Impact assessment of water and nutrient reuse in hydroponic systems using Bayesian Belief Networks

Water-saving agricultural practices can reduce negative environmental impacts in water-scarce regions all over the world. This study deals with an innovation that combines hydroponic crop production and municipal wastewater reuse for irrigation purposes. The research question was what impacts such hydroponic water reuse systems have on product confidence, economic viability, groundwater recharge, biodiversity and landscape quality. It should also be clarified under which conditions and with which measures these systems can be sustainable. To answer these questions, a number of generic hydroponic water reuse systems were modeled and assessed using a Bayesian Belief Network that included both numerical values and expert knowledge. The hydroponic water reuse systems with the most positive overall impacts are small-scale food production systems (tomatoes) equipped with lighting and heating whose products are marked with a quality label or with a label for regional products. The systems are located in a former industrial area. In addition, a wetland system and landscape integration are implemented as landscaping measures. Hydroponic systems can be operated economically viable, their products have a high level of product confidence and their ecological impacts can be positive. No tradeoffs have to be accepted between economic, social and ecological goals.


INTRODUCTION
Agriculture accounts for approximately. 70% of withdrawals of global freshwater resources and is thus by far the largest consumer of water compared with other sectors. In arid and semi-arid areas, the withdrawals can even be as high as 85% (Chmielewski ). About 40% of the agricultural food produced is irrigated (Chmielewski ). Agriculture is dependent on irrigation not only in arid and semi-arid zones.
Even in intensively cultivated agricultural areas in Germany, agriculture is confronted with problems of water scarcity. In total, 451,800 ha (or 1.3%) of agricultural land in Germany are irrigated (Statistisches Bundesamt ). In the German strategy for adaptation to climate change, it was found that regional water use conflicts could arise with regard to surface and groundwater during dry periods (BMU ). Therefore, new water-saving cultivation methods and additional water resources are of interest in Germany in order to avoid conflicts of use, but also for the protection of groundwater. Intensification of agriculture can reduce land use and therefore reduce negative environmental impacts (Ellis et al. ).
One option for the water-efficient cultivation of crops is hydroponic greenhouse production. In hydroponics, plants are grown without soil. Instead, mineral nutrient solutions dissolved in water are used (dos Santos et al. ). Crops are cultivated in such a way that only their roots are hanging in the nutrient solution. The roots can be physically supported by an inert medium such as mineral wool, volcanic stones or other substrates. One possibility of operation is that the nutrient solution continuously flows past the roots (continuous-flow solution culture). This particularly facilitates the adjustment of the nutrient concentration (Rockel ). The nutrients used in hydroponic systems can come from various sources, e.g. fish excrement, chemical fertilizers or nutrient solutions ( Jones ). Crops that are usually grown hydroponically include tomatoes and lettuce, but also ornamental plants. The main advantage of hydroponics is the significantly lower water consumption compared with conventional agriculture (Zhang et al. ). In particular, water loss through evapotranspiration is low in a closed greenhouse. This enables the cultivation of food even in extremely dry areas.
One possible resource for feeding hydroponic systems is the reuse of wastewater. In water reuse, wastewater is treated with technical means, so that it can be reused for other purposes (e.g. agricultural irrigation). The use of recycled water instead of fresh water for corresponding purposes can be a water-efficient measure. Water reuse can thus be part of sustainable water management. This can reduce water scarcity and alleviate pressure on groundwater and other water resources (Andersson et al. ). In addition, not only can water be saved when reusing wastewater, but the nutrients contained in the wastewater can also be used to fertilize the plants.
Although studies have already been carried out on hydroponic systems in which reused wastewater is used, no large-scale implementation has yet been carried out.
The use of treated wastewater as a nutrient solution has so far been little investigated (Magwaza et al. ). In the HypoWave project, a new concept was tested, which deals with the connection of treated municipal wastewater and plant production (Bliedung et al. ). The key question was to what extent municipal wastewater has to be treated in order to be able to use it in hydroponic systems with as little additional nutrient input as possible. A pilot plant with a modular set-up of different wastewater treatment processes was implemented in Wolfsburg Hattorf, Germany, on the site of the municipal wastewater treatment plant (WWTP) (Bliedung et al. ). In addition to the option of using the biologically treated (activated sludge including nitrification and denitrification) WWTP runoff, a number of alternative treatment processes have been tested. These processes include an anaerobic expanded granular sludge bed reactor and an aerobic sequencing batch reactor for nitrification. In this way, a large part of the nitrate remains in the treated water. In addition, a biological activated carbon biofilter for the removal of trace substances and, if necessary, ozonation to eliminate pathogens was used.
Depending on the process combination, water of different quality could be produced for the hydroponic system, which among other things resulted in different nutrient levels.
The hydroponic system of the pilot plant consisted of several parallel rows of pipes in which lettuce plants were cultivated. The pipes were fed with different water qualities.
In addition, lettuce was grown in one pipe in a conventional Hoagland solution with a 50% concentration for comparison. The hydroponic system was designed as a mixture of nutrient film technique and deep water culture.
In addition to the above-mentioned advantages of hydroponic water reuse systems in terms of water and nutrient efficiency, a number of impacts are also unclear, such as questions about the social acceptance of the produced crops, as well as ecological consequences and economic viability. Accordingly, the key research questions are: (1) What are the social, economic and ecological impacts of selected hydroponic water reuse systems? (2) Under which conditions are hydroponic water reuse system socially, economically and ecologically sustainable? To answer these questions, a number of generic hydroponic water reuse systems were modeled and assessed using a software-based modeling process that can make use of both numerical values and expert knowledge.

