The nitrification compartment (Compartment 3) is a cylindrical packed-bed bioreactor of 7 L operation volume (SNC-Lavalin, Brussels, Belgium and Bioprocess Technologies, Madrid, Spain). It is packed with polystyrene beads and colonized by a co-culture of Nitrosomonas europaea and Nitrobacter winogradsky that grow forming a biofilm. N. europaea is a Ammonia Oxidizing Bacteria (AOB) capable to oxidize ammonia into nitrite. Then, complete nitrification is achieved by the oxidation of nitrite into nitrate by the Nitrite Oxidizing Bacteria (NOB) N. winogradsky. 2 mol of oxygen are needed for the complete oxidation of 1 mol of N-NH₄ ⁺ making it an aerobic process. On top of the central packed-bed section, the reactor has a bottom section, mechanically stirred, where fresh liquid feeding, liquid recirculation inlet and gas sparging take place. There is as well a top section after the packed-bed for gas-liquid separation, and liquid and recirculation outlets. Both top and bottom section hold the instrumentation for on-line monitoring of the compartment (pH, T, pO₂, conductivity). It is operated in a recirculation closed gas-loop mode regulated by means of a mass flow-meter (Bronkhorst, F-202D-FA, Bronkhorst, Ruurlo, Netherlands). pH is measured by means of two sterilizable glass pH probe connected to a pH amplifier (Mettler Toledo, Inpro 3253, Greifensee Switzerland). pO₂ is measured by means of two Clark amperometric sensor (Mettler Toledo, InPro6950i/12/320, Greifensee, Switzerland) (see Figure 2).

The photobioreactor compartment (Compartment 4a) is an 83L external-loop gas lift photobioreactor (Bioengineering Ag, Wald, Switzerland). It is used for Limnospira indica cultivation. The reactor consists of two glass cylindrical tubes with 15 cm diameter and 1.5 m height with an illuminated volume of 55 L. Illumination is provided by 350 halogen lamps (MR16HM, (12 V, 20 W), Sylvania, Mississuaga, Ontario, Canada) around the glass tubes and can be adjusted by the control system of the loop in order to regulate oxygen production. The upper and lower parts of the tubes are connected by U-shaped stainless steel sections that allow improved liquid circulation and heat exchange through an external jacket. The gas phase is injected through the bottom of the right column (riser) and exits from the gas separation section situated at the top of that same column, which creates a difference of density that allows the fluid to return through the left column (downcomer) (see Figure 2).

For continuous operation of the bioreactor during the experiment, the inlet feed medium is pumped by a variable speed gear pump (Lewa, EEC0002S11, Hispania S.L., Leonberg, Germany) followed by a 0.2 µm filtration (MCY4440DFLPH4, Pall Corporation, New York, United States). Outlet flow is regulated with a second variable speed gear pump (Lewa Hispania S.L., EEC0002S11, Leonberg, Germany) Inlet and outlet air supply is regulated and measured by means of three flow-meters and controllers (Bronkhorst, F-202D-FA, Bronkhorst, Ruurlo, Netherlands). pH is measured by means of a conventional sterilizable glass pH probe connected to a pH amplifier (Mettler Toledo, Inpro 3253, Switzerland). Control of pH is performed by addition of NaOH (2 M) or H₂SO₄ (2 M) as required. The composition of the outlet gas is continuously monitored on-line by a IR analyser for CO₂ and a paramagnetic analyser for O₂ (600 series, CAI, United States).

The crew mock-up compartment (Compartment 5) is an animal isolator of 1550 L volume (Hosokawa Micron Ltd., Runcorn, UK). It is divided in three zones: main chamber, transfer air-lock and recirculation loop. It is designed to host the animal crew with artificial light at periods of 12–12 h day-night cycles. It includes six 250 mm round glove ports at the front of the isolator to perform the animal support operations.

