A simple tool to help decision making in infrastructure planning and management of phytotreatment ponds for the treatment of nitrogen-rich water

In situ experimental studies were carried out aimed at the quantitative estimation of biological processes involved in nitrogen removal such as macro-algal assimilation and bacterial denitrification and their optimisation in two experimental phytotreatment ponds colonised by the macro-algae Ulva rigida in central Italy. Results from an in situ manipulative experiment estimate that Ulva carrying capacity defined as the macro-algal biomass in which the uptake of dissolved inorganic carbon (DIC) equals the production of oxygen (O2), was close to 300g·m -2 dry biomass (dw). At this carrying capacity the experimental assessment of Ulva growth rates and Ulva assimilation rates and their optimisation with use of a logistic model estimated that maximum inorganic nitrogen removal (~0.04 mol N·d-1·m-2) was attained when Ulva biomass reached 150 gdw·m -2 and growth rate was 0.1·d-1. Denitrification rates accounted for a small amount of total nitrogen removal (~150 μmol N·m-2·h-1) although an intact core incubation experiment demonstrated that denitrification increased with increasing nitrate concentrations. Based on experimental results a series of calculations have been made by use of MATLAB algorithms to facilitate manipulation of easy-to-measure variables (infrastructural, chemical and biological) and subsequent gross estimates of their effect on biological nitrogen removal efficiency, thus providing a simple tool to help decision making for infrastructure planning and management of phytotreatment ponds.


Introduction
Phytotreament ponds with foliose macro-algae are widely employed for the treatment of ammonia-rich water due to their simple arrangement (not needing complex infrastructure) and low cost (Redding et al., 1997;Brix, 1999;Tanner, 2001;Vymazal et al., 2001;Lin et al., 2002;Schulz et al., 2003).In terms of Italian Government law, the construction of phytotreatment ponds is compulsory for wastewater treatment under particular circumstances and various activities, the main being land-based aquaculture.The aim of phytoremediation ponds is the removal of dissolved nitrogen from wastewater to be safely discharged into the natural environment.Foliose macro-algae are primarily utilised for this purpose and Ulva rigida is one of the most common due to its high net growth rates, nitrogen uptake capacity and resistance to environmental stress (Laliberté et al., 1994).Coupled to dissolved nitrogen assimilation by primary producers, removal of particulate matter must be previously achieved via sedimentation pond in order to fully exploit macro-algae assimilation (Bartoli et al., 2005) and prevent deterioration of microbiological quality of the treated water (Vezzulli et al., 2005).Despite the potential usefulness of sedimentation/phy-totreatment pond system for the treatment of nitrogen rich water one of the main reasons limiting their full exploitation is the lack of suitable methods to help decision making in pond infrastructure planning and pond management.Failing to observe these rules limits the usefulness of the pond and may lead (e.g.due to macro-algal collapse in an un-managed pond) to further deterioration of the wastewater.Therefore, the functionality of phytoremediation ponds is based on structural variables to be planned before construction (pond area, depth and water flow) and biological and chemical variables (macro-algae biomass, dissolved ammonia and nitrate concentrations in inlet water) to be experimentally estimated in order to optimise biological nitrogen removal (macro-algal assimilation and denitrification) through a correct management.Both of these variables are strongly correlated and estimation of their values and optimal setting requires expensive investigation often resulting economically unaffordable for the end-users.To provide a solution to this problem we developed a user-friendly tool able to facilitate manipulation of these variables (structural, chemical and biological) and provide subsequent gross estimates of their effect on biological nitrogen removal efficiency, thus providing a tool to help decision making for planning and management of phytotreatment ponds.The tool is based on a series of experimental studies aimed at the quantitative estimation of biological processes involved in nitrogen removal such as macro-algal assimilation and bacterial denitrification and their optimisation as a function of structural and chemical pond variables.The setting-up of the tool has been carried out during late spring (May 2002(May , 2003) ) in two experimental phytotreatment ponds for the treatment of aquaculture wastewater located in central Italy.These ponds were selected because they are representative of the average infrastructural and environmental features found in most Italian aquaculture ponds being characterised by: • A mean depth of 1 m • Presence of the macro-algae Ulva rigida growing spontaneously • Spring/summer temperature range of 15 to 30°C • A water pH range of between 7 and 8.

Experimental Experimental phytotreatment ponds
The 'Falesia' Fish Farm (Piombino, Italy) annually produces approximately 200t of gilthead sea bream (Sparus aurata) and sea bass (Dicentrarchus labrax) (Bartoli et al., 2005).Wastewater from the fish tanks is released into a lagoon system comprising three ponds connected in series.Each pond has a surface area of ~600 m 2 for a total surface of ~2 000 m 2 and a mean depth of ~1 m.Pond 1 is used for sedimentation purposes whilst Ponds 2 and 3 (~1 300 m 2 ) are colonised by a natural community of the green macro-algae Ulva rigida.Physico-chemical properties of wastewater entering the ponds are reported in Table 1.
The 'Nassa' Fish Farm (Orbetello,Italy) annually produces approximately 100t of gilthead sea bream (Sparus aurata) and sea bass (Dicentrarchus labrax) (Bartoli et al., 2006).Wastewater from the fish tanks is released into a lagoon system comprising four ponds connected in series.Each pond has a surface area of ~2 600 m 2 for a total surface area of ~10 000 m 2 and a mean depth of ~1 m.Pond 1 is used for sedimentation purposes whilst Ponds 2-4 (~8 000 m 2 ) are colonised by a natural community of the green macro-algae Ulva rigida.Chemical-physical properties of wastewater entering the ponds are reported in Table 1.

