Materials and Methods

The study was conducted on an experimental marine aquaponic system located in Sarasota, FL. The system was housed in a polycarbonate greenhouse with two exhaust fans for ventilation. The hydroponic plant raceways were covered by a corrugated polycarbonate roof, which was rated at 30% shade. To provide shade and cooler temperatures for the fish tanks, a

55% shade cloth covered approximately 50% of the greenhouse for a total of 85% shading. A system diagram is shown in Figure 3.2 and a summary of system components and volumes are presented in Table 3.3.

Table 3.3: System components and volumes in the marine aquaponic system.

System Component Volume (m³)

Fish tanks (three 3.3 m³ tanks) 9.9

Swirl separator 0.60

Upflow media filter 3.3

Backwashing sump 0.34

Moving bed bioreactor 5.4

Hydroponic beds (four 5.35 m³ rectangular raceways) 21

Pumping sump 3.6

Partially submerged sand filter 4.6

Sand filter sump 1.3

Total 50

Three 3.3 m³ fish culture tanks were stocked with red drum (Sciaenops ocellatus) on September 30, 2015, which was considered experiment day 0. Water flowed by gravity from the fish tanks to a 0.6 m³ swirl separator (Wave Vortex, WLF36, Pentair, Apopka, FL) and then to an upflow media filter. After filtration, system water entered a moving bed bioreactor (MBBR).

A target fish density of 21 kg/m³ was used to determine both the MBBR media volume and system flow rate based on the recommendations given by Losordo and Hobbs (2000). To achieve sufficient TAN removal, the MBBR was packed with 1.8 m³ Kaldnes media (Fureneset, Norway) to obtain a total surface area of 630 m² (Losordo and Hobbs, 2000). From the MBBR system, water was divided through four floating raft hydroponic plant beds operated in parallel. Four flow totalizers (Midwest Instruments, Model: 9002, Rice Lake, WI) were located in the influent pipes to the hydroponic plant beds and recorded flow rates of 38-61 L/min to each hydroponic plant bed. The hydroponic plant beds were constructed from wood and lined with polyethylene.

Each hydroponic plant bed had dimensions 12.8 m x 1.2 m for a total growing area of 61.4 m². Aeration was added to the hydroponic plant beds using 3.8 cm x 3.8 cm fine-pore diffusers

placed every 1.2 m in the plant beds. A stand pipe located at the end of each plant bed

maintained a water height of 40.6 cm. Effluent from the plant beds was recombined into one pipe and flowed to a 2.8 m diameter storage tank that contained a recirculation pump (Sweetwater High-Efficiency Pump, SHE2.4, Pentair, Apopka, FL). The system maintained a recirculation rate of 279 ± 26 L/min based on the requirements for TAN removal in the MBBR (Losordo and Hobbs, 2000). Fresh groundwater was added as needed to account for evaporation and

evapotranspiration. It was added directly to the storage tank and monitored with a water meter (Badger Recordall®, Model 25, Badger Meter Inc. Milwaukee, WI).

The upflow media filter was constructed from a 2.3 m diameter tank packed with Kaldnes media (Fureneset, Norway) to a depth of 0.3 m. An aeration grid was located on the bottom of the upflow media filter, which was used to agitate the media for backwashing. Backwashing of both the swirl separator and upflow media filter was performed twice weekly until experiment day 157, after which time backwashing was performed three times weekly. System water

removed during backwashing drained into a 1 m diameter sump tank, hereafter referred to as the solids sump. A submersible pump in the solids sump, operated by a float switch, pumped

backwash water into a sand filter with dimensions of 12 m by 1.6 m. The sand filter was

constructed in layers, with a bottom layer consisting of 15 cm depth of 3.8 cm gravel, a layer of polyethylene cloth, and a top layer consisting of 15 cm of 0.45-0.55 mm filter sand. A perforated pipe located underneath the sand collected filtered backwash, which then drained into a 1.8 diameter sump, hereafter referred to as the sand filter sump. Water in the sand filter sump was recirculated back to the upflow media filter via a submersible pump operated with a float switch.

Initially unsaturated filtration was used, such that the sand filter dried out completely between backwashing events. On day 119 a 15.2 cm stand pipe was added to the sand filter, partially

submerging the filter with 10.2 cm of standing water and providing constant saturation. Solid matter (34 kg), which had accumulated on the surface of the sand filter, was removed on day 211. Prior to removal of solid matter, backwashing was suspended for six days, allowing the sand filter to dry out and facilitating harvest of the solids.

3.3.2 Fish Stocking and Measurement

Red drum were stocked at an initial density of 2.82 kg/m³ with 200 fish per tank and an average weight of 46.5 g. On day 13 an additional 100 fish were added to each culture tank increasing the total biomass density to 4.23 kg/m³. Fish were sampled monthly to determine average biomass and feed conversion ratio (FCR). All fish were removed from the aquaponic system and held in a separate RAS during experiment days 99 to 115. During this sixteen day period the fish tanks were elevated. When removed on day 99 the biomass density was 26.1 kg/m³,which exceeded the MBBR design criteria. Therefore, when the fish were returned on day 115, the density was reduced to 22.04 kg/m³. Fish were fed a manufactured diet (45% protein, 16% lipid) at 2.6% body weight/day (BWD).

