LED Lighting and High-Density Planting Enhance the Cost-Efficiency of Halimione Portulacoides Extraction Units for Integrated Aquaculture

Halophytes are salt-tolerant plants that can be used to extract dissolved inorganic nutrients from saline aquaculture effluents under a production framework commonly known as Integrated Multi-Trophic Aquaculture (IMTA). Halimione portulacoides (L.) Aellen (common name: sea purslane) is an edible saltmarsh halophyte traditionally consumed by humans living near coastal wetlands and is considered a promising extractive species for IMTA. To better understand its potential for IMTA applications, the present study investigates how artificial lighting and plant density affect its productivity and capacity to extract nitrogen and phosphorous in hydroponic conditions that mimic aquaculture effluents. Plant growth was unaffected by the type of artificial lighting employed—white fluorescent lights vs. blue-white LEDs—but LED systems were more energy-efficient, with a 17% reduction in light energy costs. Considering planting density, high-density units of 220 plants m−2 produced more biomass per unit of area (54.0–56.6 g m−2 day−1) than did low-density units (110 plants m−2; 34.4–37.1 g m−2 day−1) and extracted more dissolved inorganic nitrogen and phosphorus. Overall, H. portulacoides can be easily cultivated hydroponically using nutrient-rich saline effluents, where LEDs can be employed as an alternative to fluorescent lighting and high-density planting can promote higher yields and extraction efficiencies.


Introduction
Halophytes are a group of plants characterized by a range of morphological and physiological features that allow them to thrive in brackish and saline environments [1]. Due to these capabilities, they have been increasingly studied for the treatment of eutrophic saline effluents, especially in the context of Integrated Multi-Trophic Aquaculture (IMTA) frameworks [2][3][4]. The major benefit of this integration pertains to the uptake and reuse of wasted nutrients generated within the production system [5,6].
Previous studies, using a variety of halophyte species, demonstrated positive outcomes in growth and extraction efficiency of nitrogen (N) and phosphorous (P) in integrated aquaculture settings [3]. This approach is associated with the principles of the circular economy, where aquaculture waste streams are valorized through the phytoremediation, harvesting and commercialization of plant biomass [7]. Several edible halophyte species can deliver food products with distinct organoleptic and functional properties, including vegetable oils and bioactive compounds [8][9][10][11][12]. Moreover, halophytes' relatively low-sodium content makes them a suitable alternative that can reduce sodium intake for populations at risk [13,14]. Given their potential socioeconomic and nutritional benefits, continued investigation into the horticultural production of halophytes within integrated aquaculture frameworks is necessary.
This paper focuses on the species Halimione portulacoides (L.) Aellen (a.k.a. sea purslane), an edible halophyte relatively widespread throughout European and Mediterranean saltmarshes with traditional human uses [15] whose aquaculture and nutritional potential has only recently been scientifically explored [16][17][18][19][20][21]. These first studies showed that H. portulacoides has one of the highest productivity rates among studied halophytes in saline hydroponic conditions and can contribute to the substantial removal of dissolved inorganic N and P, up to a rate of approximately 4.0 mg N g −1 and 0.4 mg P g −1 [16].
For an effective introduction of halophytes in integrated aquaculture systems, it is first important to understand the conditions that benefit their production and estimate key variables (e.g., growth rates, nutrient extraction rates) that can help adjust those conditions for cost-efficient solutions. In the present study, the focus lays on two key variables that influence horticultural production: light and planting density [22][23][24][25][26].
Providing optimal and energy-efficient lighting conditions is paramount for commercial plant production [27], and solid-state LED lighting is nowadays considered the most flexible and cost-efficient technology for indoor hydroponics because it provides better control over different light parameters (e.g., spectrum, irradiance, photosynthetically active radiation) [28]. Different light spectra produced by LEDs, particularly red and blue, have been shown to influence the productivity of conventional leafy-green crops, and results suggest a general improvement in performance relative to other types of artificial lighting (e.g., fluorescence) and natural lighting [29][30][31][32][33][34]. Yet, the influence of different light spectra in the development of halophyte crops is still relatively understudied [35][36][37]. In the present study we decided to test a white T5 fluorescent lamp, a common full-spectrum fluorescent light source in hydroponics, against blue-white LED tiles (6 blue and 4 white LEDs) that covered the full-spectrum (white LEDs) but also provided extra blue light (blue LEDs) to inspect potential benefits.
Planting density is known to affect individual plant development, potentially inducing competitive behavior, which can, in turn, affect production outputs [22,25,26]. In the case of halophytes, some species showed reduced individual growth as plant density increased (e.g., Atriplex prostrata Boucher) [38] while others seemed to be relatively unaffected by this variable (e.g., Batis maritima L., Cressa cretica L., Salicornia europaea L., Sesuvium portulacastrum (L.) L.) [39][40][41]. Since different species can have different biomass allocation strategies under crowding conditions, experimental growth trials can provide invaluable information about the appropriate planting densities to maximize whole plant development or promote the growth of target organs, such as fruits, seeds or leaves.
Hydroponic experiments that test halophytes performance for IMTA applications are still limited in the scientific literature and the present study is a contribution to fill this gap. The main goal of this work is to determine the conditions that further benefit the performance of the sea purslane H. portulacoides as an extractive species for IMTA by understanding the influence of lighting conditions and planting density on its vegetative development and nutrient extraction capacity. To achieve this goal, we tested the following null hypotheses: (i) increasing planting density does not affect hydroponic performance, and (ii) shifting white fluorescent lighting to blue-white LED lighting does not affect plant development. The concentration of photosynthetic pigments (chlorophylls and carotenoids) was also quantified to assess potential changes promoted by both types of light, which could have implications for the added value of halophyte products.

