Can microbial fuel cells be an effective mitigation strategy for methane emissions from paddy fields?

Ecological Engineering 60 (2013) 167–171

Contents lists available at ScienceDirect

Ecological Engineering journal homepage: www.elsevier.com/locate/ecoleng

Short communication

Can microbial fuel cells be an effective mitigation strategy for methane emissions from paddy fields? A. Rizzo ∗ , F. Boano, R. Revelli, L. Ridolfi Department of Environment, Land and Infrastructure Engineering (DIATI), Politecnico di Torino, Turin, Italy

a r t i c l e

i n f o

Article history: Received 12 March 2013 Received in revised form 29 May 2013 Accepted 4 July 2013 Available online 15 August 2013 Keywords: Microbial fuel cell Methane emission Paddy soil Process-based model

a b s t r a c t Microbial fuel cells (MFCs) are bioelectrochemical systems able to generate electricity from wetland soils, including paddies, exploiting the microbial decomposition of organic matter. Additionally, MFCs can also be applied as a novel mitigation strategy for emissions of methane (CH4 ) from paddy fields. The MFC efficiency in reducing CH4 fluxes is still poorly understood, and it is here investigated via a one-dimensional process-based model that simulates the vertical and temporal dynamics of the chemical compounds affecting CH4 fate within paddy soil. The DOC oxidation rate of MFC is modeled assuming a zero-order kinetics proportional to the generated electricity and different anode depths and current densities are tested. By assuming current densities presently achieved in paddies, MFCs are able to reduce up to −27.9%, −16.7%, and −22.0% of daily minimum, daily maximum, and total CH4 emissions, respectively. Moreover, CH4 reductions are even higher (up to −28.1%, −24.1%, and −26.5%) if we assume 5% of the current density developed on laboratory acetate-fed MFC. The system shows a limiting effect of transport processes on the mitigation of CH4 emissions at high current density. In order to maximize the reduction of CH4 emissions, simulation results suggest to place the anode in the middle portion of the superficial layer of paddy soil. The findings demonstrate the need to further investigate and develop this new technology for field-scale applications. © 2013 Elsevier B.V. All rights reserved.

1. Introduction The eco-compatibility of rice production is compromised by the large emissions of methane (CH4 – one of the most potent greenhouse gases) from paddy fields, estimated in 9÷ 19 % and 15÷ 26 % of the global and anthropogenic CH4 emissions, respectively (Denman, 2007). To tackle this issue, microbial fuel cells (MFCs) can represent a novel CH4 mitigation strategy (de Schamphelaire et al., 2008; Deng et al., 2012). MFCs can be applied within wetlands with different electrode configurations (e.g., Deng et al., 2012; Chen et al., 2012), but the most common one is composed of an anode buried in the anaerobic submerged soil linked to a cathode placed on the aerobic top of the soil and submerged by ponding water (e.g., Kaku et al., 2008; Takanezawa et al., 2010). A biofilm develops on the anode, where bacteria generate electricity, oxidizing dissolved organic carbon (DOC) and using oxygen (O2 ) available on the cathode as electron acceptor (EA) (Deng et al., 2012). MFC technology is still in an early development stage and the efficiency in electricity production is very low (Pant et al., 2010). However, MFCs can be applied for secondary aims, among which one of great interests is

the reduction of CH4 emissions from wetlands (de Schamphelaire et al., 2008; Deng et al., 2012). Indeed, DOC oxidation at the anode can be seen as an additional DOC sink in wetland environment, limiting the DOC availability for methanogens. The MFC efficiency in mitigating CH4 emissions from lake sediments has been recently confirmed by field experiments (Jeon et al., 2012), while from paddy fields data are still scarce and conflicting. Kaku et al. (2008) observed no significant MFC effects on CH4 emissions in field experiments; methanogenesis decrease is instead reported in laboratory experiments on cellulose fed-MFC (Rismani-Yazdi et al., 2013) and paddy soil MFC (Ishii et al., 2008; Cabezas, 2010). In particular, Cabezas (2010) referred a reduction in CH4 production up to 50% from laboratory analysis on incubated rice samples with sediment-MFC. Hence, in order to clarify the potential of MFCs in mitigating CH4 emissions in paddies, a preliminary investigation is developed using the process-based model proposed by Rizzo et al. (2013), which is properly updated to model MFC-induced processes within paddy soil.

