Impacts of floating urbanization on water quality and aquatic ecosystems: a study based on in situ data and observations

Climate change, population growth and increasing urbanization caused land scarcity worldwide, especially in metropolitan areas (GCA 2019; IPCC 2019). By 2050, more than 80% of the world population are expected to live in cities and 50% of people are expected to live within 100 km from the coast (Roeffen et al. 2013). The growing agglomeration of the population in urban areas leads to a continuous process of the conversion of rural land into urban land, suppressing water infiltration and creating imbalances in the urban water cycle. Additionally, rising sea levels and more extreme weather conditions due to climate change increased the vulnerability of populations and led them to be exposed to an increased flood risk in urban centres.

As a potential solution to address these challenges, floating architecture is gaining significance (Wang & Tay 2011; Silva & Costa 2014; Dafforn et al. 2015). Floating buildings could offer a solution for the future lack of building ground and reduce the vulnerability of cities to flooding by allowing water storage (Stopp & Strangfeld 2010), or provide opportunities for a wide range of multifunctional uses (Dal Bo Zanon et al. 2017), or for green/solar energy generation (Sahu et al. 2016; World Bank Group 2019). In The Netherlands, the development of floating projects is already present for some years (e.g. houseboats), including a few already-built floating communities (e.g. floating community at IJburg, Amsterdam), which are being built in unused water bodies within cities (e.g. old harbours).

Research studies addressing the impact to the aquatic ecosystems when water bodies are covered by floating structures are scarce. Water authorities currently struggle with the lack of knowledge regarding potential environmental effects, which often results in delay or even cancellation of floating solution projects (Silva & Costa 2014). To safely expand the use of floating developments, there is a need for further research on this topic that can support policies and guidelines for measures that are able to compensate possible undesired changes to the condition of ecosystems.

Aquatic systems provide water for consumption, food production, energy generation and aquatic organisms. Aquatic systems also regulate water quality and erosion and offer an attractive environment for recreation and tourism. The loss of any of these functions would result in a decrease in human well-being and ecosystem services. Most ecosystems are already heavily impacted by human activities or are completely artificial/man-made (e.g. port areas, canals). This is particularly evident in The Netherlands, where natural ecosystems underwent impactful land reclamation processes over the centuries. The environmental impact of floating structures involves complex interactions between physical, chemical and ecological components of the water system, ecosystems and human activities, and is highly dependent on local conditions of the aquatic environment. The deployment and use of floating buildings and platforms introduce changes to aquatic systems that can affect the environment negatively, but can also generate various benefits and opportunities for ecology development. Figure 1 shows a conceptual model that illustrates these changes, which are further discussed in Table 1. To make a realistic and responsible evaluation of floating urban development, it is important to consider the potential negative impacts associated with alternatives, such as conventional land reclamation (based on dredging) or land-based development. These alternatives often completely erase the entire aquatic ecosystem, or lead to other problems such as urban flooding due to the increase of paved surfaces and reduced groundwater infiltration.

Over the last few years, a number of studies reported monitoring campaigns to collect in situ data near floating structures, or a comparable infrastructure with varying characteristics. Burdick & Short (1999) investigated the growth of eelgrass beds under floating boat docks and reported severe localized impacts caused by a combination of shading caused by the docks as well as dredging/disturbances caused by boat propellers. Complete eelgrass loss under and around many docks was observed in some cases, and models indicated that less severe impacts from shading are expected under tall and narrow docks due to limited light intensity. It was also reported that the large-scale platform with small draft (around 1 m) had little impact in the surrounding flow conditions (direction, flow rates and the stratification monitored). Huguet et al. (2020) studied the influence of floating structures on the hydrodynamics of a marina, namely how tidal generated eddies are affected (attenuation). These processes affect the dispersion resuspension of organic matter and sediments, which have implications for water quality or siltation problems. A numerical hydrodynamic and ecosystem simulation of the Mega-Float in Japan indicated that the impact of the structure on flow velocity, currents, water quality parameters and in water temperature is minimal (Kyozuka et al. 2001). The introduction of a breakwater causes water stagnation, which results in increased impacts on water quality.

