2.1. Paddy Field
An existing paddy field in West Kalimantan, Indonesia (0.919215 N, 109.468182 E; elevation 100 m above sea level) was used for the experiments. This paddy field was cultivated using a common agricultural practice in that area. The rice paddy (
Oriza sativa L. var Inpari 30 Ciherang Sub 1) was transplanted twice per year which are in May and November (
Table 1). The rice paddy growth phases are shown in
Figure 1. Prior to transplantation, the rice paddy seeds were grown in a nursery site for 20–30 days. Afterward, the rice paddy were manually transplanted by farmers in the soil in lines with a distance between 15 cm and 25 cm. During the seedling period, soil was prepared by ploughing it with a hand tractor. The ploughing was preceded by flowing water into the paddy field (about 7 days) to make the soil softer. The ploughing was done per plot and in line with water flow direction from both up- and downstream. The water was kept flowing between 5–10 cm above the ground in the paddy field until ripening period. The soil was dried in the last 2–3 weeks before harvest.
Rice paddy maintenance includes several activities such as fertilization, pest control, weeding, and water management. The fertilization was applied twice during one crop season. The first fertilization was about 7 days before transplantation day (TD) after the ploughing and the second fertilization was 7 days after TD. In the first fertilization, three kinds of fertilizer were applied together, which are urea (PT.Pupuk Sriwidjaja, Palembang, Indonesia), NPK Phonska (PT.Petrokimia Gresik, Gresik, Indonesia) and Petroganik (PT.Pupuk Indonesia (PERSERO) Group, Cikampek, Indonesia). In the second fertilization, only urea and NPK Phonska were applied in the paddy field. The application of fertilizer followed the dosage stated on the package. During the fertilization, water inlet and outlet was closed for 2–3 days so that the fertilizers dissolved and seeped into the soil. After that the water was slowly put back into the paddy field with a small flow.
The pest control was done by using molluscicides and insecticides. The molluscicides (Keong Tox, PT.Santani Sejahtera, Medan, Indonesia) was applied 1 day before the first fertilization to control the snail, Pomacea spp., (known as keong mas in Indonesian). The insecticides were distributed by spraying a mixture of Mipcindo 50WP (PT. Inti Everspring Indonesia, Mangunreja-Serang, Indonesia) and Imidacloprid 96TC (PT.Catur Agro Dinamika, Pamulang Tanggerang Selatan, Indonesia). The mixing ratio between Mipcindo and Imidacloprid was 3:1 (tablespoon) for 25 L of water. The insecticide spraying was done three times during the crop season which were 12–14 days after TD, 21–25 days after TD and 40 days after TD. In addition, if walang sangit (Leptocorisa oratorius) and wereng (a general term used to designate plant-liquid sucking insects from the order Hemiptera) were still seen a lot, another insecticide spraying was done (at 60 days after TD) with a 3:2 mixing ratio.
Water management plays an important role in rice paddy cultivation. Before applying the molluscicides water was reduced to about 0.5–1 cm above the ground. This water level was kept until the first fertilization and the transplantation day. Three day after the TD water level was slowly increased to the normal level. The same mechanism was done during the second fertilization. Apart from these period, the water level was kept on the normal level (5–10 cm above the soil) following the rice paddy growth phase. The continuous flooding also functions to reduce weeds. In the studied paddy field, weeds growth was hampered because the fields were well ploughed. Therefore, weeding was only manually extracted if they were spotted. The water flow was stopped 2–3 weeks before harvest day.
The Inpari 30 Ciherang Sub 1 paddy rice is well-known as a flood-tolerant variety. It can tolerate submergence for up to 14 days [
32]. According to Indonesian Center for Rice Research, Indonesian Agency for Agricultural Research and Development, Ministry of Agriculture Republic of Indonesia (
Balai Besar Penelitian Tanaman Padi,
Badan Penelitian dan Pengembangan Pertanian,
Kementerian Pertanian Republik Indonesia), this variety has an average productivity of 7.2 ton/ha and the harvest time is 111 day after seedling establishment [
33].