Materials and Methods
Using a two-way factorial design with two levels, the effects of artificial lighting and planting density on the performance of H. portulacoides extraction units were tested in controlled hydroponic settings rather than real aquaculture settings, to provide better control over the conditions of the aquatic media. Measuring growth performance consisted of recording individual biomass gain every week (one plant per hydroponic unit) and, at the end of the experimental period, recording total biomass yield (whole plant, belowand above-ground biomass), the number of leaves, and stem length at the level of the hydroponic unit. Photosynthetic pigment concentration was also quantified as a proxy measure of the status of the photosynthetic apparatus. Nutrient extraction efficiencies were assessed through measurements and mass-balance calculations of ammonium (NH 4 -N), oxidized forms of inorganic nitrogen (NO x -N), and orthophosphate (PO 4 -P) present in the hydroponic solution, at the beginning and the end of each extraction (or remediation) cycle (1 week). The initial concentrations of inorganic nitrogen (N) and phosphorus (P) in each batch of hydroponic solution were approximately 60 mg N L −1 (4.3 mM N) and 3 mg P L −1 (0.1 mM P). Following Custódio et al. [16], these concentrations were not only analogous to those typically found in more intensive aquaculture effluents but also guaranteed non-limiting access to N and P during each extraction cycle. The data generated during this study and presented in the Results section are made available in a spreadsheet as Supplementary Material (File S1).

Plant Material
Stems of H. portulacoides were collected from wild specimens in Ria de Aveiro (Portugal) coastal lagoon (40 • 38 04.1"N 8 • 39 40.0"W) in March 2018. Six hundred cuttings, with 4 nodes each, were obtained from those stems and these were placed in small polyethylene containers filled regularly with Hoagland's nutrient solution to stimulate root development. The elemental composition of the nutrient solution was: 40 mg Ca L −1 , 60 mg K L −1 , 16 mg Mg L −1 , 56 mg N L −1 , 16 mg P L −1 , 0.28 mg B L −1 , 0.03 mg Cu L −1 , 1.12 mg Fe L −1 , 0.11 mg Mn L −1 , 0.34 mg Mo L −1 , 0.13 mg Zn L −1 . Cuttings grew under natural light and temperature for 3 months and, in June 2018, fully rooted cuttings were randomly distributed throughout the hydroponic units to initiate the acclimation period. Plants were acclimated to the new hydroponic indoor conditions for 2 weeks and during the second week were progressively adapted to a salinity of 20 ppt (0.5% increments of NaCl every second day).

