forest [31], our experimental forest is composed by different tree species, mainly by oak, ash, hornbeam, linden and herbal understory; and root biomass distribution [35]. In a study of root biomass distribution from soil cores (45 cm depth) carried at our experimental forest, was found that the highest root (0e2 mm in diameter) biomass distribution were found at the Ah and M1 horizons (unpublished data). The soil surface in the investigated site is flat and homogeneous and no soil skelet (stones) is present due to the alluvial origin of the soil. The main factors influencing the spatial variation of soil CO2 efflux are vegetation in the below canopy layer and root biomass distribution [20,35]. Apart from these factors, amount of litter, soil microbial biomass and soil pH are soil descriptors which are supposed to be related to the spatial variation of soil respiration [36]. Moreover, soil quality parameter as soil A layer, soil organic layer mass and soil texture has been identified to be significantly related to general pattern of soil CO2 efflux [27]. The parameter Q10 was estimated for temperature measured at a depth of 2 cm. Q10 is commonly used for normalizing measured CO2 efflux to a reference temperature (Eq. (2)) in order to investigate factors other than temperature [37], or it is used in carbon models to simulate soil or ecosystem CO2 fluxes (Eq. (3)) [23,25]. Our obtained Q10 value (2.23) laid between the reported range for different types of forest in other studies [23,38,39]. Several studies at field and in laboratory conditions have shown a number of biotic
and physical characteristics, which differ among forest stands, can influence soil CO2 efflux and it is strongly related to soil temperature as a classic Q10 relationship between biological activity and temperature [40]. Sim ci c et al. [26] in their study of experimental inundation and incubation temperatures, pointed out that the temperature sensitivity of inundated soils and sediments was higher of that previously measured in non-flooded terrestrial and floodplain soil [20,41], indicating that the thermal sensitivity of soil respiration most likely increases during inundation. Other studies reported different range of Q10 values of soil temperature at different depths ranging from 1 to 4 [23,29,42e44]. Nevertheless, it is necessary to point out that our Q10 value at the floodplain forest ecosystem was obtained during a season without flood event. Therefore, this Q10 should be taken as a value during low water availability at the study floodplain forest. Numerous models have been developed to express temperature sensitivity of soil CO2 efflux [23,29,40,45,46]. However, models make many simplifying assumptions and these assumptions can have a large impact on model predictions. Regardless of the simplicity of our model, it is built on data set that is underrepresent ecosystem, it follows the logic that the main abiotic driver that determines soil CO2 efflux is soil temperature. When comparing the SR model and measured data, our model shows a good agreement with our measured chamber data during the summer months (JuneeAugust) while underestimation of SR is observed during the spring and autumn months (Fig. 4). This observed underestimation was likely due to the lack of large diurnal variation of soil temperature during the measurement campaigns and it could lead to inaccurate Q10 calculation. Another possible reason could be the small variation of R10 during the year which might have been due to low activity of root growth or photosynthesis during spring and autumn. Darenova et al. [17], pointed out that SR measurements only during day time could underestimate determination of seasonal sums of CO2 released from soil. In our study, all the campaigns were done during the day time. Moreover, our model was based on a monthly measurement campaigns, leading to missing the effect of episodic rain events that sometimes results huge SR pulse through stimulation of microorganisms [47]. Nevertheless, the used model presents an overview of the soil CO2 efflux dynamic over the investigated growing season. Concerning our geostatistical analysis on our measured soil CO2 efflux positions (Fig. 5), it showed that during the campaign in May there was no high differences in soil CO2 efflux rates between sampling positions that were close together and those that were far apart. This may be connected to the hypothesis that when the favourable soil conditions decrease, the variability of soil CO2 efflux was reduced. Stoyan et al. [18] pointed out that the more structured variability observed in the deeper soil supports the hypothesis that the heterogeneity is due either to the non-uniform addition of organic material to the surface or the variations in moisture content. The soil moisture and organic material varied within the soil, resulting in smoother shifts between high and low soil respiration areas. The summer months (June, July and August, Fig. 2) showed