Kitazawa et al. (2010) assessed the environmental impact of a large temporary floating offshore airport runaway. Physical and chemical–biological parameters were measured at two monitoring stations over a 1-year period. Slight changes in the concentration of dissolved inorganic phosphorus, oxygen, nitrogen and chlorophyll-a were localized within the first 5 m away from the structure and were attributed to the colonization of the platform by sessile organisms and the inhibition of photosynthesis under the platform. The impact of floating structures is suggested to be larger in more stagnant and eutrophicated waters. The concentration of dissolved oxygen was slightly lower in the deeper column below the platform but did not reach hypoxic or anoxic levels. Slavik et al. (2019), with a strong focus on (large-scale) offshore wind turbines parks, also reports an important increase in the accumulation of filter feeders on the structures, with consequences for the pelagic primary productivity and ecosystem functioning as a whole by affecting food sources, creating new habitat or removing phytoplankton through water filtration. Alexander & Robinson (2006) measured quantifiable effects on the benthic environment resulting from the presence of floating docks, including a reduction in the numbers of organisms and biomass, loss of primary productivity (lower chlorophyll-a, attributed to shading), changes in grain size of the sediment (coarser sediments under the docks) and associated organic/nitrogen content. Physiochemical parameters and macro-zoobenthic samples near floating cage farming were analysed by Klaoudatos et al. (2006). Water flow heavily dissipated/dispersed the nutrient-rich outflow from the fish farms, resulting in little impacts in the surrounding environment (higher nitrogen and lower phosphates concentrations at the farm sites). The accumulation of organic matter-rich sediments on the bottom under the cages creating an anoxic flocculent environment, as well as clear seasonal variability for most variables, is reported. Holloway & Connell (2002) highlight differences between fixed and floating structures regarding the conditions for habitat development near the top water layer. In fixed structures, the water level on the vertical surfaces may change with tides or other variations in the water level, while in the case of floating platforms, the water level remains the same due to its floating nature. Foka (2014) analysed the impact of floating houses on the dissolved oxygen content and observed about 10% lower oxygen levels close to the floating houses (at a zone confined between different floating houses), when compared with open water. The reported differences occurred mostly in the upper layers (<1 m depth) and varied throughout the day (differences fade overnight). Regarding water temperature, the detected differences were found to be low.

The rapid implementation of novel large-scale floating solar parks worldwide generated a need for updating and adjusting legal policies and guidelines, and therefore several studies have been conducted recently that address potential impacts of floating solar panels (Armstrong et al. 2020; Haas et al. 2020; Ziar et al. 2020). Some of the studies report findings based on in situ measurements (Costa 2017; Pimentel Da Silva & Branco 2018; Ziar et al. 2020), but often rely on models without validation with in situ data. Recent monitoring campaigns in The Netherlands that investigated the environmental impacts of a floating solar panel prototype did not detect significant changes in water quality parameters (Ziar et al. 2020), such as total nitrogen, water temperature, total phosphorus, chlorophyll-a or cyanobacteria concentrations, but does report changes in plant biomass and an increase in the frequency of hypoxia conditions (concentrations of dissolved oxygen lower than 6 mg/l). Similarly, de Lima et al. (2021) and Château et al. (2019) reported no major differences in the measured key water quality parameters such as electrical conductivity or temperature below the solar panels at various water depths. The differences were mostly detectable at the upper layers as it takes a longer period for the water under the panels to warm up or cool down. A few other studies address the reduced evaporation in lakes, which could be beneficial in locations prone to drought or water shortages (Gorjian et al. 2021), or the leaching of substances/heavy metals (Mathijssen et al. 2020).