In the main chamber, an overpressure of 200 Pa is maintained by pressure control through two pressure transmitters (Rosemount, 2051 CG1,Shakopee, Minnesota, United States). An external 250 L buffer tank is connected to the isolator to compensate the effects of atmospheric pressure changes. Temperature and humidity in the isolator are monitored by the use of two transmitters (Vaisala, HMT337, Helsinki, Finland) and controlled at 22°C and 55%, respectively. A gas recirculation loop ensures a homogenous distribution of air in the isolator using a recirculation flow of 90 m³·h⁻¹. An active carbon filter is installed (Bi-On® ACPA, Bioconservación, Barcelona, Spain) in order to avoid accumulation of contaminants during long operational periods. CO₂ and O₂ are measured on-line through an IR and paramagnetic analyzers respectively (SICK-Maihak, GSM810, Hamburg, Germany). Finally, the transfer airlock operates with an independent pressure control to allow the entry and exit of materials (cages, food, bedding, etc.) without affecting the main isolator variables.

Three Wistar rats (Charles River Laboratories, Domaine des Oncins, France) with an age of 12 weeks were selected as the suitable crew model for inhabiting the isolator and running the experiments. The rational for the animals selection has been described previously (Alemany et al., 2019). A different group of rats is used for each experimental condition where different liquid flow rates of the liquid loop are tested. The rats are fed with an ab libitum diet (Teckad 2914, Envigo, United States). Weekly monitoring of the rats weight is performed.

Nitrobacter winogradsky and Nitrosomonas europaea were provided by DSMZ, (Germany) for nitrifying bioreactor culture. Limnospira indica PCC 8005 was provided by SCK CEN (Mol, Belgium) for PBR cultivation. The strains were grown axenically during the experiment in their corresponding bioreactor. Culture media was a combination of a modified Zarrouk medium (Cogne et al., 2003) and the medium defined in the literature for nitrifying cultures (Pérez et al., 2004): 1.358 g·L⁻¹ of (NH₄)₂SO₄, 1 g·L⁻¹ of K₂SO₄, 1 g·L⁻¹ of NaCl, 0.816 g·L⁻¹ of NaHCO₃, 0.71 of Na2HPO4, 0.68 of KH2PO4, 0.5 of K2HPO4, 0.1 g·L⁻¹ of MgSO₄·7H₂O, 0.18 g·L⁻¹ of (NH₄)₆Mo₇O₂·4H₂O, 0.08 g·L⁻¹ EDTA-2Na·2H₂O, 0.01 g·L⁻¹ of FeSO₄·7H₂O, 0.04 g·L⁻¹ of CaCl₂·2H₂O, 1.81 mg·L⁻¹ of MnCl₂·4H₂O, 0.22 mg·L⁻¹ of ZnSO₄·7H₂O, and 0.079 mg·L⁻¹ of CuSO₄·5H₂O.

A liquid stream was constantly supplied from a reservoir tank to the inlet of the nitrifying bioreactor. Liquid effluent from nitrifying bioreactor was constantly processed by a three-step filtration to maintain axenicity in both bioreactors. This was a critical operation to avoid any cells detached from Compartment 3 biofilm reaching Compartment 4a, with the associated contamination effects. The effluent from nitrifying bioreactor was pumped to the inlet of the PBR with a gear pump (Lewa, EEC0002S11, Hispania S.L., Leonberg, Germany). A variable peristaltic pump (Watson-Marlow 323Du, Watson Marlow Fluid Technology Group, Wilmington, United States) at the nitrifying bioreactor outlet was used to keep an overpressure in the liquid connection line at 200 mbarg. Finally, the PBR effluent was continuously collected in a harvesting tank (see Figure 2).

A set of different liquid flow rates corresponding to three NH₄ ⁺ loads (ppm·d⁻¹) into compartment 3 were initially tested: 10, 20 and 30 L·d⁻¹. A combination of testing conditions was defined by setting two illumination levels (q _o ): 120 and 285 W·m⁻². So, a total of six conditions were tested. Operational conditions in the PBR were stablished as follows: pH was set at nominal value of 8.5, CO₂ was constantly injected at 1% composition in the air inlet at a total gas flow of 2.8 L·min⁻¹ and temperature was set at 36°C. Conditions in Compartment 3 were as follows: pH was controlled at 8.1 by the addition of H₂SO₄ (0.1 M) and KOH (1.9 M), pO2 was maintained at 80% thanks to the injection of external pure O₂ into the recirculation closed gas-loop, a recirculation liquid flow between the top and bottom section of the bioreactor was set at 4.5 L·h⁻¹ to improve the homogeneity and temperature was set at 30°C.

