Eighteen individual aquaponic systems were constructed at the Aquaculture Research Lab, Purdue University. The following are the components of each system; an aquaculture tank (113.6L), a hydroponic tank (102.2L), and a biofilter tank (18.9 L; Figure 1). To avoid shrimp escape and algal growth, plastic mesh, and lids were placed on the top of aquaculture tanks and hydroponic tanks, respectively. The lids on the top of hydroponic tanks were used to inhibit light into the tank, and also used as the floating rafts to support plants. Every biofilter tank was equipped with a 25-micron filter bag and bio-balls (surface area 98 ft²/ft³; Pentair Aquatic Eco-Systems, Inc., Apopka, FL, United States). Bio-balls in the biofilter tank were used to provide high surface area for attachment and colony expansion of microbes, and the filter bag was used to filter water from shrimp culture to prevent plant roots from clogging by biofloc. Air stones were installed in aquaculture tanks, hydroponic tanks, and biofilter tanks to maintain dissolved oxygen (DO) level above 6 mg/L. To maintain the water temperature within the optimal range 26–28°C for shrimp, submersible heaters (300w; Aqueon, Wisconsin, United States) were used in aquaculture tanks. Light source for plant growth was provided via LED light tubes (40w, 5000 lumens, 4000K daylight white; Kihung LED, Guangdong, China), that were suspended at a height of 16.5 cm over the plant growth bed. Quantum sensor (MQ-500 Full-Spectrum Quantum Meter; Apogee Instruments, Inc., UT, United States) was used to measure light intensity. The photosynthetically active radiation averaged 234 μmolm^–2s^–1. The photoperiod was 14 h light and 10 h dark.

A 3 × 2 experimental design was established in this study; 3 ratios of shrimp to plants (2:1, 3:1 or 5:1) and 2 C/N ratios (12 or 15). Treatments were designated 2:1–12 (stocking density ratio 2:1 with C/N ratio 12), 2:1–15, 3:1–12, 3:1–15, 5:1–12, and 5:1–15. Treatments were randomly assigned to three replicate experimental systems. The study was conducted for 4 weeks, from July 4 to August 1, 2020. Before the experiment started, all experimental systems were seeded with Bacillus spp. (EZ-Bio; Zeigler Brothers, Gardners, PA, United States), and inoculated water from established systems used in prior research. Sea salt (Instant Ocean^®, Blacksburg, VA, United States) was used to adjust the salinity to 15 ppt. One week prior to the experiment, shrimp were weighed and placed in aquaculture tanks to produce nutrients for plants. The stocking density of shrimp was 200 shrimp/m² (40 shrimp/tank), 300 shrimp/m² (60 shrimp/tank), or 500 shrimp/m² (100 shrimp/tank). The average weight of individual shrimp was 1.50 g. The stocking density of plants was 100 plants/m², which was equivalent to 24 plants (8 plants per species) in each hydroponic tank. Commercial shrimp feed (Zeigler Brothers, Gardners, PA, United States) was provided twice a day at 8 a.m. and 5 p.m., with a total daily amount of 3.0% of body weight divided into equal aliquots. Guaranteed analysis of the feed was 35% protein, 7% fat, and a maximum of 2% fiber.

Probiotics (EZ-Bio; Zeigler Brothers, Gardners, PA, United States) were used to manage water quality and the microbial community within every system. EZ-bio (Bacillus spp.) was inoculated at 10 mg/L into every system once a week prior to starting the experiment. As soon as shrimp were moved into aquaculture tanks, additional doses of probiotics were added every other day in the first week, twice per week in the second week, and once per week beginning in the third week continuing until the end of the experiment (Crab et al., 2010; Chu, 2014). Molasses (Hawthorne Gardening Co., Vancouver, WA, United States) was added to aquaculture tank as an organic carbon source to adjust the C/N ratio in the water with the same frequency of the probiotic inoculations. The amount of molasses added was based on the carbon-nitrogen content of shrimp feed and the carbon content of the molasses to adjust the C/N ratio to 12 or 15 (Xu et al., 2016). Potassium bicarbonate was added to maintain the alkalinity above 60 mg/L, and 10% sulfuric acid was applied to keep the pH below 8. Throughout the 4-week experiment, there was no water discharged or exchanged except for replacement due to evaporation.