The objective of this paper is to contribute a better understanding of how floating urbanization solutions affect water systems, to serve as a support for water authorities and municipalities to create a policy framework, and to regulate and facilitate the development of new projects. This paper reports findings based on the collection of in situ water quality data underneath several floating buildings of different types and scales. The monitoring campaigns took place in The Netherlands over a period of 6 years (2014–2020). The research focuses on aspects such as the effect on dissolved oxygen levels, water temperature or aquatic life. Opportunities and recommendations for new designs of floating urbanizations and infrastructure projects that could potentially avoid or minimize undesired environmental changes and the loss of ecosystem services are also discussed.

Dissolved oxygen, temperature and (visual) ecological indicators were selected as the key parameters for this research. These parameters are critical in most aquatic processes, as they affect water quality cycles and the health of aquatic life and can be highly impacted by the presence of floating constructions. Dissolved oxygen can be affected by reduced photosynthesis due to shadow/blocking of direct sunlight, changes in turbulence/currents and (wind-related) mixing, reduction in the area available for aeration and air–water interactions or the increase of oxygen demand for decomposition of organic matter due to heavy sessile colonization of surfaces. Temperature changes may occur due to reduced vertical heat exchanges between air and water, reduced exposure to solar radiation, heat exchanges between the (heated) houses and the water. Finally, with regard to ecology, floating houses may cause changes in the behaviour of fish and benthic species (also linked with water quality and temperature conditions), limit the growth of fauna and flora due to the obstruction of light penetration or affect biodiversity by creating conditions for invasive species. The ease of measurement of these parameters with sensors that can collect data at high frequency and that can be deployed in the field for a long period of time was also a decisive factor. Other parameters such as nutrient levels (e.g. nitrogen/phosphorous), pollutants or bacteriological parameters still require the collection of samples and laboratory analysis, which are costly and not suitable for long-term continuous monitoring.

For this study, multiple locations where different types of floating structures are located were selected for water quality and ecology measurement campaigns. Table 2 shows an overview of the characteristics of the floating constructions and of the water systems where they are located. The locations vary from small ponds with little water circulation (e.g. location 1: Harnaschpolder) to larger water bodies connected to streams (e.g. location 13: Maasvillas). Different methodologies used in selected locations were selected for specific measurements and methodologies (also indicated in the last column of Table 2). The selection of the three locations for the installation of the fixed continuous sensors considered aspects such as safety of the equipment (risk of theft of equipment) and representativity of floating buildings and of the type of water system (e.g. minimum water depth and hydrodynamics).

This study compares water quality parameters measured under/near floating structures with the corresponding parameters measured at the same depth at an open water location. The latter is located at the same water body, away from the influence of the floating object (minimum of 10 m away from the floating object). Data were collected by (i) performing scans with underwater drones equipped with in situ sensors and video cameras during a short period of time, which collected water quality data from the three positions (open water and near/under the floating buildings) and (ii) fixing two sets of continuous measuring in situ sensors for a period of several days/months at both positions. Figure 2 schematizes the position where the continuous sensors were installed, and the areas that were scanned with underwater drones.

The small underwater remotely operated vehicles (ROV) used in this project belong to the mini/exploration class, taking into account their weight and dimensions. These vehicles were equipped with the same sensors that were used for the continuous measurements (pressure, temperature, electrical conductivity and dissolved oxygen). Polystyrene blocks were used to adjust the buoyancy to compensate for the extra weight of the payload. The position of the drone under the floating building was estimated based on tether length and angle of entry, as well as from an onboard pressure sensor (depth). The drones had semi-autonomous features that allow them to remain at constant depth (based on pressure sensor data), allowing the drone to remain at each position for the desired time.

The underwater drones were remotely operated to collect data from the three positions (open water and near/under the floating buildings). The data were collected at similar water depths, even at different locations, varying between 1.25 and 1.75 m of water depth. At each site, the measurements were taken within a maximum of 3 h due to battery limitations. Some locations were visited, more than once, in different days/seasons. The sensors/loggers were set to collect measurements at a high frequency measurement interval of 5 s. However, the relatively slow response time of sensors, such as dissolved oxygen, requires that the drone remains at each position for several minutes to allow the values to stabilize.