Experimental Setup
The growth trial took place over 10 weeks, during which plants were allowed to develop with minimal disturbance on a deep-water culture hydroponics configuration using extruded polystyrene floating-rafts without intermediary harvests during this period. An overview of experimental units under different lighting systems at week 1 and week 9 are displayed in Figure 1.
The hydroponic units were made of opaque polypropylene boxes (300 × 200 × 170 mm) with an overflow outlet to keep water volume at the 5 L mark. The base for the hydroponic solution was artificial seawater prepared by dissolving commercial Red Sea salt (Red Sea, Cheddar, UK) in freshwater purified by reverse osmosis (V2Pure 360 RO System, TMC, Hertfordshire, UK) until achieving a salinity of 20 ppt. The experimental hydroponic media was a modified version of Hoagland's solution described previously, where N and P concentrations were modified to resemble realistic values of dissolved inorganic N and P as measured in fish-farming effluents [16]. The detailed nutrient composition of the experimental hydroponic solution is presented in Table S1 (Supplementary Materials).
The experimental design consisted of a two-way factorial design with two levels, resulting in 4 treatments. The levels of artificial lighting were 'fluorescent lights' and 'LEDs' and the levels of plant density were '110 plants m −2 ' and '220 plants m −2 '. Treatment labels were as follows: F110 = fluorescent lights + 110 plants m −2 ; F220 = fluorescent lights + 220 plants m −2 ; L110 = LEDs + 110 plants m −2 ; and L220 = LEDs + 220 plants m −2 . Each treatment was assigned to five replicate hydroponic units, resulting in a total of 20 units, each having a surface area of 0.0455 m 2 . Therefore, the F110 and L110 units were assigned 5 plants each, and the F220 and L220 units were each assigned 10 plants, for a total of 150 plants distributed across all hydroponic units. To do so, 150 fully rooted cuttings of similar weight were selected, identified by a number from 1 to 150 and randomly assigned to the hydroponic units following a randomized sequence of integers (from 1 to 150) generated by a tool provided at https://www.random.org (accessed on 18 June 2018). Each plant was individually photographed and weighed at the beginning of the experiment. The average initial biomass was 6.8 ± 1.8 g plant −1 . Afterward, the cuttings were fixed to the floating rafts using natural cotton, to hold each plant in place at the base of the aerial portion. The hydroponic medium was continuously aerated to maintain aerobic conditions, and the hydroponic units were refilled as needed with freshwater purified by reverse osmosis to compensate for evapotranspiration. The hydroponic units were made of opaque polypropylene boxes (300 × 200 x 170 mm) with an overflow outlet to keep water volume at the 5 L mark. The base for the hydroponic solution was artificial seawater prepared by dissolving commercial Red Sea salt (Red Sea, Cheddar, UK) in freshwater purified by reverse osmosis (V2Pure 360 RO System, TMC, Hertfordshire, UK) until achieving a salinity of 20 ppt. The experimental hydroponic media was a modified version of Hoagland's solution described previously, where N and P concentrations were modified to resemble realistic values of dissolved inorganic N and P as measured in fish-farming effluents [16]. The detailed nutrient composition of the experimental hydroponic solution is presented in Table S1 (Supplementary Materials).
The experimental design consisted of a two-way factorial design with two levels, resulting in 4 treatments. The levels of artificial lighting were 'fluorescent lights' and 'LEDs' and the levels of plant density were '110 plants m −2 ' and '220 plants m −2 '. Treatment labels were as follows: F110 = fluorescent lights + 110 plants m −2 ; F220 = fluorescent lights + 220 plants m −2 ; L110 = LEDs + 110 plants m −2 ; and L220 = LEDs + 220 plants m −2 . Each treatment was assigned to five replicate hydroponic units, resulting in a total of 20 units, each having a surface area of 0.0455 m 2 . Therefore, the F110 and L110 units were assigned 5 plants The fluorescent light was provided by tubular fluorescent lamps (Philips 54W/830 Min Bipin T5 HO ALTO UNP), while the LED light was provided by solid-state LED lighting tiles (AquaBeam 1500 Ultima NP Ocean Blue Light). Their photometric and colorimetric information is presented in Table 1. Each fluorescent lighting system was composed of two fluorescent lamps and each LED system was composed of three LED tiles. The spectral profiles of artificial lighting sources are presented in Figure S1 as Supplementary Materials. The photoperiod was set at 14 h light: 10 h dark and the photosynthetically active radiation (PAR) was adjusted at the beginning of every week so that all artificial lighting sources delivered identical PAR to the plants. PAR, the fraction of electromagnetic radiation that can be used by plants during photosynthesis, is found within the spectral range of 400-700 nm. It is normally expressed as photosynthetic photon flux density (PPFD, µmol photons m −2 s −1 ). The PAR values reaching the top of the canopy of stocked plants were measured twice a week with a spherical microquantum sensor (US-SQS/L) connected to a Universal Light Meter ULM-500 (Heinz Walz, Pfullingen, Germany). The average PPFD throughout the experiment was 371.0 ± 12.0 µmol photons m −2 s −1 . Water temperature and pH were measured using a multi-parameter portable meter (ProfiLine pH/Cond 3320, WTW, Weilheim, Germany) and dissolved oxygen was measured by a portable oxygen meter (Oxi 3310, WTW, Weilheim, Germany). Measurements were performed twice a week (day 2 and day 7) and average weekly values are presented in Figure S2 (Supplementary Materials). Overall, hydroponic units displayed an average water temperature of 22.9 ± 0.7 • C, a pH of 7.8 ± 0.2, and a dissolved oxygen concentration of 6.7 ± 0.6 mg L −1 (= 81.1 ± 7.3% saturation).