The animal isolator was connected to the PBR at the entrance and output of the compartment so that closed gas loop was established by using a diaphragm vacuum pump (GAST, 22D1180-202-1005, Benton Harbor, Michigan, United States).

After completion of previous tests, involving Compartments 3 and 4a in the liquid phase, a new series of experiments was carried out. Now, the animal isolator was connected in the gas phase to the PBR, and liquid connection between nitrifying bioreactor and PBR was maintained (see Figure 3). Therefore, the three compartments were connected. In this scenario, the liquid flow-rate was increased to 20, 30 and 40 L·d⁻¹ and the N-NH₄ ⁺ concentration was maintained at 300 ppm at 20 and 30 L·d⁻¹, while it was decreased to 240 ppm at 40 L·d⁻¹. The main control variable of these experiments was the light intensity of the PBR. The control system adjusted the light intensity to maintain an O₂ concentration set-point in the animal isolator and compensating the animals oxygen needs along the day/night cycles. For each liquid loop flow rate condition, the O₂ set-point in the isolator was sequentially changed following this sequence: 21–19-20–21%. The rationale between this operation schedule was to test the capacity of the control system to adjust the illumination conditions in the PBR each time a new O₂ set-point was fixed in the animal isolator. Additionally, the experimental period at a given flow rate between the nitrifying bioreactor and PBR coincided with a new group of rats in the isolator.

Off-line monitoring of biomass in the PBR was carried out by cell dry weight (CDW) determination and OD measurement. For CDW, 25–50 ml of culture broth were filtered through a 47 mm-glass microfiber (GMFC-52047, Scharlab, Barcelona, Spain), dried at 105°C until constant weight. OD measurement was performed at 750 nm in an optical spectrophotometer (UV-VIS DR6000, Hach, Düsseldorf, Germany).

Filtered fractions for biomass determination in the PBR were used for NO₃ ⁻ and TIC analysis. Samples were diluted 1:100 with milli-Q H₂O and NO₃ ⁻ was analysed by spectrophotometry with LCK 339 kits (Hach, Weinheim, Germany). For TIC determination a 1:5 dilution with milli-Q H2O was required. 1.5 ml of sample were loaded to a multi N/C2100S equipment (Analytik Jena, Jena Germany). 200 µL from the sample were injected into the acidic TIC reactor and 0.5 ml of H₃PO₄ at 10% were added to the chamber. CO₂ generated from the acidification was detected by the NDIR detector and quantified by triplicate.

In Compartment 3, NO₃ ⁻ at the bioreactor outlet was analysed following the same method as in Compartment 4a. NH₄ ⁺ and NO₂ ⁻ were analysed with LCK 303/304 and LCK 342 (Hach, Weinheim, Germany).

Outlet gas composition from the PBR was measured on-line by an IR analyser for CO₂ coupled to a paramagnetic analyser for O₂ (600 series CAI, Orange, California, United States). In the case of Compartment 4a and 3 liquid connection, oxygen volumetric production rate (g·L⁻¹·h⁻¹) was calculated on-line based on continuous monitoring of outlet gas composition in the PBR: r O 2 = ( ( F in − F CO 2 ) ⋅ 0.2089 − ( F out ⋅ O 2 out / 100 ) ) / ( 22.4 ⋅ M W O 2 ⋅ V R ) (1) where F _in is the inlet air flow rate, F _CO2 is the inlet CO₂ flow rate, F _out is the total outlet gas flow rate, O _2out is the outlet mole fraction of O₂ in %, MW_O2 is the molecular weight of O₂ (32 g·mol⁻¹). The flow rates are expressed at normal conditions (T = 273.15 K; p = 1 atm). In the case of the second integration step (Compartments 5 and 4a connected in the gas phase and 3 and 4a connected in the liquid phase), the inlet gas composition of PBR is analysed by means of the gas analyser in compartment 5. Then, oxygen volumetric production rate in PBR is calculated as follows: r O 2 = ( ( F i n ⋅ O 2 i n − ( F o u t ⋅ O 2 o u t / 100 ) ) / ( 22.4 ⋅ M W O 2 ⋅ V R ) (2) where O _2in is the inlet mole fraction of O₂ (%) in the PBR.