3.3.3 Plant Stocking and Measurement

The saltwater vegetables, sea purslane (Sesuvium portulacastrum) and saltwort (Batis maritima), were stocked over two days on day 0 and day 2. Approximately 7-10 cm long cuttings

of the two saltwater plant species were planted in net pots packed with coconut fiber and supported on polystyrene rafts floating in the raceways. Both plant species and coconut fiber media were selected based on the results of Chapter 2. Two to three cuttings per pot were added to obtain a total planting density of 47 plants/m² and a functional density of 19.5 net pots/m². Two hydroponic plant beds were stocked with saltwort and two with sea purslane. The saltwort was slow to adapt to the system, and one hydroponic plant bed of saltwort was replaced with sea

Figure 3.2: System diagram with sampling locations. Arrows indicate direction of system flows.

Blue lines represent “main system” components and green lines represent “solids treatment”

components.

purslane on day 129. Also at this time, in the remaining saltwort hydroponic plant bed, any dead or small saltwort plants were replaced with cuttings from the surviving saltwort plants.

Plant samples were collected twice monthly during the first 90 days of operation. Three net pots were randomly collected from each hydroponic plant bed using a coordinate grid system

Sample location key:

1. Effluent from fish tanks.

2. Effluent from solids removal (after upflow media filter).

3. Effluent from biofilter tank.

4. Effluent from hydroponic plant beds.

5. Effluent from solids sump.

6. Effluent from sand filter sump.

and the random number generator in Microsoft Excel. Each contained two or three plant samples.

On day 244 and 302, six additional net pot samples of each plant were collected randomly from among the hydroponic plant beds. Each plant was washed carefully to remove debris from the roots, separated into above and below ground weight, and dried at 80°C for 24-48 hours or until a constant weight. On day 108 harvesting of hydroponic plant beds began for sale and

distribution where harvest is defined as trimming 15-40 cm pieces from the top of plants. The sea purslane was more productive and therefore it was harvested more regularly than the saltwort. In general, plant harvests were not structured and occurred when orders were placed by vendors or when labor was available, and at a sufficient frequency to maintain stable plant heights at the top of the hydroponic plant beds. During this period, high quality biomass was bundled into 113 g bunches (0.25 lb) for distribution and sale.

3.3.4 Water Sampling

Water samples were collected in triplicate, in 1 L acid washed (10%HCL) HDPE bottles from six points located throughout the system (Figure 3.2). During the first 83 days, samples were collected once weekly and analyzed for total ammonia nitrogen (TAN), nitrite (NO2-), total nitrogen (TN), total phosphorous (TP), chemical oxygen demand (COD), total suspended solids (TSS), and volatile suspended solids (VSS). Nitrate (NO3-) was analyzed twice weekly during this period. Beginning on day 118, sampling was reduced to twice monthly, for a total of six samples. At this point the biofilter was established and large variations in water quality were not expected. On day 188 sampling frequency was reduced to once monthly, for a total of three samples.

3.3.5 Analytical Methods

All water quality tests were adapted for use with saline water. Standard curves were made with a background salinity concentration of 15 ppt or 1.5 ppt depending on the chloride

interference levels of the test. TAN was analyzed based on the method outlined in Bower and Holm-Hansen (1980) (method detection limit (MDL), 0.04 mg/L TAN); NO2- was analyzed using a combination of Standard Methods 4500-N B and Strickland and Parsons (1972) (MDL, 0.01 mg/L NO2--N); NO3- was analyzed based on Zhang and Fischer (2006) (MDL, 0.15 mg/L NO3--N); TN was analyzed based on a persulfate digestion in Standard Methods 4500-N C (MDL, 1.3 mg/L TN), TP was analyzed based on Standard Methods 4500-P B (MDL, 0.33 mg/L TP); COD was analyzed with Standard Methods 5220 D, an additional 0.5 g of mercury sulfate was added to sample vials to eliminate chloride interference (MDL, 3.1 mg/L COD); TSS and VSS were analyzed based on Standard Methods 2540 D & E. Total iron was measured once, on a grab sample collected from the fish tanks. The method was based on Standard Methods 3500-Fe B. No MDL was determined for this test, although the method suggested a MDL of 0.02 mg/L Fe. The off-flavor compounds geosmin and 2-methylisoborneol (MIB) were measured once, on the same grab samples collected to analyze Fe. Measurements were completed as described in Pettit et al. (2014).

Dried plant samples were finely ground with a burr grinder then analyzed for TN and TP.

TN was analyzed on a TN 3000 Total Nitrogen Analyzer (Thermo Scientific, MA). The NIST standard reference material apple leaves (SRM #1515) were used to create the calibration line and peach leaves (SRM #1547) were used as an accuracy check. TP was analyzed using a persulfate digestion (Standard Methods 4500-P J) and an ascorbic acid colorimetric

determination (Standard Methods 4500-P E). A standard solution of potassium phosphate was used for the calibration line and apple leaves (SRM #1515) were used as an accuracy check.

3.3.6 Statistical Methods

Water quality data were presented as the mean of three replicate samples ± the standard deviation. Statistical analyses were completed with Minitab 16 Statistical Software (State College, PA) where ρ < 0.05 was considered significant for all analyses. Influent and effluent to system components such as the hydroponic plant beds were evaluated with paired t-tests. If necessary data was log transformed to meet assumptions for normality.