METHODS AND MATERIALS
General methodology and state of the art The assessment of social, ecological and economic impacts of hydroponic water reuse systems was carried out using Bayesian Belief Networks (BBNs). BBNs are statistical multivariate models based on the theorem of Thomas Bayes (1740) for calculating conditional probabilities.
BBNs combine a qualitative, graphical representation of a system with a quantitative, probabilistic evaluation of the interactions between the variables of such a system The structure of the BBN (i.e. system variables and their interrelations) was pre-developed together with external experts and project partners. For this purpose, a total of six expert interviews with researchers from different disciplines (e.g. civil and environmental engineering, plant science, ecology and landscape architecture) were carried out. Five social, economic and ecological impact variables were jointly chosen, namely product confidence, economic viability, groundwater recharge, biodiversity and landscape quality. The data required to model the BBN were collected through qualitative or semi-quantitative expert surveys in the case of product confidence, biodiversity and landscape quality (Table 1)

Calculation of the economic viability
The economic viability was determined for a total of 16 different generic hydroponic water reuse systems. These systems differed in terms of product (food vs. non-food crops), size (small-scale vs. large-scale), production process (with or without lighting and heating) and the water regime applied (open vs. closed mode of operation). Furthermore, landscaping measures (i.e. wetland systems and landscape integration) were considered in the cost calculation. Additional costs for construction and operation of the landscaping measures, however, were not considered due to the diversity of possible options (e.g. wetland systems, hedges and trees).