PBR CO₂ consumption was calculated for each condition considering on-line gas analysis and off-line TIC concentration in the steady state. In such situation, no carbon accumulation takes place and CO₂ can be calculated in g·L⁻¹·h⁻¹ by means of the following mass balance: r CO 2 = C O 2 L o u t l e t − ( C O 2 g a s   t r a n s f e r ) (3) r C O 2 = C L o u t ⋅ D + F i n ( X C O U T − X C I N ) / 22.4 ⋅ M W C O 2 (4) where C _LOUT is the carbon concentration (g·L⁻¹) in the liquid outlet converted into CO₂ by means of molecular weight, D is the dilution rate (h⁻¹), F _in (L·h⁻¹) is the bioreactor gas flow rate, X _COUT is the molar fraction of CO₂ in the outlet gas, X _CIN is the molar fraction of CO₂ in the inlet gas and MW_CO2 is the molecular weight of O₂ (44 g·mol⁻¹). When PBR and animal compartment where connected in the gas phase, X _CIN was the corresponding CO₂ concentration in the animal compartment.

Considering that O₂ was the controlled variable in the animal compartment, O₂ consumption in animal compartment was assumed to be the same as O₂ production in the PBR. In the case of CO₂ consumption (g·h⁻¹) of the rats, calculation was as follows: r C O 2 = F i n ( X C O U T 5 − X C I N 5 ) / 22.4 ⋅ M W C O 2 (5) where X _COUT5 and X _CIN5 are the molar fractions of CO₂ at the outlet and inlet of animal compartment respectively.

One-way ANOVA analysis assuming differences in variances was performed to assess variability of O₂ generation of the system. This statistical approach was also used to study the differences in body weight of the animals. A significance value of p < 0.05 was considered in all cases.

Nitrifying bioreactor and PBR are connected in the liquid phase in continuous operation mode for a total period of 276 days. The nitrification compartment provides the necessary nitrogen source for the growth of cyanobacteria (Clauwaert et al., 2017) A total of six different conditions were tested corresponding to three different ammonium loads: 435, 870, and 1300 ppm·d⁻¹ into Compartment 3 at 10, 20, and 30 L·d⁻¹ respectively. These flow rate conditions are combined with two levels for q _o in the PBR: 120 and 285 W·m⁻².

Figures 4D–F presents the operation profile of the PBR in terms of oxygen production and biomass cell density for the complete series of experimgents. In all the tested conditions steady state in the PBR is achieved as it is maintained with stable parameters for a period of at least 3 Hydraulic Residence Times (HRT) for each condition. The highest biomass concentration is obtained for 10 L·d⁻¹ condition (Figure 4E with values between 2.1 and 2.2 g·L⁻¹, while the lowest cell density is reached at 30 L·d⁻¹ (Figure 4F) with values between 0.9 and 1.1 g·L⁻¹. On the contrary, the highest biomass productivity results from 30 L·d⁻¹ conditions, followed by 20 and 10 L·d⁻¹ respectively. On-line oxygen production monitoring is also presented in Figure 4. The dynamics of the system can be observed for each q _o increase followed by a stabilisation profile. The highest oxygen production is reached at 30 L·d⁻¹ condition (between 2.1 and 2.7 g·h⁻¹) while the lowest production occurs at 10 L·d⁻¹ (1.4–1.7 g·h⁻¹). The same pattern is also observed from specific production values (q _O2 ). Detailed data are presented in Table 1. Figure 5 reflects the influence of light and flow in the performance of the PBR in terms of oxygen and biomass production. A Pareto’s ANOVA was performed to study the influence of those factors. The liquid flow is the factor showing the greatest effect on biomass and oxygen production. The influence of flow on biomass concentration is 85%, while it is 69% for oxygen production. In contrast, light intensity shows little influence in biomass concentration (4%), while its contribution increases up to 21% regarding oxygen production. The interaction effect between factors in both cases is at 11 and 10% respectively.