During the experiment, dissolved oxygen, temperature (OxyGuard Handy Polaris DO meter, Farum, Denmark), and pH (pHTestr™ 10 Pocket pH Tester, Vernon Hills, IL, United States) were measured twice per day at 8 a.m. and 5 p.m. before feeding. Salinity (Vital Sine™ Salinity Refractometer, Pentair Aquatic Eosystems, Apopka, FL, United States) was measured once per day at 8 a.m. Water samples were collected twice per week from the aquaculture tank before feeding, to determine the concentrations of total ammonia nitrogen (TAN), nitrite-N (NO₂^–), nitrate-N (NO₃^–), phosphate (PO₄^3–), and alkalinity using HACH reaction kits (HACH, Loveland, CO, United States). Total suspended solids (TSSs) and volatile suspended solids (VSSs) were measured once a week by United States EPA method 1684.

Shrimp and plant growth performance, nutrient content in plants, and water quality parameters were analyzed using JMP v14.0 (SAS Institute Inc., Cary, NC, United States). Treatment means were compared by two-way analysis of variance (ANOVA). Statistical differences between means was determined by Tukey’s honestly significant difference test (HSD) at p ≤ 0.05.

The survival of shrimp in all treatments was above 95% and there were no significant (p > 0.05) differences among treatments. The growth of shrimp was significantly (p < 0.05) impacted by stocking density, but not by the C/N ratio or the interaction of factors (Table 1). Shrimp raised in 2:1–12 treatment had better growth performances (final weight, weight gain, and SGR) and a lower FCR. The mean values of final weight, weight gain, and SGR were significantly greater (p < 0.05) in the treatment 2:1–12, averaging 2.68g, 73.9%, and 1.97, respectively, compared to 5:1–12 and 5:1–15 treatments, yet, values were not significantly (p > 0.05) different from that of the 2:1–15, 3:1–12, and 3:1–15 treatments. Similarly, FCR was significantly lower (p < 0.05) in shrimp raised in 2:1–12 treatment compared to shrimp raised in 5:1–12 and 5:1–15 treatments, whereas it was not significantly (p > 0.05) different from the other treatments.

Survival of red orache, okahijiki, and minutina was 100% in all treatments. Stocking density ratio significantly (p < 0.05) affected the growth of all plants, and C/N ratio significantly (p < 0.05) affected water content (WC) of red orache (Table 2), the final fresh weight (FFW), final dry weight (FDW), RGR and WC of okahijiki (Table 3), and the FFW and RGR of minutina (Table 4). The interaction of the two factors exerted no significant (p > 0.05) effect on growth of plants.

During the experiment, temperature and dissolved oxygen (DO) were maintained at 28–29°C and 6.1–7.3 mg/L in all treatments, respectively. The salinity was monitored and controlled every day to maintain at the desired level, 15 ppt. The alkalinity in all treatments was not significantly different (p > 0.05) among each other. There were no significant differences (p > 0.05) found in the concentrations of total suspended solids (TSS) or volatile suspended solids (VSS) (Table 6). The pH was affected by the stocking density ratio, the lower stocking density ratio tended to have a higher pH than the higher stocking density ratio. Starting on day 9, the value was significantly higher (p < 0.05) in the stocking ratio of 2:1–12 and 2:1–15 treatment than that of 5:1–12 and 5:1–15 (Figure 2).