2.2. Electricity Generation
On 28 October 2017, three tubular plant microbial fuel cells (tubular plant-MFCs) were installed in the paddy field (
Figure 2). The plant-MFC reactors were installed in lines next to each other from west to east (
Supplementary Information Figure S1). The distance between reactors was 30 cm. The tubular plant-MFC was manufactured by Plant-e (Wageningen, The Netherlands) according to a similar design used in [
17]. As base a transparent silicon tube (VMQ silicone, 12/16 mm inner/outer diameter; rubbermagazijn.nl, Zoetermeer, The Netherlands) was used to supply oxygen. Consequently the cathode felt, the spacer and finally the anode felt was wrapped around this tube. Both the anode and the cathode were made of carbon felt (KFA-5 mm thickness, SGL Carbon GmbH, Bonn, Germany). The electrode length of each tubular plant-MFC was 1 m and the width for the anode and the cathode was 19 cm and 10 cm, respectively. The spacer was made from non-conductive materials to prevent short circuiting, but is otherwise completely permeable (air filter cloth DA/290, DACT Filter- & Milieutechniek, Kerkrade, The Netherlands). Titanium wire (grade 2, 0.5 mm; Titaniumshop BV, Kampen, The Netherlands) was used as current collector both in the anode and the cathode side. The current collector was tied and wrapped around the cathode and the anode as shown in
Figure 2.
The tubular plant-MFCs were installed manually by hoeing the top soil of the paddy field prior to the transplantation. The tubular plant-MFCs were placed about 10–15 cm below the soil (
Figure 3). Both ends of the silicon tubes were bent down with the open hole facing the ground to avoid rainwater from getting into the tube, which would affect the air, i.e., oxygen, supply to the cathode.
2.3. Measurements and Analysis
Plant-MFC reactor performances were evaluated based on anode potential, cathode potential and cell potential, in combination with different applied external loads. In the first three crop seasons, data were irregularly measured with a digital multimeter (Fluke, Fluke Europe B.V., Eindhoven, The Netherlands). All potentials were measured and reported against a 3M KCl Ag/AgCl reference electrode (QIS, Oosterhout, Netherlands). The anode reference electrode was fixed on the anode surface with a cable tie and the cathode reference electrode was inserted in the tube between the cathode and spacer. In the fourth crop season, data (the anode potential, the cathode potential, the cell potential and temperature) were automatically logged using LoRa technology (AE Sensors B.V, Dordrecht, The Netherlands). The temperatures were measured 50 cm above the ground using the a temperature sensor integrated in the same LoRa data acquisition technology.
Rainfall data were obtained from the two nearest weather station, which are the Mempawah Climatology Station (0.07500 N, 109.19000 E; 2 m above sea level) about 100 km to the south-southwest of the research area and the Paloh Climatology Station (1.74000 N, 109.30000 E; 15 m above sea level) about 100 km north-northwest of the research area.
Soil samples were collected (on 30 June 2018) from six different locations for microbial community analysis (
Figure 4). Samples were grouped into three: Group I (Samples A and C) was soil that attached on the anode from mid plant-MFC; Group II (samples B and D) was soil that attached on the anode from end of the plant-MFC; and Group III (samples E and F) was from soil with 2 m distance from plant-MFC 1 and 2. After collection, samples were kept in a 30 mL-tube container and keep in a 4 °C fridge. The next day samples were transported for 48 h with a cool-ice box for DNA extraction to Genetika Lab, Jakarta (PT. Genetika Science Indonesia, Jakarta, Indonesia), a partner company of 1st BASE Axil Scientific Pte Ltd., Singapore.
Sequencing steps were performed by 1st BASE [
34] as follows: universal primers that targeted the V3V4 regions were used for amplification. The quantity and quality of the PCR product that targeted the V3V4 regions were measured using Tapestation 4200, picogreen and nanodrop. All the samples passed the QC measurement and proceed straight to a library preparation. The libraries were prepared using Illumina 16s metagenomics library prep kit and their quality and quantity were determine using Agilent Tapestation 4200, Picogreen and qPCR. These libraries were then pooled according to the protocol recommended by the Illumina and proceed straight to sequencing using MiSeq platform at 2x301PE format by 1st BASE Axil Scientific Pte Ltd., Singapore.
During the harvest time from the third and the fourth crop season, aboveground biomass was collected from 12 different locations. Three locations (above plant-MFC) were a 1-square-meter area above each plant-MFC reactors and another six location (1 metre from plant-MFC) were a-1-square-metre area both to the north and to the south from each reactor (
Supplementary Information Figure S2). In the fourth crop season, the biomass was only collected from the north part of the plant-MFC reactors. Biomass was cut about 5 cm from the ground. After collection, wet biomass was weighed using a manual 10-kg counterweights scale with a precision of 100 g.