Values from carbon dioxide consumption in the PBR are also presented in Table 1, showing that carbon dioxide consumption by L. indica increases in correspondence with oxygen production. 2.3 g CO₂·h⁻¹ are consumed when the cells are exposed to the maximum q _o of 285 W·m⁻² at a continuous flow of 30 L·d⁻¹. The minimum carbon dioxide consumption is found at the lowest flow rate of 10 L·d⁻¹ and minimum light intensity (120 W·m⁻²). Photosynthetic quotient ranges between 1.3 and 1.8 molO₂·molCO₂ ⁻¹, which is in agreement with theoretical values between 1.2 and 1.4 when NO₃ ⁻ is used as nitrogen source according to the stoichiometry (Cornet, 2007).

In terms of N balance, all the nitrogen supplied to compartment 4 is in form of NO₃ ⁻ previously converted from NH₄ ⁺ in compartment 3. The ammonium conversion is maintained at 99% during most of the experiment (Figures 4A–C). Only minor transitory peaks occur because of operation activities in the bioreactor focused on biomass removal with an external circulation system coupled with filtration for biomass retention. During these short periods the maximum N-NH₄ ⁺ and N-NO₂ ^- concentration in the PBR is 0.1 and 1 ppm respectively (data not shown) and therefore can be considered negligible. The outlet N-NO₃ ^- concentration from the nitrifying bioreactor is between 275 and 300 N-ppm (Table 2). Yield values are higher than 90%, whilst conversion of ammonium is at 99% during all steady states (transitory peaks are not considered). The yield is higher (96–99%) with the lowest ammonium load of 435 N-ppm·d⁻¹, but the yield decreases (91–93%) when the load increases to 870 and 1300 ppm·d⁻¹. Inlet N-NH₄ ⁺ into nitrifying bioreactor was maintained constant at 300 N-ppm for all conditions, resulting in different ammonium loads depending on the liquid flow rate. The output from nitrifying bioreactor was the inlet of the PBR, so Y _N/X was calculated considering that input. Results show that Y _N/X is maintained constant no matter the experimental condition used. 0.10–0.11 g of N-NO₃ ^- are needed to generate 1 g of biomass. In terms of oxygen as a product, Y _N/O2 is between 0.06 and 0.07 g·g⁻¹ in all tested conditions. The yields obtained fit with the reaction stoichiometry described in previous works (Cornet et al., 1998).

The obtained results prove that liquid connection between nitrifying and photosynthetic bioreactors was successful. Nitrogen requirements of L. indica were met thanks to the nitrate supplied by the nitrifying bioreactor after a complete nitrification process performed by the co-culture of N. europaea and N. winogradsky. Indeed the operation of the photobioreactor was very stable and robust in the range of 10–30 L·d⁻¹ and q _o of 120 W·m⁻² and 285 W·m⁻². Stability for each condition was maintained at steady-state to obtain reliable data. In terms of nitrification activity, the system has been proved to be robust enough to maintain the ammonium conversion at 100% for 200 days no matter the load. There were minor accumulation peaks that were rapidly recovered and had no impact on photosynthetic activity of L. indica. These peaks were most likely caused after biomass removal from the packed-bed reducing the amount of active biomass. Then, once the previous active levels of biomass were reached, the nitrification activity recovered to 100%. Additionally, residual N-NO₂ ^- concentration in the PBR supernatant was below the limit for drinking water (World Health Ogranization, 2017).