Expert surveys and estimations of social and ecological impacts
A total of 12 experts from research and practice were surveyed using questionnaires regarding product confidence, biodiversity and landscape quality, covering different disciplines and sectors. Even if the number of 12 experts cannot be representative, their positions and experience mean that they are multipliers who can provide a wellfounded appraisal of the impacts surveyed.
Consumer acceptance of the food and non-food products made using the hydroponic water reuse system was considered an important factor. The decision of food and flower retailers to add products to their range depends heavily on how they assess consumer acceptance. Therefore, the product confidence of corresponding retailers was chosen as an indicator of consumer acceptance. Additionally, measures to increase product confidence were taken into account. In addition to taking no action, these measures comprise products with a quality label (e.g. products are harmless to health, sustainable or without genetic engineering), products that are labeled as regionally produced as well as branded products (e.g. Bonduelle and Ardo).
The product confidence was determined by using standardized expert surveys. In particular, grocery retailers from organic supermarkets and farmers' markets as well as florists were interviewed. The interviewees were asked to rate the product confidence on a 5-level Likert scale from 1 to 5, with 1 meaning that the respondent would include the product in their range without hesitation and 5 that the respondent would under no circumstances do so.
Two separate questionnaires for two separate groups of interviewees were used for vegetables (tomatoes) and ornamental plants (Chrysanthemum). One question, for example, was: 'Would you include foods (e.g. tomatoes) produced with the described hydroponic system in the assortment of your market if they are labelled as regional products?' The responses were converted to probabilities by first calculating the mean and then performing a linear interpolation where a mean of 1 corresponded to a product confidence of 100% and a mean of 5 to a product confidence of 0%.
The implementation of hydroponic greenhouses can have adverse ecological impacts, for instance increased land consumption or impervious surfaces. Landscaping measures to mitigate these effects can be the creation of wetland systems around the greenhouses and the integration of greenhouses into the landscape by using hedges, trees or extensive farming elements such as short rotation forestry.
Both measures can also be combined and might have positive impacts on biodiversity and landscape quality.

Selected hydroponic water reuse systems and their impacts
The hydroponic water reuse system with the most positive overall impacts is a small-scale food production system (3-6 ha, tomatoes) equipped with lighting and heating whose products are marked with a quality label ( Table 2).
The system is located in a former industrial area, and both above-mentioned landscaping measures (wetland system and landscape integration) are implemented. With regard to the investigated impacts, it is irrelevant whether it is a system in which the water is circulated (closed system) or through which the water flows only once (open system).
Such a system would achieve a very high product confidence (90%), a high economic viability (6% rate of return), a greatly improved groundwater recharge, a very positive impact on biodiversity (77%) and a very positive effect on the landscape quality (84%). A slight variation in this system in which the tomatoes are marked with a label for regional products would achieve a slightly lower product confidence (70%). Apart from that, the system achieves the same very positive results. Small-scale food production systems without lighting and heating, however, have a very poor economic viability (À12% rate of return).
Large-scale food production systems (tomatoes with quality label) in former industrial areas also perform fairly well but only without lighting and heating (Table 2). In this case, the economic viability is just positive (Table 2).  Apart from this, the large-scale system scores slightly worse than the small-scale system with regard to the criteria biodiversity (76 instead of 77%) and landscape quality (80 instead of 84%) ( Table 2). Large-scale food production systems with lighting and heating, however, do not appear to be economically viable (rate of return À5%, Table 5).
As in the case of small-scale systems, product confidence drops from 90 to 70% if the tomatoes are labeled as regionally produced. Food systems without consumer acceptance measures (regardless of whether they are large-or small-scale) lead to a loss of product confidence (56 instead of 90 or 70%). Branded products even lead to a product confidence of only 42%.
In the case of food production systems in former agricul- regime cannot be drawn either for food production systems or for systems for the production of ornamental plants.

Social impacts
The highest product confidence is found in ornamental plants (Chrysanthemum) with (100%) and without (92%) consumer acceptance measures (quality label and/or labeled as regionally produced) as well as in food products (tomatoes) with quality labels (90%) followed by ones that are labeled as regionally produced (70%). In general, food retailers were positive about potential products from hydroponic water reuse systems. They said that they could imagine including the products in their range. The best measure to improve the product confidence was a quality label which can be compared with labels that certify sustainable production or GMO-free labels. Labeling as a regionally produced product also improved product confidence.
Branded food products even had a lower product confidence (42%) than products without any consumer acceptance measure (56%). A sensitivity analysis of the factors influencing product confidence shows that the influence of acceptance measures is significantly greater than that of the product type.
Florists did not mention concerns with selling ornamental flowers produced with hydroponic water reuse systems.
The use of labels (quality label or labeling as a regional product) was preferred over no label. However, it was also argued that labels are not required in case of ornamental plants since customers would not be concerned. They were said to prefer quality and price over other factors.
In some cases, it was said that costumers would ask where the flowers were made. This might indicate an interest in regional products and their social or environmental impacts.
The results indicate that the production of flowers from hydroponic water reuse systems would be well accepted.
Flowers labeled as regionally produced could even influence the consumer's purchase decision in a positive way if they are interested in the social and environmental sustainability of the production. Apart from this, the attitude of grocery retailers towards food crops from hydroponic systems was predominantly positive. Nevertheless, quality labels or labeling as a regional product might be required to increase product confidence. Big supermarket chains could not be reached by means of this study. Further research is necessary to understand how the big players will position themselves towards the circular economy and future developments in food production. To our knowledge, there are no other detailed studies on product confidence concerning products from hydroponic systems with water reuse.