After the fieldwork campaigns, the data were organized by study location, averaged for each position and used to build comparison chart plots with the detected differences. Statistical tests (paired t-tests) were applied to assess the statistical significance of the results.

To fix the sensors underneath the floating houses, a sensor (and float) was attached to a tensioned cable from both sides of the building (Figure 2). Underwater drones were used to transport the cable across the floating buildings, which was then used to position the sensor approximately underneath the centre point of the building. At the open water location, a mooring system involving a buoy and anchor was used to fix the cable with sensors (deployed from a boat). The two sets of sensors were placed at approximately the same water depth (1.5 m, positioned immediately after the draft of the floating building) and collected data simultaneously from the two positions. The sensors measured parameters such as pressure (depth), temperature, electrical conductivity and dissolved oxygen. The frequency of the measurement intervals selected varied from 15 s to 5 min, depending on the intended duration of the measurement campaign at each location in relation with the battery/storage capacity of the loggers. Considering that meteorological conditions, such as wind or temperature, are important driving factors for mixing of the water column and on the water temperature, meteorological data were retrieved from nearby (public) weather stations, or from a weather station installed on the location by the research team (Leeuwarden).

Continuous sensors were installed at the following four different locations: (i) Floating Pavilion (location 15) between August and October 2020 (43 consecutive days); (ii) Himpenser Wielen (location 2) for 5 days in September 2017, (iii) IJburg (Steigereiland; location 3) and (iv) Drijvende Kas (location 6), both for a duration of 3 days in May 2014. Continuous monitoring also took place at three other locations, although for shorter periods of time. After the field measurements, and during the processing/analysis of the data, statistical tools such as boxplots and paired t-tests were used for a better characterization of the datasets and to assess its variability and statistical significance.

In addition to water quality sensors, the underwater drones were equipped with video cameras to collect underwater footage of the aquatic ecosystem in the vicinity of the floating structures. The drones included onboard cameras (HD; 1080p), which were mostly used for navigation purposes, and that could be tilted 90° in order to look up or down. Additional cameras were installed on the drones, enabling the capture of higher-quality underwater images and the positioning of the cameras facing the desired direction. The cameras were aimed upwards to inspect the biofouling on the surface underneath the platforms, or downwards to record the benthic ecosystem. These cameras have a wide angle of view (up to 120°) and are enclosed inside watertight casing with a depth rating of 40 m. These specifications were sufficient for this research, where a maximum depth of 8 m was reached near the visited floating buildings. To ensure adequate illumination conditions, the underwater drones were also equipped with two sets of high intensity (dimmable) diving lights of up to 1,500 lumens.

Most of the underwater filming took place during the spring/summer months between April and September, to ensure that macrophytes are already visible. Following the recording of the images, the footage was trimmed in clips, and a combination of video and image editing software was used to extract and enhance stills/screenshots of the surfaces.

The different types of equipment used in this study are listed in Table 3 and Figure 3. These include the various water quality sensors and loggers that were installed at the study locations for continuous measurements of water quality, the underwater drones that were used as a platform to position the water quality sensors underneath the floating houses and the underwater camera system that was coupled to the unmanned vehicles that provided visual insights into the condition of the aquatic ecosystems.

Figures 4 and 5 show the results of the scans with the underwater drone. In Figure 4, three values are shown for each study-site location, corresponding to the average of the values (temperature and dissolved oxygen) measured by the sensors underneath the floating structures, near the floating structures and at open water. The standard deviation for each value is also represented in the graph. The data show that different locations have different temperature and dissolved oxygen levels, which are logical considering the varying weather conditions and season. In most cases, differences between the three zones were detectable and characterized by lower dissolved oxygen concentrations underneath the floating structures. Low dissolved oxygen levels were found in Oosterdok (location 10; both under the floating building and at open water), which are linked with overall poor water quality of the water body. The recorded temperatures at the three positions showed little variation in most cases, but in three to four locations the differences were detectable, with slightly warmer water found underneath the floating buildings. Paired t-test results between temperature and dissolved oxygen data from underneath the floating structures and from open water show high correlations between the tested pairs, but only the mean differences of dissolved oxygen are statistically significant (95% confidence interval of the differences of dissolved oxygen concentrations between −0.95897 and −0.24982 mg/l).