The effect of light and dilution rate have been proved to contribute in a different degree to the performance of the PBR. Liquid flow, and consequently dilution rate, is the factor having the greatest effect on biomass and oxygen productivity. It is an indication that what really influences the behaviour of cyanobacteria is light availability rather than the amount of light. Figure 5 shows that light availability was directly proportional to the amount of light, but inversely proportional to the amount of cells. Results in terms of oxygen production were very relevant for the definition of the next integration step including the gas connection of the animal compartment to the PBR. Animal compartment in the MELiSSA loop aims to mimic the human activity. In this case the rats are the mock crew. As reported previously (Alemany et al., 2019), the oxygen needs for 1 rat is 0.48–0.7 g·h⁻¹. The analysis of the obtained results for Compartment 3 and 4a liquid connection shows that depending on the liquid flow between bioreactors, different number of rats could be used in the experimental connection with the animal compartment. When working at 10, 20 or 30 L·d⁻¹, the maximum oxygen productions achieved in the steady state were 1.6, 2.44 and 2.76 g·h⁻¹ respectively. Under this scenario, the maximum number of rats that could be selected in the animal compartment is 4 with a theoretical minimum and maximum oxygen demand of 1.92 and 2.8 g·h⁻¹ respectively. Nevertheless, under this scenario the system would be at the maximum capacity to meet the maximum oxygen requirements. It is for this reason that 3 rats were selected for the experiments where nitrifying bioreactor and the PBR were connected in the liquid phase, and the PBR was connected through the gas phase with animal compartment. This solution provides some margin to the system to react against potential perturbations or changes imposed by the user in the gas phase composition of the animal isolator. The selection of 3 rats imposes a theoretical oxygen demand into the system of 1.44–2.1 g·h⁻¹ for minimum and maximum values. In order to meet this demand, the minimum and maximum carbon dioxide needs of the PBR would be 1.32 and 1.93 g·h⁻¹ respectively assuming a middle range experimental Q _P of 1.5 molO₂·molCO₂ ⁻¹ (Table 1). The animal compartment would provide 1.78–2.60 g·h⁻¹ of carbon dioxide if a theoretical RQ of 0.9 molCO₂·molO₂ ⁻¹ is considered (McGregor and Lee, 1998). So, the carbon needs of the PBR would be met. In terms of N needs, no lack of nitrate is expected in the PBR to meet the O₂ requirements if the same N-NH₄ ⁺ loads in nitrifying bioreactor are maintained. It has been demonstrated that according to yields displayed in Table 2, the amount of N-NO₃ ⁻¹ needed in the PBR would be between 0.09 and 0.12 g·h⁻¹. Hence, any of the 3 tested loads conditions in the nitrifying bioreactor are acceptable.

According to previous rationale, the best hydrodynamic conditions for the next integration phase would be selecting a liquid flow of 20 and 30 L·d⁻¹ between nitrifying bioreactor and the PBR. The liquid flow of 10 L·d⁻¹ should be discarded as the system could be in the limit in terms of oxygen production, especially during high activity periods of the rats. Additionally, the exploration of a wider experimental liquid flow domain could provide of relevant information for further optimisation of the MELiSSA loop. This is why a liquid flow of 40 L·d⁻¹ was also selected as an experimental condition for the second phase tests.

After a 110 days experiment, where the PBR and the animal isolator were connected in the gas phase, it has been demonstrated that the system is able to cover the oxygen needs of the rats and carbon needs of L. indica while operating in a closed gas loop. This condition is in agreement with previous results (Alemany et al., 2019). The added value of the current work is that nitrogen requirements of L. indica to cover the oxygen demand, have been covered by the continuous operation of a nitrifying bioreactor. This is particularly relevant since it represents a further step in the completion of the overall closed-life support system.

Focusing on the gas phase dynamics, the system could maintain a constant oxygen concentration in the animal isolator. Oscillations (±0.04%) in the measured value are the direct cause of circadian activity of the rats. In all oxygen set-points (21,19, 20%), the activity of the rats was higher during night periods because of its photoperiod (Antle and Mistlberger, 2005). In all cases the RQ of the rats ranged between 0.9 and 1.0 molO2·molCO2⁻¹, which is consistent with the experimental data reported in the literature (McGregor and Lee, 1998; Jørgensen et al., 2010).

Transition periods, where oxygen set point was increased, took less than 10 h in two groups of rats and less than 20 h in the last group. This proves the fast response of the system to adjust to a change in the parameters of the crew compartment. It should be reminded that the aim of Compartment 4a (PBR) in the loop is to respond to oxygen demand variations thanks to its fast dynamics. The explanation of a higher time period for the transition in the last group of rats is the higher oxygen demand of the rats. It was demonstrated by the average oxygen produced in the PBR for each rat’s group (Figure 7), which increased up to 2.1 g·h⁻¹ in that case. Indeed, the body weight gain of the rats, which could be an indication of the physiological activity of the rats, has been correlated with the oxygen demand of the system. In contrast, the transition periods for the first group of rats was not lower than in the second group, even a lower oxygen demand (1.5 vs. 1.9 g·h⁻¹). This effect could be explained by the differences in the PBR capabilities. In the first case, the biomass concentration in the PBR was higher than in the second case (1.24 g·L-1 vs. 0.98 g·L⁻¹), potentially limiting the light availability to the cells and decreasing the photosynthetic activity at the same light irradiance. This hypothesis is plausible considering the average q o ¯ , which is higher in the first group (79 W·m⁻²) than in the second the group of rats (75 W·m⁻²) even the lower oxygen need. From this observation, it is confirmed that the cells were in a “physical limitation regime” where the light is the limitation parameter of the growth kinetics (Cornet, 2007). The higher light intensity observed in the last group could be explained by the high oxygen demand.