2. Methods ∗ Corresponding author. Tel.: +39 0110905674. E-mail addresses: [email protected] (A. Rizzo), [email protected] (F. Boano), [email protected] (R. Revelli), luca.ridolfi@polito.it (L. Ridolfi). 0925-8574/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ecoleng.2013.07.033

The hydro-biogeochemical process-based model (Rizzo et al., 2013) quantifies the dynamics of chemical compounds affecting CH4 production, oxidation, and emission. Many physico-chemical

168

A. Rizzo et al. / Ecological Engineering 60 (2013) 167–171

processes and paddy soil features have been included in the model, such as paddy soil stratigraphy, detailed root compartment modeling, transport flows, biogeochemical reactions, gas transport, and respiration within roots. In order to simulate the effect of MFCs on CH4 dynamics, the model is now coupled with an MFC model based on the assumption of zero-order kinetics for DOC oxidation driven by the MFC. In the following sections, the main features of the original model are briefly revised (see Rizzo et al., 2013 for more details), while the MFC model is described in detail. 2.1. Hydro-chemical model The paddy soil is schematized as three soil layers with homogeneous physico-chemical features (e.g., hydraulic conductivity, bulk density, porosity). The three layers are denoted as muddy, hard pan, and non-puddled layers and the subscripts k = 1, 2, 3 are used to distinguish their respective variables. The muddy and the hard pan layers are assumed in saturated conditions, while the non-puddled layer is unsaturated. The depth below ground surface of the lower interface of each layer is denoted as zk (see the online Supplemetary Materials). Roots are assumed to be present only in the muddy layer and are treated as variable in space and time via a root density distribution and a plant–root development function, respectively. Root density is modeled with an exponential decrease over depth, with root area and volume density differentiated between primary and secondary roots and empirically linked to root biomass. The plant–root development function modulates the variation over time of plant–root features such as root biomass, root exudation, plant transpiration, and tiller area extension; this function considers a linear increase of plant features from the germination up to a maximum plant development stage, followed by a linear decrease up to maturity. The infiltration rate, qk , is modeled with the Darcy’s equation for the saturated layers, while it is assumed constant, and equal to the value of hard pan layer, in the unsaturated non-puddled layer (see Rizzo et al., 2013 for more details). A term for root water uptake driven by plant transpiration is included in the water mass balance equation. The dissolved species considered in the model are: DOC, O2 , NO− , NH+ , Mn2+ , Fe2+ , SO2− and CH4 , where DOC is the dissolved 3 4 4 organic carbon pool. The modeled solid species are: SOC, SOCr , MnO2 , and Fe(OH)3 ; SOC is the soil organic carbon, while SOCr is the pool of organic carbon originated by root dieback. For each ith dissolved species and jth solid species the mass balance equations, referred to bulk volume of the kth layer, are

 k + k

s ∂Ci,ads

∂Cil



∂Cil ∂t

=−

∂ ∂z

 qk Cil − k Dh,i

∂Cil ∂z

 + Ri

(1)

The plant-mediated gas exchange between soil and atmosphere is modeled via a gas mass balance equation within root aerenchyma, referred to soil bulk volume, expressed as Va

g ∂Cr,o ∂ =− ∂t ∂z



Va Dd,o

g ∂Cr,o ∂z



+ Ro ,

(3)

g

where Cr,o is the concentration of the oth gas (CH4 and O2 ) within root aerenchyma, Va is the aerenchyma pore volume, and Dd,o is the coefficient of molecular diffusion of the oth gas in the air phase. Ro is the sum of all sink terms involving the oth gas within aerenchyma, which include root respiration for O2 and the exchange flux at rootsoil interface for both O2 and CH4 . The top boundary condition of Eq. (3) models the plant-mediated flux of oth gas. The other two minor pathways for CH4 emissions from paddy field, i.e. ebullition and diffusion, are also included. Ebullition is modeled assuming the CH4 solubility concentration as threshold value and removing all CH4 in excess, while the diffusion fluxes is assumed equal to the dispersive transport flux at the top of muddy layer. Moreover, the model is updated to simulate the water temperature vertical profile in paddy soil solving the heat transport equation