2.4. Long Distance Data Acquisition
Eight wireless voltage meters (AE Sensors, Dordrecht, The Netherlands) were installed for data acquisition on 14 February 2019 between crop season 3 and 4. Each sensor has three inputs; one common ground and two completely differential inputs. For each installed tube, a reference sensor (3M KCl Ag/AgCl reference electrode, QIS, Oosterhout, The Netherlands) was installed which was connected to the common ground. Both the anode and cathode were measured completely differential against the reference input. Each voltage meter is built into a watertight junction box and is powered by two AA batteries, the projected operating time at Borneo conditions with four measurements per hour is at least one year. The voltage meters have a RM186-SM module (LoRa/BLE 868MHz LoRa EU, Laird Connectivity, London, UK) implemented and can be accessed by Bluetooth through the Laird Toolkit app to check it’s status and connectivity (
Figure 5). Through the same module, data can be sent through the Long Range Wide Area (LoRaWAN) network. Since the voltage sensors, including the Laird module, were manufactured in The Netherlands, they used the EU LoRaWAN protocol which cannot be used outside of the EU. Moreover, there was no LoRaWAN network installed yet at the research site. We therefore also installed a gateway (Laird RG186 LoRa Gateway, Laird Connectivity) on site that was connected through the locally installed WiFi network dongle (ZTE MIFI Router, InternationalSIM, Terborg, The Netherlands) (
Figure 5). This WiFi network was finally made possible through the available 3G mobile phone network (IM3 Ooredoo, PT. Indosat Tbk, Jakarta, Indonesia). Due to the lack of available electricity on site, the whole system is powered by a locally installed battery system (100 Wp solar panel, 100 AH 12 V Rechargeable Sealed Lead Acid Battery; PWM20 Solar Charge Controller) on solar panels. Data is temporary stored by the sensors and sent in CSV-format on a daily basis to pre-defined email addresses of the involved researchers in The Netherlands. This data logging equipment was co-designed and/or manufactured with support by Plant-e B.V (Wageningen, The Netherlands) & AE Sensors B.V (Dordrecht, The Netherlands).
2.5. Calculation
Current generation was calculated according to Equation (1):
where
I is the current production in Ampere (A),
V is the cell potential in Volt (V) and
R is the applied load in Ohm (Ω).
Power generation was calculated according to Equation (2) or Equation (3):
where
P is the power output in Watt (W),
V is the cell potential in Volt (V),
I is the current production in Ampere (A) and
R is the applied load in Ohm (Ω).
Current density (A/m2) and Power density (W/m2) were obtained by dividing the current production and the power output with plant growth area (PGA). The PGA was 0.0585 m2. Energy density (Wh/m2) was obtained by multiplying the power density with the time it was generated (h).
Internal resistance (
Rint) is calculated according to Equation (4) [
4]:
where
EOCP is the open cell potential in V,
Ecell is the measured cell potential in V,
i is the current density in A/m
2 and
Rint is the internal resistance in Ω·m
2.
Internal resistance is the accumulation of resistances in the Plant-MFC system due to anode overpotential (ŋ
an), cathode over potential (ŋ
cath) and membrane potential (Em) [
4]. Cathode internal resistance (
Rcath) and anode internal resistance (
Ran) are calculated according to Equations (5) and (6), respectively:
where
EOCP,cath is the cathode potential at open cell potential (V),
EOCP,an is the anode potential at open cell potential (V),
Ecath is the measured cathode potential (V) and Ean is the measured anode potential (V). The internal resistance is reported normalized to the PGA.
In this study, the theoretical cathodic reduction reaction potential of oxygen to water (−0.494V vs. Ag/AgCl) is considered as the cathode potential at open cell condition (
EOCP,cath). The theoretical anodic oxidation reaction of acetate (0.6V vs. Ag/AgCl) is used as the anode potential at open condition (
EOCP,an). Thus, the open cell potential is 1.094V vs. Ag/AgCl [
2,
3].
The maximum power generation was evaluated by a polarization technique. Polarization was conducted on 29 November 2017. First, the plant-MFCs were operated under open cell conditions (the external load was disconnected) for 10 min. After that the external load was reconnected and changed every 10 min from high to low in order. The external loads used for the polarization were 1000, 470, 220, 100 and 10 Ohm. Cell potential generated from the plant-MFCs for every operated external load were measured with a multimeter after 10 min of operation. From these cell potential, current generation and power generation were calculated according to Equations (1) and (2) and normalized to PGA. Note that this is not an indicator for the actual long term performance of the Plant-MFC since this method does allow to take capacitive current into account.