Determination of Ulva carrying capacity
Evaluation of Ulva carrying capacity defined as the macro-algal biomass in which the uptake of dissolved organic carbon (DIC) equals the production of oxygen (O 2 ) was measured by an in situ manipulative experiment carried out on Pond 3 at the Falesia Fish Farm which was fully colonised by Ulva rigida.Three consecutive 24 h cycles of investigation were started on May 2003 in this pond.In order to investigate the effect of different macro-algal biomass approximately 20% of fresh biomass was rapidly removed (in 1 h ) at the end of Cycles 1 and 2 and a 6 h interval was taken before starting a new sampling to allow for pond stabilisation.Ulva biomass was estimated before the beginning of each investigation cycle.Eight replicates were randomly collected in Pond 3 with a cylinder (internal diameter 700 mm, height 1 200 mm) trapping the algae along the whole water column.The macro-algal biomass within the cylinder was then collected, gently washed with pond water and weighed fresh and after oven desiccation at 70°C for 4 to 6 h.Ulva for fresh weight determinations was harvested from the pond with long rakes and packed in a series of perforated tanks generally used for fish harvesting.All the material was drained to remove most of the water and then weighed.Estimated macro-algal biomass was ~8 kg•m -2 at Cycle 1, ~6 kg•m -2 at Cycle 2 and ~4 kg•m -2 at Cycle 3. Inlet and outlet pond water were sampled over the three cycles approximately every 3 h for a total of 48 collected samples and analysed for temperature, pH, salinity and oxygen with an YSI 556 multiple probe.Water samples (200 mℓ) were filtered with GF/F Whatman filters within 1 h of collection and immediately frozen for later analyses.Filtered water was analysed for ammonia (NH 4+ ) (Bower & Holm-Hansen, 1980), NO 2 - (Golterman et al., 1978), NO 3 -as NO 2 -after reduction with cadmium and dissolved inorganic carbon (DIC).Ten mℓ of unfiltered water were analysed for total dissolved inorganic carbon (DIC) by titration with 0.1 M HCl (Anderson et al., 1986).

Ulva growth rates (UGR)
Ulva growth rates were experimentally assessed from Ulva samples collected from the Nassa and Falesia experimental ponds in May 2002 and May 2003 respectively.Fresh thalli were gently washed to remove epiphytic material and cut into ~300 discs (diameter = 50 mm); 20 discs were individually weighed fresh and after desiccation in an oven at 70°C whilst 4 series of 50 discs were suspended in 4 cylindrical cages (internal diameter 200 mm, height 400 mm) and positioned within the Ulva mats at a density close to the carrying capacity (250 g dw •m -2 ).Three days later the cages were removed and all the discs were gently washed, dried and weighed for the determination of growth rates.

Denitrification rates (DR)
Denitrification rates and correlation with nitrate concentrations were measured by means of bare sediment incubations carried out only in the dark at the Nassa experimental pond with approximately 200 to 300 g dw m -2 macro-algal biomass.Increasing amounts of labelled nitrate (final concentration comprised between 20 and 150 µM) were added to the water phase of the cores for an evaluation of the sediment denitrification potential.

607
The rates of denitrification (D15) were calculated according to the equations and assumptions of Nielsen (1992).

Ulva carrying capacity
Results from the in situ manipulative experiment showed the effectiveness of the removal of macro-algal biomass on the phytotreatment pond changing from a strong heterotrophic condition to almost autotrophic (dissolved inorganic carbon [DIC] and oxygen [O 2 ] daily balance shifted from 2 686.5 to 1 066.5 mol DIC•d -1 and from -518.6 to -13.0 mol O 2 •d -1 for Cycle 1 and Cycle 2 respectively) (Fig. 1).At the end of the 3 rd cycle the pond became almost a sink for NH 4 + (11.4 mol NH 4 + •d -1 ) and the carrying capacity defined as the macro-algal biomass in which the uptake of DIC equals the production of O 2 was identified as 300g dw •m -2 of macro-algal biomass.

Assessment and optimisation of Ulva growth rates (UGR) and Ulva assimilation rates (UAR)
Close to the carrying capacity (250 g dw •m -2 ) weight of dry Ulva discs augmented on average of ~2.3 mg•d -1 in the cage compartments (0 to 40 cm depth), resulting in a specific growth rate of about 4%•d -1 .Nitrogen assimilation by Ulva has been calculated indirectly by growth rates experiments assuming for healthy algae growing in a not limiting environment a constant nitrogen content of 4% dw.Based on experimental data, the temporal evolution of macro-algal biomass and nitrogen removal from water has been simulated with a logistic model assuming an instantaneous growth rate of 0.2•d -1 and the in situ estimated carrying capacity of 300 g dw •m -2 .The amount of nitrogen removed per pond per day was then calculated as a function of the in situ Ulva biomass.With the chosen logistic model and in situ parameters we calculated that the maximum inorganic nitrogen removal (~0.04 mol N•d -1 •m -2 ) was attained when Ulva biomass reaches 150 g dw •m -2 and growth rate is 0.1•d -1 (Fig. 2).