The concentrations of TAN, and NO₂^–, which are toxic to aquatic animals and plants, remained at safe and low concentrations (both were below 0.2 mg/ L) throughout the experiment in all treatments, although the treatments with higher stocking density ratio, 5:1–12 and 5:1–15, had significantly higher (p < 0.05) concentrations than that of lower stocking density ratio, 2:1–12 and 2:1–15 (Figures 3A,B). On the other hand, concentrations of NO₃^– and PO₄^3– continued increasing throughout the experiment. In general, 5:1–12 and 5:1–15 treatments had significantly higher (p < 0.05) concentrations of NO₃^– and PO₄^3– than 2:1–12 and 2:1–15 treatments (Figures 3C,D).

The results of this study indicate that an increase in stocking density ratio was a negative factor on shrimp growth. In addition, the FCR increased with the increasing stocking density, which increases the cost of production. Similar results were observed in other studies that investigated the effect of stocking density on shrimp production (Moss and Moss, 2004; Esparza-Leal et al., 2010; Neal et al., 2010; Krummenauer et al., 2011; Sookying et al., 2011; Façanha et al., 2016; Araneda et al., 2020; Fleckenstein et al., 2020). However, survival of shrimp in our experiment was above 95%, while other researchers reported that survival decreased with increasing density (Neal et al., 2010; Krummenauer et al., 2011; Araneda et al., 2020).

The reduced survival and growth of shrimp at higher stocking densities can be related to several factors including the availability of space for growth, competition for feed, cannibalism, and increased waste excretion leading to degraded water quality (Arnold et al., 2006; Wang and Gu, 2010; Ferreira et al., 2020; Luo et al., 2020). The high survival at high stocking density in this study might be attributed to the application of probiotics and molasses, because probiotics improve the resistance to adverse environments and the ability to tolerate stress (Martínez Cruz et al., 2012; Buruianǎ et al., 2014; Nemutanzhela et al., 2014; Olmos et al., 2020). The use of molasses increased the C/N ratio in the environment, which provided a beneficial environment for probiotics to assimilate nitrogenous waste (Avnimelech, 1999; Crab, 2010; Crab et al., 2012) and compete with pathogens, such as Vibrio spp. (Panigrahi et al., 2018). The increased C/N ratio also led to the generation of biofloc, which can be a supplementary food for shrimp (Avnimelech, 1999; Browdy et al., 2012; Crab et al., 2012). Moreover, the application of molasses improved the ability of plants to uptake nutrients from the water column (White, 2012). The combined effect from the application of probiotics and molasses maintained the water quality and led to higher survival in the present study, albeit, with reduced growth. It is unclear if the increased density and potential increases in marketable shrimp would offset the slower growth and increased feed costs associated with super-high-density culture, but this possibility needs to be explored.

In general, the growth performance of the three halophytes was better with the increasing stocking ratio and C/N ratio. The nutrient balance is one of the key factors to the success of aquaponics; too few animals (inadequate nitrogen) is limiting to plants, and excessive animal density will result in nitrogen excretion in excess of the plants’ ability to uptake compounds, and may lead to chronic or even acute toxicity to both animals and plants (Somerville et al., 2014).

In our study, plants were benefiting from a higher shrimp stocking density, in which more nutrients are provided for plants. All plants had greater yield in treatment 5:1 (Table 5). A similar trend was also reported by Shete et al. (2015); the plant production was higher in fish to plant ratio of 1:1, followed by1:2 and then 1:3. Furthermore, the concentration of macronutrients (N, P, and K) in plant tissues were also higher in both 5:1 treatments than the other stocking density ratio treatments. The higher concentrations of TAN, NO₂^–, NO₃^–, and PO₄^3– found in the 5:1 treatment was likely the result of the higher feed inputs into those treatments. The steady increase in NO₃^–, and PO₄^3– in all treatments may indicate the saturation of plant’s assimilation limits. Further research is needed to evaluate a longer culture duration or a different strategy on plant harvest (sequential stocking) since the concentration of all N- and P-compounds in the water will likely increase after every harvest (Yang and Kim, 2020b), which might be a concern for shrimp culture.