∂T ∂ c = ∂t ∂z



∂T  ∂z



− cw w qk

∂T ∂z

(4)

where T is the water temperature in soil, c = [(1 − nk )cs k + k cw w ] is the bulk specific heat of the soil per bulk soil density, and  = [(1 − nk )s + k w ] is the bulk heat conductivity of the soil. nk is the soil porosity, and cs , cw , s , w , s , w are the specific heat, heat conductivity, and density, respectively (subscripts s and w denote soil and water components). The boundary conditions for Eq. (4) are T |z=0 = Tw (t) and ∂T/∂z|z=z3 = 0 (heat convection only) for the top of muddy layer and the bottom of the non-puddled layer, respectively, where Tw (t) is the temperature of ponding water depth. The initial soil temperature is set equal to Tw (t = 0). The ponding water temperature is assumed Tw (t) = Tair (t) + Tw , where Tair (t) is the air temperature and Tw is a constant difference between ponding water and air temperature. Tair (t) is assumed to follow a sinusoidal behavior during the year Tair (t) = A sin (ωt + ) + T air

(5)

where A is the temperature amplitude, ω is the angular frequency, ϕ is the phase, and T air in the mean air temperature. The kinetic biogeochemical parameters, the conductivity of plant-mediated CH4 emissions, and the diffusion coefficients are modulated with temperature following a Q10 approach. 2.2. MFC model

∂Cjs ∂t

= Rj ,

(2)

where t is time, z is depth,  k is soil moisture, k is soil density, s Ci,ads is concentration adsorbed on solid matrix, Cil is concentration of dissolved species, Cjs is concentration of solid species, and Dh,i is the coefficient of hydrodynamic dispersion. Ri and Rj are the sum of all sink terms involving the ith dissolved species and the jth solid species, respectively, i.e. soil organic matter decomposition, primary and secondary biogeochemical reactions driven by DOC oxidation in anaerobic conditions and radial oxygen loss from roots, root solute uptake, and root exudation. A list of the biogeochemical reactions described by the model and a scheme of the input/outputs for each chemical compound are reported in the online Supplementary Materials.

The MFC is composed of a planar horizontal anode buried at a specific depth, z* , linked to a cathode placed on the top of the soil (see the online Supplementary materials) (Kaku et al., 2008; Takanezawa et al., 2010). From the point of view of paddy soil biogeochemistry, the MFC is treated as an additional pathway of DOC decomposition, which is modeled by a DOC oxidation rate due to ∗ ∗ MFC, RDOC , localized at the anode depth. RDOC is assumed to be * the only DOC oxidation process between z − s/2 and z* + s/2, with s anode thickness, due to the following reasons: (i) aerobic microorganisms form a biofilm on the anode surface which uses O2 as EA (provided by the cathode) for DOC decomposition, so we suppose that other anaerobic microorganisms cannot compete for the same DOC at the anode depth; (ii) the anode projection area is assumed to cover 75% of corresponding bulk horizontal area (Kaku et al., 2008),

A. Rizzo et al. / Ecological Engineering 60 (2013) 167–171

so the paddy soil portion not covered by anode can be neglected. Hence, the overall DOC oxidation rate, RDOC , is

RDOC,s ,

and z ≤ z ∗ + s/2,

0.8

(6)

otherwise

where RDOC,s denotes the microbial DOC oxidation rate developed ∗ in soil, modeled as reported in Rizzo et al. (2013). RDOC is modeled with a zero-order kinetics as follows



∗ RDOC

IA∗p np

DOC = min , eDOC F t

(7)