Figure 5 plots the differences of dissolved oxygen concentrations and of temperature between the values measured at the reference point and the values measured under the floating structures and at the same water depth. Most points are above (or near) the horizontal axis, which indicates that the measured differences were lower than 1 mg/l. Most points with differences above this value correspond to measurements taken at a single location (location 1: Harnaschpolder). This location is occasionally dredged to ensure sufficient water depth under the floating houses and was therefore characterized by high water turbidity and by small space available under the floating houses. The low dissolved oxygen readings may have been influenced by the contact of the sensors with the (anoxic) sediment layer at the bottom of the pond while positioning the drone under the building (sediment disturbances visible on the video recordings from the underwater cameras). The other point indicating a 2.08 mg/l difference corresponds to Drijvende Kas (location 6), a floating greenhouse that is located at a small and shallow pond with little water circulation, which also has limited space under the floating structure. The sediment layer may again have been disturbed by the underwater drone.

The dataset from the longer monitoring period is presented in Figure 6. Both temperature and the concentration of dissolved oxygen are shown. It can be observed that the data are characterized by daily (and seasonal) fluctuations, and that the parameters at both locations show similar behaviour. For both parameters, the paired t-tests indicated high correlation between the pair of variables, and that the differences are statistically significant for both parameters (95% confidence interval between −0.13 and −0.12 for temperature and between 0.17 and 0.18 for dissolved oxygen concentrations).

Figure 7 represents the daily averages of dissolved oxygen concentrations and temperature of the water in combination with local meteorological data. The dissolved oxygen differences between the two locations seem to be less noticeable as the air temperature decreases. Water temperature seems to follow the same trend at both locations, although consistently slightly lower at the open water location. The difference was lower than 0.5 °C during most of the monitoring period, with a few moments with differences of approximately 1 °C. Air temperature, which decreases over time reaching a peak on 16th September, was the guiding factor as it influences both the temperature of the water (decrease over time) and the concentration of dissolved oxygen (increase over time). This is in accordance with the inverse relationship that exists between the two parameters. The sudden increase in precipitation may have contributed to an increase in dissolved oxygen immediately after the rainfall event, but longer datasets are needed for further assessing this correlation. Windspeed during the monitoring period was low for the Netherlands; therefore, the influence of floating structures during windstorms could not be evaluated.

In addition to seasonal/monthly variations, the data are characterized by daily/hourly variations. Figure 8 shows the data averaged by hour for the complete dataset and for a selection of warmer (14–18th September) and colder (1–4th October) days. It can be observed that the differences in temperature were lower during the warmer hours of the day (from 1 pm to 6 pm), while the differences in dissolved oxygen are higher between 7 am and 6 pm. The latter coincides with the sunlight hours; therefore, the differences are possibly linked with reduced photosynthesis: floating structures block direct solar leading to a reduction in the production of dissolved oxygen underneath them. However, it is still unclear if the cause for lower oxygen content underneath the floating houses is caused by a reduction of photosynthesis or is due to other factors such as reduced reaeration due to limited air–water interface. During the warmer period, it was noticeable that temperature at open water was higher during the night than under the floating building. During the colder period, the variations in dissolved oxygen seem to be related mostly with the variations in water temperature.

Figure 9 shows results averaged to an hourly representation. It can be observed that the datasets follow similar diurnal cycles to those previously observed at the Floating Pavilion, with an increasing temperature and dissolved oxygen levels as the day advances. The dissolved oxygen concentrations, although similar during the night, were lower during the day under the floating structures at all locations. The only exception was at the Himpenser Wielen (location 2), with lower dissolved oxygen concentrations under/near the floating houses throughout the complete monitoring period. The peaking temperatures were not consistent at the different locations, with the higher temperatures being observed either at open water or under the floating building. Increases and decreases in temperature appear to be softened under the houses, as less fluctuation occurred, when compared with open water. The floating houses seemed to act as a source of heat, with higher temperatures found in their vicinity. Additionally, the limited air–water interaction caused by the presence of the floating houses may also contribute to store energy in the confined area underneath them, and to reduce losses of energy to the atmosphere.