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
higher differences in soil CO2 efflux rates among sampling positions regarding distance compared to late spring (May). This could be explained by the activity of decomposers working on leaf litter from previous fall in a moisture environment and higher soil temperature in summer resulted in soil activity in our floodplain forest that was higher in summer than in late spring. November showed low differences in soil CO2 efflux rates between sampling positions that were close together but higher differences in rates for those that were far apart. We consider that a possible explanation of this could be connected with the vicinity of trees to individual measuring position, we observed that position closed to tree showed higher soil CO2 efflux compared to those far of them. The same situation was regarding the influence of herbal understory in positions where vegetation was presented compared to positions without vegetation. In a study of spatial variability of soil CO2 efflux, Brito et al. [48], pointed out that depending on the topographic position of the experiment, soil CO2 emission could present differences in range values, which represents the changes in scale of each property. However, it is worth to point out that geostatistical analyses provide powerful analytical tools to capture the horizontal variability of a property and have received increasing interest by soil biologists in recent years. In its simplest form geostatistics define the degree of autocorrelation among the measured data points and interpolates values between measured points based on the degree of autocorrelation encountered [49], but it is tricky to define the reason of this autocorrelation. As was stated before, the present pilot study was designed to address the general need for basic information about soil CO2 efflux heterogeneity in floodplain forest and to gain information about the factors influencing this heterogeneity. Our results represent a novel information about soil CO2 efflux in this kind of ecosystem that has been missing concerning carbon fluxes studies. The magnitude of soil CO2 effluxes and its heterogeneity in floodplain forest is mainly caused by the rich biodiversity of this kind of ecosystem and it is significant compared to upland forest ecosystems. Our results also provide a conceptual understanding of soil CO2 efflux dynamics in floodplain forests and the main environmental factors influencing these dynamics in order to improve the knowledge of carbon forest floor emission in this kind of terrestrial ecosystem. Our pilot study also demonstrated the importance of floodplain forest in terms of carbon issues and the need for longer and deeper investigations. 5. Conclusions Our study provides evidence of the influence of soil temperature, as a main factor, on SR and the seasonal variability of SR at floodplain forest ecosystem. In our floodplain forest ecosystem, soil moisture was not a limiting factor of SR. The spatial heterogeneity of SR at our investigated forest was 20%. The cumulative amount of carbon forest floor released from our experimental forest site was 7.4 (±1.1) tC ha1 y1 for the 2016. The majority of carbon cycling studies including modelling are mainly carried out in typical forest ecosystem and do not take into account the specific conditions of floodplain forest ecosystem. Even though, our study was carried out in a season without floods, the results of our study help to improve our knowledge and models concerning carbon issues in this kind of ecosystem during period of low water availability. Such data are important for evaluating the effect of climate change or other possible influences on carbon cycling and carbon sequestration in lowland floodplain forest ecosystems. Acknowledgements This work was supported by the Ministry of Education, Youth
41