Ecological impacts
Implementing both landscaping measures (i.e. wetland system and landscape integration) leads to the most positive impacts in terms of biodiversity and landscape quality. This is irrespective of whether the hydroponic system was set up in a former industrial or agricultural area or whether it is a small-scale or large-scale system (Table 3)

Economic impacts
As mentioned above, good economic viability (6% rate of return) is associated with small-scale food production systems with lighting and heating (  Table 5 are based on rather conservative assumptions.
A further sensitivity analysis was made with different assumptions considering energy supply and market prices.
If prices and energy production are used as initially suggested by the model, six scenarios are highly profitable (15-35%), further four scenarios are positive, two scenarios are negative, and four scenarios are very negative (À6 to À13%).
Furthermore, a combined heat and power plant for the production of heat and electricity would result in a surplus and therefore an income from sales of electricity. The market prices for Chrysanthemum (with lighting and heating) were assumed to be 0.38 EUR according to the model.
Costs for disposal of residues from water treatment were not considered in the calculations. In systems with closed water regimes, reverse osmosis is suggested to recycle the greenhouse runoff. It would be possible to irrigate a short rotation plantation in the landscape park which is much less sensitive to salinity than crops in hydroponic systems.
In this way, water treatment costs and the question of disposal of residue would not arise.

CONCLUSIONS
The results of the study show that hydroponic water reuse systems can be operated in an economically viability way, that their products have a high level of product confidence and that their ecological effects can be positive if appropriate landscaping measures are undertaken. This applies in particular to the small-scale food production system described above. The results mean that no tradeoffs have to be accepted between economic goals on the one hand, as well as social and ecological goals on the other. In order to achieve this, however, the results suggest that landscaping as well as acceptance measures should be carried out to accompany hydroponic crop production. Land use through the construction of greenhouses should be Small-scale Open system 3 ha 3 ha 3 ha 3 ha 6% À12% À17% À12% Closed system 6 ha 6 ha 5 ha 5 ha 6% À12% À19% À13% compensated in particular by measures such as wetland systems and landscape integration in order to avoid or at least mitigate undesirable effects on the environment.
The BBN method allowed a rough, but nevertheless comprehensive and quick overview of the interdependencies of generic hydroponic systems. On the one hand, the modeling of generic systems has the advantage that the results can offer a general orientation and can be transferred to a specific case under certain conditions. On the other hand, the circumstances of a specific case can be such that the effects of the modeled systems might not apply. From a business perspective, however, such a model can serve as a decision support for companies (e.g. horticultural companies) or investors for a pre-feasibility study. The advantage of using a method such as the BBN is that quantitative assessments (e.g. cost-benefit analysis) are expanded to integrate qualitative expert knowledge. This means that impacts that are difficult to assess and to quantify can be included, which are often neglected in purely quantitative assessments.
Further investigations could address questions that remained open. One of these aspects is to clarify how an adequate quality management for hydroponic crop production including water reuse should be. In connection with this, the possibilities for the certification of the products with regard to consumer acceptance could be examined in more detail. Another aspect would be to include further assessment criteria such as measuring the carbon footprint in order to get a more complete picture of the impacts of hydroponic systems. A life cycle analysis could, in particular, provide insights into energy and material consumption for water reuse and hydroponic crop production in greenhouses. Unfortunately, this was not possible within the scope of the study. Nevertheless, the findings of this work may contribute to a successful take-off of the discussed innovation and the associated transformation processes in agriculture and water management.