Growth Performance
To measure weekly growth, one plant per hydroponic unit (five plants per treatment) was randomly chosen at the beginning of the growth trial, and the total weight was systematically measured at the end of every remediation cycle of 1 week. Only one plant per unit was weighed every week to minimize handling time and avoid stress to the other plants. At the end of the experiment (end of week 10), all plants were individually photographed, divided into roots, stems, and leaves and weighed. The three plant organs (roots, stems, and leaves) were pooled per experimental unit and stored at −80 • C. Image-analysis software (ImageJ 1.51) was used to measure the length of stems and count the leaves.

Nutrient Extraction Efficiency
The efficiency of H. portulacoides in extracting N and P was determined using a retention time (the amount of time the solution remains in the hydroponic unit) of one week (7 days). The time frame allowed for nutrients to be taken up by the plants was crucial to the performance of a hydroponic unit and the suitable retention time could be highly variable. Previous nutrient extraction studies with halophytes used a wide range of retention times, from 12 h to 5 weeks [20,[41][42][43]. Following Custódio et al. [16], a one-week retention time was considered appropriate for the present study.
By the end of each extraction period, water samples from each hydroponic unit were collected and filtered (Whatman GF/C filters) into flasks, and final concentrations of ammonium (NH 4 -N), oxidized forms of inorganic nitrogen (NO x -N) and orthophosphate (PO 4 -P) in the hydroponic media were determined using a San++ Continuous Flow Analyzer (Skalar Analytical, Breda, The Netherlands) following Skalar's standard automated method (NH 4 -N: modified Berthelot reaction for ammonia determination; NO x -N: Total UV digestible nitrogen/nitrate + nitrite/nitrite; PO 4 -P: Total UV digestible phosphate/orthophosphate). After sampling, the medium in each unit was renewed with a new batch of its corresponding treatment solution. The real initial concentrations of NH 4 -N, NO x -N, and PO 4 -P in every new batch were determined the same way as described above. Nutrient extraction efficiency was estimated (both weekly and in total) for dissolved inorganic nitrogen (DIN-N), calculated as the sum of NH 4 -N and NO x -N, and dissolved inorganic phosphorus (DIP-P), equivalent to PO 4 -P.

Photosynthetic Pigments
Biomass samples were taken from the pool of leaves produced in each hydroponic unit and freeze-dried. Pigments were extracted using 95% cold-buffered methanol (2% ammonium acetate). Before extraction, samples were ground with a mortar and 2-3 mg of the sample were weighed into Eppendorf tubes. Subsequently, 1 mL of extraction solvent was added to each tube, followed by 45 s sonication and 20 min incubation at −20 • C in the dark. The extracts obtained were filtered through 0.2 µm PTFE membrane filters, and 50 µL were injected into an HPLC equipment with an SPD-M20A photodiode array detector (Shimadzu, Kyoto, Japan). The chromatographic separation of pigments was achieved using a Supelcosil C18 column (Sigma-Aldrich, St. Louis, MO, USA) following Cruz et al. [44]. Pigments were identified from absorbance spectra, and retention times and concentrations were calculated using linear regression equations obtained from pure crystalline standards (DHI, Hørsolm, Denmark).

Statistical Analysis
Statistical analysis was performed using R v3.4.3 (64-bit) with R Studio and statistically significant differences were considered at p < 0.05. A two-way ANOVA was employed to assess the effects of plant density and artificial lighting on outcome variables. Post-hoc Tukey's HSD test for individual means comparison was used when significance was observed. All data were checked for normality (Shapiro-Wilk test) and homogeneity of variance (Levene's test). A repeated-measure ANOVA was used to assess treatment differences in cumulative biomass gain and N/P extraction efficiency across time points. The Geenhouse-Geisser correction was applied when the sphericity assumption was violated, and the Bonferroni correction was used when performing multiple pairwise comparisons.