Denitrification rates (DR)
Denitrification rates accounted for a small amount of total nitrogen removal at the Nassa experimental ponds with in situ values of ~150 µmol N•m -2 •h -1 .However, results from the concentration series experiment clearly demonstrated that denitrification increased at higher 15 NO 3 -denitrification (D 15 ) concentrations and a saturation value cannot be calculated being the highest D15 value (>800 µmol N•m -2 •h -1 ) still linearly correlated with the water column 15 NO 3 -.The relationship occurring between 15 NO 3 concentrations (0 to 150 µM) and denitrification rates were modelled using linear regression and resulted in the following: .8; r 2 =0.94.(Fig. 3)

Discussion
Considering the optimal condition of 150 g dw •m -2 of macro-algal biomass, 0.04 mol N•d -1 •m -2 Ulva assimilation rates and 0.1•d -1 These are calculated using structural and chemical input variables from ponds: The current version is set up for modelling phytotreatment ponds having a depth of ~1 m, optimal macro-algal biomass of 1.5 kg•m 2 and environmental conditions characterised by summer temperature ranging 15 to 30°C, daylight ranging between 500 and 2 000 μE•m -2 •s -1 , water pH ranging between 7.0 and 8.0 and a certain nitrification potential to allow for ammonium oxidation to nitrate (Bartoli et al., 2005).Adaptations of the tool to different environmental conditions and seasonal variations can be achieved by setting up key variables such as Ulva growth rates and Ulva assimilation rates on the basis of newly collected experimental data.Finally, to facilitate access in a user-friendly way the MATLAB version of the tool has been translated into a Windows compatible format (Ulva.xls,MS Excel)(Fig.4) and is available upon request from the Department of Biology of the University of Genoa.

Conclusions
Figure 4 shows the output of the developed tool that has been run using input variables measured at the Falesia and Nassa phytotreatment ponds respectively.At the Falesia Fish Farm even following an optimal management strategy (~195 kg fresh macro-algal biomass to be removed daily) the estimated total nitrogen removal accounts for less than ~2% of the total nitrogen inlet, with macro-algal assimilation representing the main removal processes.The low remediation potential of the Falesia system is depending upon the strong unbalance between structural and chemical pond variables such as the small area (1 300 m 2 ), the high water flow (250 ℓ•s -1 ) and the high nitrogen load (179 μM) in wastewater.In contrast at the Nassa fish

609
farm the estimated total nitrogen removal in the managed pond (~1 200 kg fresh macro-algal biomass to be removed daily) may account for up to ~50% of total inlet due to the higher pond area (8 000 m 2 ) and the reduced nitrogen load in wastewater (61 μM).Again macro-algal assimilation constitutes the major removal processes while denitrification only accounts for a small percentage (~1%) of nitrogen removal due to the low nitrate concentration in pond water.By manipulating the input variables and running the tool the user can rapidly calculate the best conditions in order to obtain 100% ammonia removal from wastewater in the two ponds.As a general estimate the surface of phytotreatment ponds needed for 100% ammonia removal in wastewater using the macro-algae Ulva rigida can be calculated as being as least ~50 m 2 /pe for a load per capita of ~30 gN•d -1 under optimal conditions.This estimate can be further improved if we consider an increase of denitrification rates, for example, by stimulating nitrification rates in pond water (e.g.oxygen supply).Of course this corresponds to the daily harvesting of ~150 g•m -2 dry macro-algal biomass that has to be managed as well (150 g•m -2 dry macroalgal biomass corresponds to 1500 g•m -2 of fresh biomass).The management of harvested Ulva is a matter of debate up to now and its use for consumption by humans, by secondary cultured macro-algivores (such as abalone and sea urchins) and by other fish, has been proposed recently (Neori et al., 2004).
Successful commercial exploitation of biological systems relies on minimising development time and cost while simultaneously delivering substantial process efficiencies.The tool we have developed enables this concept by providing improved strategies for the optimisation and control of bioprocesses involved in nitrogen removal by phytotreatment pond.Recently a conceptual model for the application of bioremediation strategies in organic-rich ecosystems has been developed and is based on the mobilisation (decomposition of insoluble organic polymers) and removal of macro-elements (e.g.carbon and nitrogen) from accumulation areas (Vezzulli et al., 2004).Among systems for biological removal the use of foliose macro-algal material is considered a primary candidate, thus enhancing the potential interest of the tool we have developed within the field of applied biotechnology.

Figure 3 Figure 1
Figure 3Rates of 15 NO 3 -denitrification (D 15 ) as function of labelled nitrate concentration in the water column (fromBartoli et al, 2006)