3. Results and discussion The MFC efficiency in reducing CH4 emissions is investigated running simulations with different MFC anode depths, z* , and current intensities, I. Four different z* are considered (Kaku et al., 2008; Takanezawa et al., 2010): 2 cm, 5 cm, 10 cm, and 15 cm. Four different I are chosen according to the minimum, mean, and maximum values observed in literature from MFC applied in paddy fields (Kaku et al., 2008; Takanezawa et al., 2010): 20 mA m−2 , 80 mA m−2 , 160 mA m−2 , and 400 mA m−2 . The highest current density is chosen observing that the maximum I currently developed in paddy soil is only 2% of those registered from MFC chamber fed with acetate (i.e. the same typical substrate of paddy soil; Rizzo et al., 2013) in laboratory experiments (Pant et al., 2010). Hence, we assume a technical MFC improvement which allows to reach I equal to the 5% of that obtained in laboratory, i.e. 400 mA m−2 . y The simulations are called Sx with the subscript x and superscript y representing the values of anode depth z* (in cm) and current density I (mA m−2 ), respectively (e.g., S220 represents the simulation with z* = 2 cm and I = 20 mA m−2 ). The notation with only a subscript and supercripts refer to the set of simulations with the same z* and I, respectively (e.g., S5 and S400 denote all the simulations with z* = 5 cm and I = 400 mA m−2 , respectively). The base simulation without MFC is denoted as S0. The values of parameters z* and I of each simulation are summarized in Table 1. For each simulation, s, A∗p , and np are set equal to 3 mm, 337 cm2 plant−1 , and 22 plant m−2 , respectively (Kaku et al., 2008). The values of temperature model parameters are resumed and discussed in the online Supplementary Materials, while for the other values see Rizzo et al. (2013). Eqs. (1)–(4) are resolved via a finite element approach. Details about mesh size and discretization, time step, values of boundary and initial conditions are reported in Rizzo et al. (2013). The simulated CH4 emissions are reported in Table 1, showing the values of daily minimum, daily maximum, and total flux over the whole growing season. The CH4 emissions during the growing season are strongly affected by the ability to generate high current densities, as visible in Fig. 1 where daily CH4 fluxes of S10 simulations are shown. Interestingly, all simulations with minimum I (Kaku et al., 2008) (S20 ) show an insignificant reduction of CH4 (always lower than 2.5%), in agreement with the observations of Kaku et al. (2008). Generally, higher I always leads to higher mitigation of CH4 emissions (see Table 1). However, the relationship between CH4 emissions and current density (Fig. 2, upper panel), exhibits an asymptotic plateau with increasing I. This suggests that the beneficial effect

0.4

0.2



where I is the current density per anode projection area, A∗p is the anode projection area per plant, np is the number of plants per square meter of paddy field, F is the Faraday constant, eDOC is the mole of electrons donated by a mole of DOC, and t is the time step; the minimum function is used to avoid that the zero-order kinetics results in DOC concentration below zero.

0.6

4

z≥z ∗ − s/2

CH flux [g m−2 d−1]

RDOC =

∗ RDOC ,

1

0 0

20

40

60 t [d]

80

100

120

Fig. 1. Temporal dynamics of CH4 emission for simulations with MFC anode depth at 10 cm (S10 ) and different current density: 0 mA m−2 , i.e. no MFC application (S0 20 80 – thin continuous line); 80 mA m−2 (S10 – – thick continuous line); 20 mA m−2 (S10 160 400 – dash-dotted line); 400 mA m−2 (S10 – dotted line). dashed line); 160 mA m−2 (S10

of MFCs on CH4 emissions is limited by the transport processes at higher I, i.e. the rate of DOC oxidation by the MFC is limited by the rate of DOC supply to the anode by advection and dispersion. The spatial profiles of DOC and CH4 production exposed in Fig. 3 also confirm the limiting effect of transport processes, showing differences between S80 and S160 results more pronounced than those between S160 and S400 . The depth of MFC anode, z* , strongly influences the capability of the MFCs to reduce CH4 emissions due to the DOC vertical profile within paddy soil, characterized by higher concentrations at the bottom of muddy layer (Rizzo et al., 2013). As shown in the leftupper panel of Fig. 3, when the MFC anode is positioned near the top of the muddy layer (S2 ) the DOC available for oxidation by the MFC is low; consequently, the reduction of CH4 production is also low (see right-upper panel of Fig. 3) and the mitigation of CH4 emissions is scarce even at high I (S2400 , −14.2%, −11.7%, and −12.0% on daily minimum, daily maximum, and total CH4 emissions over the whole growing season, respectively). Conversely, in the case of an anode buried deeper in the soil, the MFC influences the DOC spatial profile more effectively due to the higher available DOC concentration (S10 – see the left-lower panel of Fig. 3); in this way, there is a greater reduction of CH4 production (see right-lower panel of Fig. 3) which leads to a significant mitigation of CH4 emissions (up to −28.1%, −24.1%, and −26.5% on daily minimum, daily maximum, and total CH4 emissions, respectively). However, an anode positioned near the bottom of muddy layer (i.e. S15 ) does not result in the lowest 55 50 −2 total CH4 flux [g m ]