Growth Parameters and Productivity
The initial biomass was 6.80 ± 0.04 g per plant and significant differences in growth parameters were detected after 10 weeks (Table 2). At the individual level, F110 produced significantly (p < 0.05) higher aboveground biomass than L220 and higher stem growth than F220 and L220. At the unit level (pooled individual weights), on the other hand, both F220 and L220 produced significantly higher (p < 0.05) aboveground and belowground biomasses, number of leaves and stem growth than did F110 and L110. The total final biomass in F220 was higher (p < 0.05) than in both L110 and F110; meanwhile, L220 only differed significantly (p < 0.05) from L110. The same trend was observed in total productivity, with F220 expressing the highest productivity rate, at 56.6 ± 14.0 g m −2 day −1 . Regarding relative growth rate, the treatments showed similar values even though the low-density treatments had a slightly higher rate on average. Cumulative biomass gains over time ( Figure 2) showed a large variance in individual weight across treatments, and, as a result, no statistical main effects of either plant density nor artificial lighting were detected. Nonetheless, a within-subject effect of the variable week was detected [F (76.4,1222.7) = 120.58, p = 2.88 × 10 −9 , generalized η 2 = 0.76].     Figure 3A). Regarding the total quantity of DIN-N extracted during the totality of the experiment ( Figure 4A), a main effect of plant density (p < 0.05) was detected: high-density units extracted a total of 875.3 ± 187.7 mg and low-density units extracted in total 686.4 ± 94.2 mg. Overall, high-density units extracted more DIN-N on average than low-density units and extraction efficiencies (%) were: F110 = 21.6 ± 1.8%; F220 = 26.3 ± 6.9%; L110 = 21.8 ± 4.1%; L220 = 29.1 ± 5.1%.  After normalizing the total quantity of DIN-N removed to the total biomass produced ( Figure 4C), the main effects of plant density (p < 0.001) and artificial lighting (p < 0.05) were detected: low-density units and LED units resulted in higher DIN-N extraction per gram of biomass produced. Pairwise comparisons show that F220 removed significantly less (p < 0.01) DIN-N per gram of biomass (4.7 ± 0.2 mg g −1 ) than either F110 (6.0 ± 0.9 mg g −1 ) or L110 (6.4 ± 0.6 mg g −1 ), yet the L220 extraction rates (5.5 ± 0.3 mg g −1 ) were not significantly different from those of low-density units.
Regarding the total DIP-P removed per gram of biomass ( Figure 4D), neither main effects nor the treatment effects were detected, and values ranged between 0.30 and 0.34 mg DIP-P extracted per gram of biomass produced. Regarding the total DIP-P removed per gram of biomass ( Figure 4D), neither main effects nor the treatment effects were detected, and values ranged between 0.30 and 0.34 mg DIP-P extracted per gram of biomass produced.