and Sports of the Czech Republic within the National Sustainability Program I (NPU I), grant number LO1415 and grant number LD15040. We would like to thank to Mr. Ryan McGloin for the linguistic revision. References camps, The ecology of interfaces: riparian zones, Annu. Rev. [1] R.J. Naiman, H. De Ecol. Syst. 28 (1997) 621e658. [2] K. Tockner, J.A. Stanford, Riverine flood plains: present state and future trends, Environ. Conserv. 29 (2002) 308e330. [3] K. Tockner, F. Malard, J.V. Ward, An extension of the flood pulse concept, Hydrol. Process. 14 (2000) 2861e2883. [4] K. Tockner, S.E. Bunn, C. Gordon, R.J. Naiman, G.P. Quinn, J.A. Standord, Flood plains: critically threatened ecosystems, in: N.V.C. Polunin (Ed.), Aquatic Ecosystems: Trends and Global Prospects, Cambridge University Press, Cambridge, 2008, pp. 45e61. [5] D.P. Turner, S.V. Ollinger, J.S. Kimball, Integrating remote sensing and ecosystem process models for landscape- to regional-scale analysis of the carbon cycle, BioScience 54 (2004) 573e584. [6] J. Cerm ak, Leaf distribution in large trees and stands of the floodplain forest in southern Moravia, Tree Physiol. 18 (1998) 727e737. [7] M. Kazda, J. Salzer, I. Reiter, Photosynthetic capacity in relation to nitrogen in the canopy of Quercus robur, Fraxinus angustifolia and Tilia cordata flood plain forest, Tree Physiol. 20 (2000) 1029e1037. [8] J. Glaser, M. Wulf, Effects of water regime and habitat continuity on the plant species composition of floodplain forests, J. Veg. Sci. 20 (2009) 37e48. [9] T.T. Kozlowski, Responses of woody plants to flooding and salinity, Tree Physiol. 1 (1997) 1e29. [10] M. Southwell, M. Thoms, Patterns of nutrient concentrations across multiple floodplain surface in a large dryland river system, Geogr. Res. 49 (2011) 431e443. [11] T.J. Battin, S. Luyssaert, L.A. Kaplan, A.K. Aufdenkampe, A. Richter, L.J. Tranvik, The boundless carbon cycle, Nat. Geosci. 2 (2009) 598e600. [12] D. Baldwin, G.N. Rees, J.S. Wilson, M.J. Colloff, K.L. Whitwoth, T.L. Pitman, T.A. Wallace, Provisioning of bioavailable carbon between the wet and dry phases in a semi-arid floodplain, Oecologia 172 (2013) 539e550. [13] M. Reichstein, M. Bahn, P. Ciais, D. Frank, M.D. Mahecha, S.I. Seneviratne, J. Zscheischler, C. Beer, N. Buchmann, D.C. Frank, D. Papale, A. Rammig, P. Smith, K. Thonicke, M. van der Velde, S. Vicca, A. Walz, M. Wattenbach, Climate extremes and the carbon cycle, Nature (2013) 287e295. [14] I. Rieger, F. Lang, B. Kleinschmit, I. Kowarik, A. Cierjacks, Fine root and aboveground carbon stocks in riparian forests: the roles of diking and environmental gradients, Plant Soil 370 (2013) 497e509. [15] J.H. Van’ t Hoff, Lectures on theoretical and physical chemistry, in: R.A. Lehfeldt (Ed.), Part I Chemical Dynamics, Edward Arnold, London, 1898, pp. 224e229. [16] S. Linder, E. Troeng, The seasonal variation in stem and coarse root respiration of a 20-year-old Scots pine (Pinus sylvestris L.), in: W. Tranquillini (Ed.), €ume, Mitt. Forstl., vol. 142, Bundesversuchsanst Wien, Dickenwachstum der Ba 1981, pp. pp125e140. [17] E. Darenova, M. Pavelka, M. Acosta, Diurnal deviations in the relationship between CO2 efflux and temperature: a case study, Catena 123 (2014) 263e269. € hm, G.P. Robertson, E.A. Paul, Spatial heterogeneity [18] H. Stoyan, H. De-Polli, S. Bo of soil respiration and related properties at the plant scale, Plant soil 222 (2000) 203e214. [19] J.R. Ribeiro, P.J. Diggle, geoR: a package for geostatistical analysis, R-News 1 (2) (2001). ISSN 1609e3631. [20] N. Buchmann, Biotic and abiotic factors controlling soil respiration rates in Picea abies stands, Soil Biol. Biochem. 32 (2000) 1625e1635. [21] D. Gaumont-Guay, T.A. Black, H. Mccaughey, A.G. Barr, P. Krishnan, R.S. Jassal, Z. Nesic, Soil CO2 efflux in contrasting boreal deciduous and coniferous stands and its contribution to the ecosystem carbon balance, Glob. Change Biol. 15 (2009) 1302e1319. [22] A. Shvaleva, R. Lobo-do-Vale, C. Cruz, S. Castaldi, A.P. Rosa, M.M. Chaves, J.S. Pereira, Soil-atmosphere greenhouse gases (CO2, CH4 and N2O) exchange in evergreen oak woodland in southern Portugal, Plant Soil Environ. 57 (10) (2011) 471e477. [23] B. Bond-Lamberty, A. Thomson, A global database of soil respiration data, Biogeosciences 7.6 (2010) 1915e1926. [24] J. Gomez, P. Vidon, J. Gross, C. Beier, J. Caputo, M. Mitchell, Estimating greenhouse gas emissions at the soileatmosphere interface in forested watersheds of the US Northeast, Environ. Monit. Asses 188 (5) (2016) 1e16. [25] M. Khomik, M.A. Arain, J.H. Mccaughey, Temporal and spatial variability of soil respiration in a boreal mixewood forest, Agric. For. Meteorol. 140 (2006) 244e256. [26] T. Sim ci c, H. Mori, C. Hossli, C.T. Robinson, M. Doering, The response in floodplain respiration of an alpine river to experimental inundation under different temperature regimes, Hydrol. Process. 29 (2015) 5438e5450. [27] J.J. Dilustro, B. Collins, L. Duncan, C. Crawford, Moisture and soil texture effects on soil CO2 efflux components in southeastern mixed pine forests, For. Ecol. Manag. 204 (1) (2005) 87e97.