While not as pronounced as the effect of shrimp stocking density, the C/N ratio also exerted an impact on plant production characteristics. Additional carbon for amending the growing environment improves nutrient uptake by plants, increases crop yield, and alleviates phytotoxicity, caused by trace metals, salinity, pesticides, phytotoxins, or allelochemicals (Rodríguez-Kábana, 1986; Schenck, 2001; White, 2012; Pyakurel et al., 2019; Li et al., 2020). Overall, the effect of C/N ratio on plant growth, yield, and mineral nutrient concentrations in plants was relatively minor compared to stocking density.

To manage the water quality well, a robust microbial community is indispensable in aquaponics. Promoting the establishment of microbial flora by inoculation of probiotics or inoculations from mature water, used-biomedia, or biofloc from stabilized systems to new systems are efficient practices (Otoshi et al., 2011; Xu and Pan, 2012; Pinheiro et al., 2020; Chu and Brown, 2021). Xu and Pan (2012) inoculated bioflocs, characterized by Bacillus sp. as the predominant bacteria, into experimental tanks before their study. The concentrations of toxic nitrogenous waste, TAN and NO₂^–, in their study were maintained below 0.51 mg/L and 1.25 mg/L, respectively. In the present study, the concentration of TAN and NO₂^– remained lower than 0.2 mg/L throughout the experiment, even in the high stocking density treatments. Additional research needs to be conducted examining the frequency of probiotic application and varying harvest scenarios, as after every plant harvest, the TAN and NO₂^– will likely increase (Yang and Kim, 2020b).

Concentrations of TAN and NO₂^– were higher in the higher C/N treatments while the NO₃^– concentrations were lower. Similar results were also reported in other studies (Xu et al., 2016, 2018). The additional organic carbon facilitates growth of heterotrophic bacteria, which compete for nutrients and space inside the biofilm or biomedia with the nitrifying bacteria (NB), which are composed of ammonia-oxidizing bacteria (AOB) and nitrite-oxidizing bacteria (NOB). The rate of reproduction of heterotrophic bacteria is much faster than NB, while the reproduction rate of AOB is faster than NOB (Hu et al., 2015). Due to the competition with heterotrophic bacteria, and the slower growth rate of AOB and NOB, the efficiency of nitrification decreases with the increasing C/N ratio, resulting in a decrease in TAN removal rate and NO₃^– productivity (Zhu and Chen, 2001; Michaud et al., 2006; Luo et al., 2020). However, in our study, all nitrogenous waste products remained below concentrations considered toxic.

The concentrations of PO₄^3– continued to accumulate throughout the experiment, similar to previous results (Boxman et al., 2018; Yang and Kim, 2019, 2020a,b; Chu and Brown, 2021; Huang et al., 2021), and was higher in the treatments with the higher stocking densities and C/N ratio, likely due to higher feed inputs. Another explanation could be the higher C/N ratios resulted in bioflocs dominated by heterotrophic bacteria and less algae (Xu et al., 2016). Moreover, the accumulation of PO₄^3– suggests the saturation of plant’s assimilation ability. PO₄^3–, one of the compounds that causes eutrophication, is an issue for aquaponics; hence, more research is required to improve the management of PO₄^3– in aquaponics and determine how to improve plant’s ability to assimilate PO₄^3–.

In aquaponics, pH is another vital parameter that can be affected by nitrification, nutrient assimilation by plants and heterotrophic bacteria, and CO₂ excretion by aquatic animals, as well as other factors (Yang and Kim, 2019; Li et al., 2020). The process of nitrification and nitrogenous waste assimilation by bacteria, and CO₂ released through the respiration of aquatic animals and microorganisms tends to decrease the pH. Conversely, CO₂ removal and nutrient assimilation by plant tends to raise the pH (Ebeling et al., 2006; White, 2012; Somerville et al., 2014). This may explain why the pH level was lowest at the stocking density of 5:1, followed by 3:1, then 2:1. Also, the lower pH can be another possible reason for a better plant growth in higher stocking density treatments, because nutrient availability increases with the decreasing pH (Somerville et al., 2014).