Artificial Lighting
The two tested artificial lighting systems displayed two distinct light spectra that were expected to induce changes in H. portulacoides development. Nonetheless, according to the results, no differences could be associated with the type of lighting system. In contrast, previous studies showed that some types of artificial lighting and their associated light spectra can affect different stages of halophytes' development. For instance, the seedling germination rates and shoot development of Atriplex halimus L., Atriplex hortensis L., and S. europaea were higher when irradiated with a combined red and blue LED system compared with fluorescent lighting [36], and the vegetative development of A. hortensis also improved under red and blue LEDs [45]. Similarly, the edible Mesembryanthemum crystallinum L. (ice plant) showed improved vegetative development under red and blue LEDs compared to red or blue LEDs alone [35], as well as under red and white LEDs compared with other combinations of white, blue, red, and far-red LEDs [37].
The effect of LED lighting on plant hydroponic production has been more intensively studied using more traditional crops [29,46]. The use of blue and red LEDs (alone or combined) seems to improve the quality and yields of several vegetables and fruit when compared with fluorescent lighting [31]. In aquaponics, for instance, the productivity of kale (B. oleracea) and Swiss chard (Beta vulgaris L.) is higher with LEDs [47] and some studies suggest that blue LEDs seem to stimulate leaf area enlargement and the aboveground development of lettuce (Lactuca sativa L.) and other vegetables (e.g., Chinese cabbage, spinach and coriander) [28,48].
In Table 4, a non-exhaustive summary of hydroponic trials looking at the different effect of LEDs and fluorescent lighting in the vegetative growth of leafy greens is presented, but the results were varied and inconclusive. For instance, L. sativa var. "crispa" grew better under red LEDs compared to either blue or red and blue LEDs [49]; meanwhile, the var. "capitata" seemed to grow better under blue LEDs [50]. The reality is that the effects of light quality in regulating growth processes and physiology in plants is tremendously complex. Even within the same species, outcomes can be different among varieties and time points [51,52].
The analysis of pigment concentrations suggested that the photosynthetic pigments in H. portulacoides leaves were unaffected by the type of light spectra after a long-term exposure of 10 weeks. The only indication of an effect came in fact from the planting density, where the concentrations of antheraxanthin and zeaxanthin were higher in low-density units. Being products of the photoprotective xanthophyll cycle [53], one possible explanation for this observation is that leaves from low-density plants were probably exposed to higher irradiance for a longer time-period than high-density plants due to less shading.
Even though the type of light did not seem to affect photosynthetic pigments in H. portulacoides, an increase in the concentration of carotenoids, which include xanthophylls and carotenes, has been observed in leafy vegetables exposed to blue LEDs [29,54]. For instance, several L. sativa varieties, cabbages (Brassica rapa L. and B. oleracea varieties) and water spinach (Ipomoea aquatica Forssk.) displayed an increase in the concentration of chlorophylls and other pigments when growing under blue LEDs [49,[55][56][57][58][59]. Nonetheless, some L. sativa varieties displayed similar concentrations regardless of exposure to fluorescent lights or LEDs [60,61], highlighting the complexity of the effects of light quality in plant physiology.
One key aspect to consider when choosing the type of lighting system to cultivate plants in a controlled environment is energy efficiency. In the present study, the wattage of one fluorescent lamp was 54 and one LED unit was 30 W. Therefore, operating on a 14:10 light-dark photoperiod, a fluorescent lamp consumed 0.76 and a LED unit 0.42 kWh day −1 . Considering that one fluorescent lighting system was composed of two lamps consuming a total of 1.52 kWh day −1 , and one LED lighting system was composed of three LED tiles consuming 1.26 kWh day −1 , lighting energy costs were reduced by 17% just by using LEDs.
Operating costs could be driven down even further by employing LEDs in H. portulacoides hydroponic production since their lifespan and maintenance costs are typically lower than fluorescent lights [28]. Despite requiring a higher initial investment, LEDs are likely more cost-efficient given their potentially higher energy efficiency, longer lifespan, and lower maintenance cost [62]. Since H. portulacoides was seemingly unaffected by the type of artificial lighting tested, the LEDs can be seen as a suitable cost-efficient alternative to fluorescent lights. Table 4. Hydroponic-based studies of vegetative growth performance and photosynthetic pigment accumulation of leafy greens under different LED spectra and fluorescent lighting. DWC--deep water culture; NFT--nutrient-film technique; FL--fluorescent lighting.