169

45 40 20

80

160

400 −2

I [mA m ] 55 50 45 40 2

5

10

15

*

z [cm]

Fig. 2. Total CH4 emission over the whole growing season as a function of different current densities (upper panel) and MFC anode depths (lower panel). In the upper panel, simulations with different anode depths are shown: 2 cm (S2 – continuous line and squares), 5 cm (S5 – dashed line and circles), 10 cm (S10 – dash-dotted line and diamonds), 15 cm (S15 – dotted line and upward triangles). In the lower panel, simulations with different current densities are shown: 20 mA m−2 (S20 – continuous line and downward triangles); 80 mA m−2 (S80 – dashed line and rightward triangles); 160 mA m−2 (S160 - dash-dotted line and leftward triangles); 400 mA m−2 (S400 – dotted line and pentagrams).

170

A. Rizzo et al. / Ecological Engineering 60 (2013) 167–171

Table 1 CH4 emission results for simulations with different MFC configurations: no MFC (S0); anode depth equal to 2, 5, 10, 15 cm (S2 , S5 , S10 , S15 , respectively); current density equal to 20, 80, 160, 400 mA m−2 (S20 , S80 , S160 , S400 , respectively). The deviations between the simulated emissions of each simulation and those of S0 are exposed as percentages in parentheses.

S0

I [mA m−2 ] 0

z* [cm] no

Min CH4 flux [g m−2 d−1 ] 0.26

Max CH4 flux [g m−2 d−1 ] 0.95

Tot CH4 flux [g m−2 ] 54.55

S220

20

2

0.24 (−5.5%)

0.94 (−1.2%)

53.46 (−2.0%)

S280

80

2

0.22 (−15.2%)

0.86 (−9.7%)

48.70 (−10.7%)

S2160

160

2

0.22 (−14.9%)

0.84 (−11.6%)

48.17 (−11.7%)

S2400

400

2

0.22 (−14.2%)

0.84 (−11.7%)

48.00 (−12.0%)

S520

20

5

0.24 (−5.5%)

0.93 (−1.9%)

53.31 (−2.3%)

S580

80

5

0.20 (−21.8%)

0.87 (−8.8%)

47.87 (−12.3%)

S5160

160

5

0.20 (−22.8%)

0.80 (−16.0%)

44.48 (−18.5%)

S5400

400

5

0.20 (−22.6%)

0.77 (−18.7%)

43.61 (−20.1%)

20 S10

20

10

0.24 (−5.3%)

0.93 (−2.3%)

53.23 (−2.4%)

80 S10

80

10

0.21 (−20.4%)

0.87 (−8.5%)

47.94 (−12.1%)

160 S10

160

10

0.19 (−27.9%)

0.79 (−16.7%)

42.54 (−22.0%)

400 S10

400

10

0.29 (−28.1%)

0.72 (−24.2%)

40.09 (−26.5%)

20

15

0.25 (−4.6%)

0.93 (−2.4%)

53.13 (−2.6%)

20 S15 80 S15

15

0.21 (−17.0%)

0.88 (−7.2%)

48.82 (−10.5%)

160

15

0.20 (−23.0%)

0.81 (−14.2%)

44.27 (−18.9%)

400 S15

400

15

0.20 (−23.2%)

0.75 (−20.9%)

42.27 (−22.5%)

z−depth [m]

80

160 S15

0

0

0.1

0.1

0.2 0

1

2

3

0.2 0

0

0

0.1

0.1

0.2 0

1 2 DOC [mol m−3]

3

0.2 0

3

6

3 CH production [µmol m−3]

6

4

Fig. 3. Vertical profile of DOC (left column) and CH4 production (right column) for simulations with MFC anode depth at 2 cm (S2 – upper panels) and 10 cm (S10 - lower panels) at different current densities: 0 mA m−2 , i.e. no MFC application (S0 – thick continuous line); 20 mA m−2 (thin continuous line); 80 mA m−2 (dashed line); 160 mA m−2 (dash-dotted line); 400 mA m−2 (dotted line). The reported results correspond to the 60th day of the growing season.