42
M. Acosta et al. / European Journal of Soil Biology 82 (2017) 35e42
[28] J. Chen, Q. Wang, M. Li, F. Liu, W. Li, L. Yin, Effects of deer disturbance on soil respiration in a subtropical floodplain wetland of the Yangtze River, Eur. J. Soil Biol. 56 (2013) 65e71. [29] J.W. Raich, C.S. Potter, D. Bhagawati, Interannual variability in global soil respiration, 1980e94, Glob. Change Biol. 8 (8) (2002) 800e812. [30] S.A. Billings, D.D. Richter, J. Yarie, Soil carbon dioxide fluxes and profile concentrations in two boreal forests, Can. J. For. Res. 28 (12) (1998) 1773e1783. [31] B.E. Law, F.M. Kelliher, D.D. Baldocchi, P.M. Anthoni, J. Irvine, D. Moore, S. Van Tuyl, Spatial and temporal variation in respiration in a young ponderosa pine forest during a summer drought, Agric. For. Meteorol. 10 (2001) 27e43. [32] Y. Kosugi, T. Mitani, M. Ltoh, S. Noguchi, M. Tani, N. Matsuo, S. Takanashi, S. Ohkubo, A.R. Nik, Spatial and temporal variation in soil respiration in a Southeast Asian tropical rainforest, Agric. For. Meteorol. 147 (2007) 35e47. [33] E. Darenova, M. Pavelka, L. Macalkova, Spatial heterogeneity of CO2 efflux and optimization of the number of measurement positions, Eur. J. Soil Biol. 75 (2016) 123e134. [34] J. Tang, D.D. Baldocchi, Spatialetemporal variation in soil respiration in an oakegrass savanna ecosystem in California and its partitioning into autotrophic and heterotrophic components, Biogeochemistry 73 (1) (2005) 183e207. [35] F.E. Moyano, L.K. Werner, C. Rebmann, Soil respiration fluxes in relation to photosynthetic activity in broad-leaf and needle-leaf forest stands, Agric. For. Meteorol. 148 (2008) 135e143. [36] P.J. Hanson, N.T. Edwards, C.T. Garten, J.A. Andrews, Separating root and soil microbial contributions to soil respiration: a review of methods and observations, Biogeochemistry 48 (2000) 115e146. [37] R.S. Jassal, T.A. Black, M.D. Novak, D. Gaumont-Guay, Z. Nesic, Effect of soil water stress on soil respiration and its temperature sensitivity in an 18-yearold temperate Douglas-fir stand, Glob. Change Biol. 14 (2008) 1305e1318. [38] W. Borken, Y.J. Xu, E.A. Davidson, F. Beese, Site and temporal variation of soil respiration in European beech, Norway spruce, and Scots pine forests, Glob. Change Biol. 8 (2002) 1205e1216. [39] G. Saiz, K. Black, B. Reidy, S. Lopez, E.P. Farrell, Assessment of soil CO2 efflux and its components using a process-based model in a young temperate forest site, Geoderma 139 (2007) 79e89.
[40] J. Lloyd, J.A. Taylor, On the temperature dependence of soil respiration, Funct. Ecol. 8 (1994) 315e323. [41] E. Samaritani, J. Shrestha, B. Fournier, E. Frossard, F. Gillet, C. Guenat, P.A. Niklaus, K. Tockner, E.A.D. Mitchell, Heterogeneity of soil carbon pools and fluxes in a channelized and a restored floodplain section (Thur River, Switzerland), Hydrol. Earth Syst. Sci. 15 (2011) 1757e1769. [42] E. Davidson, I. Jansens, Y. Luo, On the variability of respiration in terrestrial ecosystems: moving beyond Q10, Glob. Change Biol. 12 (2) (2006) 154e164. [43] P.J. Hanson, N.T. Edwards, C.T. Garten, J.A. Andrews, 2000. Separating root and soil microbial contributions to soil respiration: a review of methods and observations, Biogeochemistry 48 (2000) 115e146. [44] M. Acosta, M. Pavelka, L. Montagnani, W. Kutsch, A. Lindroth, R. Juszczak, D. Janous, Soil surface CO2 efflux measurements in Norway spruce forests: comparison between four different sites across Europedfrom boreal to alpine forest, Geoderma 192 (2013) 295e303. [45] M. Reichstein, A. Rey, A. Freibauer, J. Tenhunen, R. Valentini, J. Banza, P. Casals, Y. Cheng, J.M. Grünzweig, R. Israel, J. Irvine, R. Joffre, B.E. Law, D. Loustau, F. Miglietta, W. Oechel, J.M. Ourcival, J.S. Pereira, A. Peressotti, F. Ponti, Y. Qi, S. Rambal, M. Rayment, J. Romanya, F. Rossi, V. Tedeschi, G. Tirone, M. Xu, D. Yakir, Modeling temporal and large-scale spatial variability of soil respiration from soil water availability, temperature and vegetation productivity indices, Glob. Biogeochem. Cycles 17 (2003) 1104. [46] M. Tuomi, P. Vanhala, K. Karhu, H. Fritze, J. Liski, Heterotrophic soil respiration -comparison of different models describing its temperature dependence, Ecol. Model. 211 (2008) 182e190. [47] X. Liu, S. Wan, B. Su, D. Hui, Y. Luo, Response of soil CO2 efflux to water manipulation in a tallgrass prairie ecosystem, Plant Soil 240 (2) (2002) 213e223. [48] L.D.F. Brito, J. Marques Júnior, G.T. Pereira, N. La Scala Junior, Spatial variability of soil CO2 emission in different topographic positions, Bragantia 69 (2010) 19e27. [49] G.P. Robertson, Geostatistics in ecology: interpolating with known variance, Ecology 68 (3) (1987) 744e748.