Species
Hydroponic Technique

Planting Density
Despite promoting lower individual growth, high-density treatments L220 and F220 produced higher yields at the level of the hydroponic unit. Doubling planting density from 110 to 220 plants m −2 increased hydroponic unit productivity 52-57% for an average value of approximately 54-57 g FW m −2 day −1 . Higher productivity values were observed by Custódio et al. [16], who reported average yields of 63-73 g FW m −2 day −1 under identical planting densities but by using different combinations of N and P concentrations. On the other hand, Buhmann et al. [42] reported lower productivity values of 33 g FW m −2 day −1 using very different experimental conditions. These first studies with H. portulacoides (Table 5) already suggest contrasting outcomes, most likely influenced by different culture conditions, genetic variability and the use of different ecotypes [64].
In other halophyte species, planting density seemed to have had little effect on the hydroponic performance (Table 5). For instance, Webb et al. [41] observed that S. europaea displayed no difference in total productivity between 200 and 10,000 plants m −2 . This lack of a plant density effect could be ascribed to the high morphological plasticity of Salicornia plants [41,65]. Still, some species of Salicornia seem to be more productive than others under similar nutrient conditions. For instance, in the same study presented above, S. europaea Salicornia species have annual life cycles, contrary to a perennial plant like H. portulacoides, which indicates major differences in life histories and growth strategies [69]. As such, comparisons between very different species are only meaningful from a horticultural perspective, where some species can be considered better than others in productivity and nutrient extraction rates, for instance. Other edible perennial halophytes investigated for their productivity under different plant densities (Table 5) [71].
Increasing planting density also affected DIN and DIP extraction efficiencies, as highdensity units extracted more in total, with average extraction efficiencies of 28% DIN and 36% DIP. Under similar nutrient conditions and using the same fluorescent lighting, highdensity and retention time, Custódio et al. [16] reported extraction efficiencies of 35% DIN and 32% DIP, values very close to the present study. However, using radically different hydroponic conditions can produce very different outcomes as shown in previous studies where H. portulacoides displayed extraction efficiencies of 50% DIN and 45% DIP [42] and 65% DIN and 0% DIP [20]. This stresses the fact that culture parameters must be finetuned for the specific conditions of a hydroponic or IMTA system, based on the nutrient availability and nutrient extraction rates.
Interestingly, Boxman et al. [39] observed an improvement in DIN extraction efficiency by B. maritima when planted at a low density compared with high-density units; meanwhile, Webb et al. [41] reported no changes in extraction efficiencies after a 50-fold increase in S. europaea planting density, which demonstrated that increasing density did not always equate with an increase in nutrient extraction capacity and that resource competition in plants is complex [72]. The type of substrate used to grow halophytes is also a key factor that influences the performance of a hydroponic unit [73]. Most studies mentioned so far used some type of inert solid substrate (e.g., quarry sand, coconut fiber, expanded clay, perlite) that substantially increased the area available for microbial communities to establish and influence nutrient dynamics in water through complementary processes such as denitrification and adsorption. Confounding factors that affect N and P dynamics are still present in systems using soilless media, as in the present study, such as aeration, water mixing and the presence of microorganisms in the rhizosphere and biofilms [74]. The retention time (exctraction cycle) is another important factor that enables these processes [75] and is a key parameter to consider in these types of studies.
The optimization of hydroponic conditions for halophyte production is indeed very complex due to the interplay of many different environmental variables. Therefore, carefully deliberated experimental designs are key to controlling the many confounding factors present in these systems, which become even more complex in real, rather than simulated, integrated aquaculture settings.

Conclusions
As edible halophytes continue to reveal their potential as crops with a role to play in the future of sustainable food production, the conditions for their commercial cultivation must continue to be explored. This is especially true in the context of integrated saline aquaculture where their role could be i) as phytoremediation units that can recover wasted dissolved nutrients and ii) as cash-crops that are nutritional, material and energy sources. By shedding light on the most suitable hydroponic conditions for growing H. portulacoides, future halophyte producers can make informed decisions that translate into sustainable and profitable cultures. Halimione portulacoides displayed productivity values at the higher end of those exhibited by other edible halophytes, given that appropriate planting density and nutritional conditions were present. Potential biomass allocation trade-offs should nonetheless be taken into consideration since certain densities can promote undesirable phenotypes. Regarding artificial lighting. H. portulacoides grew similarly under white fluorescent lighting and blue-white LED lighting, suggesting that the LEDs can be more cost-efficient solution given their similar performance and have potentially lower operating costs. Future trials should continue to explore how different LED spectral profiles can improve development and stimulate the accumulation of bioactive compounds to add functional value to halophyte products. It is also crucial to promote the design and engineering of integrated experimental and commercial setups and the standardized reporting of results to allow a more reliable comparison between hydroponics/IMTA studies and provide more robust data to help determine which halophytes are the most suitable for integrated aquaculture frameworks.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/app11114995/s1, Figure S1: Experimental growth systems with A) fluorescent lights and B) LEDs, Figure S2: Average water A) pH, B) temperature and C) dissolved oxygen, Table S1: Chemical composition of the hydroponic medium. The data generated and presented in this study is made available as supplementary dataset(File S1) in the form of an Excel file, namely growth data, nutrient extraction data and photosynthetic pigments data Portugal 2020 and European Union, through the European Regional Development Fund, and by the project "AquaMMIn-Development and validation of a modular integrated multitrophic aquaculture system for marine and brackish water species" (MAR-02.01.01-FEAMP-0038) co-funded by Portugal 2020 and the European Union through Mar 2020, the Operational Programme (OP) for the European Maritime and Fisheries Fund (EMFF) in Portugal.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are made available as supplementary material.

Conflicts of Interest:
The authors declare no conflict of interest.