In terms of carbon dioxide concentration, the levels were always maintained below 15,000 ppm. This was below the acceptance limit of 3% for rats (Krohn and Hansen, 2002; Krohn et al., 2003). However, future experiments in the system should consider lowering this level and make it closer to the limits accepted for humans (<0.5%). This point will address in future integration steps of the MELiSSA loop.

Results from liquid connection experiments between Compartment 3 and 4a showed that maximum oxygen production in the PBR was around 2.4–2.8 g·h⁻¹ with a flow rate of 20–30 L·d⁻¹. However, during the second phase of the experiments (liquid connection between Compartment 3 and 4a and gas connection between Compartment 4a and 5) the system has proved to be over the previous determined limits. Light irradiance was maintained around 30% of its capacity with enough margin to meet the oxygen demand during transition periods. Indeed, the maximum oxygen production in those transition phases exceeded by 25% the reported values of the first experimental phase. This difference evidences that transition phases should be treated differently than steady state. Comparing biomass concentration at maximum light intensity condition in both experimental phases shows that cell density is lower in the second phase, leading to a higher light availability. From a biological perspective this effect might be explained by the molecular composition of the cells. The photosynthetic capacity of L. indica is influenced by the pigment content. When light availability is increased, it takes some hours (Tomaselli et al., 1997) to reduce the pigment content to protect the cells against potential photoinhibition (Muramatsu and Hihara, 2012; Weiwen Zhang, 2018). It is during that transition that the closed loop benefits from the still high pigment content to exploit the maximum photosynthetic activity of the cells. It can be concluded that the capacity of the system in transition steps cannot be extrapolated at long-term, when the system would evolve to reach a new steady state with lower oxygen generation capacity.

Regarding nitrogen needs of L. indica to meet the oxygen requirements of the animal crew, it has been demonstrated that nitrifying bioreactor is capable to fulfil them. Indeed, nitrate concentration in the PBR outlet was always kept higher than 20 N-ppm, which is considered a potentially limiting concentration (Depraetere et al., 2015). This scenario avoided any metabolic drift in L. indica, allowing an optimal performance in terms of air revitalisation. However, the limiting capacity of the nitrifying bioreactor was reached in the last condition (40 L·d⁻¹), so ammonium load had to be reduced in order to avoid nitrogen species accumulation. One of the outputs from this work is that the maximum capacity of nitrifying bioreactor in the MELiSSA loop is 1400 ppm·d⁻¹, which should be taken into consideration when it comes to introduce the urea in the loop. Once the ammonium load was decreased, the full nitrification capacity was fully recovered. Although complete nitrification was dominant during most of the test, three partial nitrification events occurred. Nitrite and ammonium reached the PBR resulting in different consequences. Whilst nitrite was present at detectable levels, no ammonium was detected at the PBR outlet. This result shows that L. indica can use ammonium as nitrogen source as previously reported in the literature (Sachdeva et al., 2018). Even nitrite concentration in the PBR reached up to 78 N-ppm during peaks periods, no irreversible effects were detected in L. indica culture. In the first and second accumulation events (78 and 58 N-ppm), the presence of nitrite had an effect on photosynthetic activity. The control system adjusted the lights up to the maximum q _o of 285 W·m⁻² during 2–3 h in order to cover the oxygen needs of the animal compartment. This behaviour was not observed in the last accumulation event (36 N-ppm) suggesting that a concentration higher than 40 N-ppm of nitrite have an inhibitory effect on L. indica photosynthetic activity. Nevertheless, this event shows the robustness of the system in such a wide range of conditions and in front of potential perturbations. The main function of air revitalisation in Compartment 4a was never at risk and oxygen needs of the rats were fulfilled during 110 days of continuous operation.