Nitrogen accountancy in space agriculture

Abstract

Food production and pharmaceutical synthesis are posited as essential biotechnologies for facilitating human exploration beyond Earth. These technologies not only offer critical green space and food agency to astronauts but also promise to minimize mass and volume requirements through scalable, modular agriculture within closed-loop systems, offering an advantage over traditional bring-along strategies. Despite these benefits, the prevalent model for evaluating such systems exhibits significant limitations. It lacks comprehensive inventory and mass balance analyses for crop cultivation and life support, and fails to consider the complexities introduced by cultivating multiple crop varieties, which is crucial for enhancing food diversity and nutritional value. Here we expand space agriculture modeling to account for nitrogen dependence across an array of crops and demonstrate our model with experimental fitting of parameters. By adding nitrogen limitations, an extended model can account for potential interruptions in feedstock supply. Furthermore, sensitivity analysis was used to distill key consequential parameters that may be the focus of future experimental efforts.

Fig. 1. Mars-based agriculture overview.

a Scheme for deploying agriculture systems on Mars using In Situ Resource Utilization (ISRU) with an expansion of candidate crops within habitat. The expansion of crop systems includes example groupings of crops and hydroponic reactor logistics. b Breakdown of crop parameters for nine MEC-modeled space crops: water content fractions and edible (harvest index) and inedible fractions of biomass on a dry weight (DW) basis. Values are out of 1. c Carbon content reference values and approximate oxygen and hydrogen fractions typical in cultivated plants; “Other" fraction does not correspond numerically to (d, e) as shown here. d Compiled values for nitrogen and other macronutrient fractions in field-grown plants^,. e Micronutrient fractions in field-grown plants^,.

Fig. 2. Modified energy cascade model calculations.

a MEC model total crop biomass per area, ${\stackrel{⌢}{m}}_{\mathrm{T}}$ (blue), and crop growth rate per area, ${\stackrel{⌢}{\dot{m}}}_B$ (orange), for (from top left) dry bean, lettuce, rice, soybean, tomato, wheat, peanut, sweet potato, and white potato with parameters Φ_γ = 500 μmol_γ m⁻² s⁻¹, c_CO2 = 1200 $\mu {\mathrm{mol}}_{\mathrm{CO}2}{\mathrm{mol}}_{\mathrm{air}}^{-1}$. Crop-specific time points [d_AE]: t_A, canopy closure; t_M, harvest/maturity; t_E, onset of organ formation; t_Q onset of canopy senescence. b MEC model contours of edible biomass accumulation, ${\stackrel{⌢}{m}}_{\mathrm{E}}$, for each crop terminating at t_M, across Φ_γ and c_CO2.

Fig. 3. Hybrid MEC-NP modeling.

a Algorithm by which the growth curves of the MEC and NP models act as limiting factors in a hybrid growth curve. b Example of hybrid growth curve, limited first by nitrogen, then by photosynthetic photon flux, atmospheric CO₂ concentration, or both.

Fig. 4. Nitrogen Productivity Studies.

a Measured areal biomass and MEC model curve calculated with Φ_γ = 225 μmol_γ m⁻² s⁻¹ and c_CO2 = 525 ppm. b Measured fresh weight percentage of total nitrogen in plants over time. A single measurement was performed for each condition at 20 d_AE. c Nitrogen productivity calculated from measured biomass and nitrogen content. Error bars represent propagated error. a–c Time range highlighted in gray is specified harvest time, t_M, plus 5 d. Error bars represent one SD. For each data point, 5 ≤ N ≤ 10. d–f Concentration and change in concentration of nitrogen measured in the NSS over time by experimental condition (deficient, normal, excess). Error bars represent 1 SD. N = 3 for each data point. g–i Photos of lettuce crops during main growth phase at 35 d_AE grown in deficient, normal, and excess nitrogen conditions, respectively.

Fig. 5. Lettuce model comparison.

a Experimental data (circles) and NP model fits (lines) for lettuce biomass predicted by the MEC model and grown in 3 different nitrogen conditions. b Fitting of m_N by values of r, K, and α to experimental mass of N in plant. c Fitting of ${\stackrel{°}{Y}}_{\mathrm{N}}$ by functions for η_N and μ_N to nitrogen productivity calculated from experimental data. d Sensitivity analysis result based on the ranges defined by the fitting procedure. The y − axis denotes the variable of interest wile the x − axis represents the corresponding variable’s index value. e Fitted NP parameter values for MEC baseline and experimental N conditions. r: governing rate in [d⁻¹]; K: limiting value of m_N in [g_N]; α: dimensionless scaling factor; η_N: nitrogen use efficiency in [${\mathrm{g}}_{\mathrm{DW}}{\mathrm{g}}_{\mathrm{N}}^{-1}$]; μ_N relative nitrogen accumulation rate in [d⁻¹].

Fig. 6. MEC and NP Model Projection Comparison.

a NP model fits to MEC growth curves (Φ_γ = 225 mol_γ m⁻² s⁻¹, c_CO2 = 525 ppm) for dry bean, peanut, rice, soybean, sweet potato, tomato, wheat, white potato. b Sensitivity analysis result based on the ranges defined by the fitting procedure. The y − axis denotes the variable of interest wile the x − axis represents the corresponding variable’s index value.