CH4 emissions (see the lower panel of Fig. 2), since the reduction of CH4 production is localized in a soil portion that contribute less to total CH4 emissions (lower root density, lower contribution to plant-mediated patways Rizzo et al., 2013). Finally, the reductions of CH4 fluxes driven by the maximum I currently developed within paddy environment (Takanezawa et al., 2010) are also very interesting (S160 – up to −27.9%, −16.7%, and −22.0% on daily minimum, daily maximum, and total CH4 emissions, respectively) confirming the potential of MFCs as novel mitigation strategy for CH4 emissions from paddy fields.

4. Conclusion The study of MFCs as novel strategy to face CH4 emissions from paddy fields is in a very infant stage. Many issues still need to be investigated, including the feasibility of the large-scale application in real paddy fields. However, our results confirm the potential of this technique, since we have obtained reductions up to −28.1%, −24.1%, and −26.5% for daily minimum, daily maximum, and total CH4 emissions, respectively. Moreover, our simulations provide

other important insights about MFC effectiveness in reducing CH4 emissions, such as the limiting effect of within-soil transport processes at high current density and the identification of middle of the muddy layer as the best position for the MFC anode in order to minimize the CH4 emissions. Acknowledgment A.R. thanks the Italian Ministry of University and Research (MIUR) for funding the Ph.D. project. Appendix A. Supplementary Data Supplementary data associated with this article can be found, in the online version, at doi:http://dx.doi.org/10.1016/ j.ecoleng.2013.07.033. References Cabezas, A., 2010. Chapter 6: Taming methane emissions from rice field soil with microbial fuel cells. PhD Thesis. University of Marburg.

A. Rizzo et al. / Ecological Engineering 60 (2013) 167–171 Chen, Z., Huang, Y.-c., Liang, J.-h., Zhao, F., Zhu, Y.-g., 2012. A novel sediment microbial fuel cell with a biocathode in the rice rhizosphere. Bioresour. Technol. 108, 55–59. de Schamphelaire, L., van den Bossche, L., Dang, H.S., Hofte, M., Boon, N., Rabaey, K., Verstraete, W., 2008. Microbial fuel cells generating electricity from rhizodeposits of rice plants. Environ. Sci. & Technol. 42 (8), 3053– 3058. Deng, H., Chen, Z., Zhao, F., 2012. Energy from plants and microorganisms: progress in plant-microbial fuel cells. ChemSusChem 5, 1006–1011. Denman, K.L., et al., 2007. Couplings between changes in the climate system and biogeochemistry. In: Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Eds.), Climate Change 2007: The Physical Science Basis. Contribution of Working Goup I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Ishii, S., Hotta, Y., Watanabe, K., 2008. Methanogenesis versus electrogenesis: morphological and phylogenetic comparisons of microbial communities. Biosci. Biotechnol. Biochem. 72 (2), 286–294.

171

Jeon, H., Choi, Y., Kumaran, R., Kim, S., Song, K., Hong, S., Kim, M., Kim, H., 2012. Electrochemical control of methane emission from lake sediment using microbial fuel cells. Bull. Korean Chem. Soc. 33 (2401–2404), 813–824. Kaku, N., Yonezawa, N., Kodama, Y., Watanabe, K., 2008. Plant/microbe cooperation for electricity generation in a rice paddy field. Appl. Microbiol. Biotechnol. 79, 43–49. Pant, D., Van Bogaert, G., Diels, L., Vanbroekhoven, K., 2010. A review of substrates used in microbial fuel cells (mfcs) for sustainable energy production. Bioresour. Technol. 101, 1533–1543. Rismani-Yazdi, H., Carver, S.M., Christy, A.D., Yu, Z., Bibby, K., Peccia, J., Tuovinen, O.H., 2013. Suppression of methanogenesis in cellulose-fed microbial fuel cells in relation to performance, metabolite formation, and microbial population. Bioresour. Technol. 129, 281–288. Rizzo, A., Boano, F., Revelli, R., Ridolfi, L., 2013. Role of water flow in modeling methane emissions from flooded paddy soils. Adv. Water Resour. 52, 261–274. Takanezawa, K., Nishio, K., Kato, S., Hashimoto, K., Watanabe, K., 2010. Factors affecting electric output from rice-paddy microbial fuel cells. Biosci. Biotechnol. Biochem. 74 (6), 1271–1273.