While, in some study locations, high water turbidity hampered the use of underwater footage for ecological observations, in a few locations high water transparency allowed clear images of the aquatic ecosystem underneath and in the vicinity of floating structures (Figure 10). It was observed that the turbidity of the water was negatively influenced by heavy rainfall/wind during the days that preceded fieldwork activities, and that clear water was usually present in bivalve-dominated ecosystems. In some locations, occasional dredging is required around the floating houses, which causes resuspension of sediments, and consequently high water turbidity and damages ecosystems. Figure 10 shows underwater images at different parts of the ecosystem (e.g. sediment surface and surfaces of floating objects) and from different locations. These images were selected by the authors and provide a representative impression of the observed coverage and the characteristics of habitats including the condition of macrophytes, biofouling/algae, fish population or the benthic ecosystem, as discussed in the following sub-sections.

Many sessile organisms were found attached to the structures (e.g. dreissenid mussels). The floating structures also seem to be attractive for zooplankton and Mysidae (Neomyses). This was observed not only in older structures (e.g. floating holiday houses in Roermond (location 12), built over 15 years ago), but also in more recently built developments (e.g. floating homes in location 13: Ohe en Laak, project completed 2 years before the measurements). Depending on the location, the submerged walls/facades of the floating house were fully covered either by algae/biofouling or by colonies of mussels (Dreissena). The observation of different floating houses in the same water system also indicated that different materials have an influence on the type and amount of vegetation on the surface of these structures.

Under the floating foundations, underwater footage showed different levels of coverage by bivalves, which get attached and grow directly on the surface of the floating structure. In some cases (e.g. Floating Pavilion), the mussels grew on top of each other, resulting in strings of mussels that were hanging down from the platform. Less frequently, scattered groups of mussels were only partly covering the surface (20–30%). In addition to the colonization by bivalves, bright globular sponges were also visible in most locations, frequently growing over the mussels.

At the bottom of the water body, three main areas with different benthic characteristics were identified as follows: (i) areas away from the floating structures, where the benthic ecosystem varied from bare sand to the presence of macrophytes, green filamentous algae or even scattered mussel beds, depending on each location. As the drone approaches the edge of the platforms, the sediment surface becomes increasingly populated by mussels, often with macrophytes (e.g. Ceratophyllum demersum) and algae growing in-between. These (ii) partly shaded areas seem to have different characteristics at different sides of the floating houses, especially in the presence of macrophytes, which are likely due to different degrees of shading. In the sides that have higher solar exposure during the day, the presence of macrophytes frequently reaches the wall of the floating object, whereas in shaded parts this type of aquatic vegetation becomes less dense. The occurrence of macrophyte plants demonstrates that the light intensity that reaches the bottom in these semi-shaded areas is sufficient to sustain photosynthesis. While the transitions between the different benthic habitats are generally continuous, the (iii) benthic ecosystem under the floating houses is generally characterized by a lack of macrophytes, which seem to disappear abruptly, forming a clear line near the edge of the floating object. Under the floating platforms, dead mussel shells could be found, which are likely detached from the bottom of the platform. Also, bare sand is frequently visible in the images. At most locations, the dominant mussel species was Dreissena (as observed in most Dutch freshwater systems). At the study locations that were deeper, these changes and patterns are less visible, which is logical considering the relatively small scale of the floating constructions at these sites, and considering that the water is usually less stagnant due to currents and waves.

Additionally, different species of fish of varying sizes were also visible in the images, including smaller fish, as well as zooplankton, or water fleas. Information about the species that are present underneath and in the vicinity of floating structures also provides information about water quality and ecosystem health (Kroes et al. 2015). However, considering that underwater drones generate noises and disturbances that cause aquatic life to disperse, other methods may be more adequate for this type of study.

At the water surface, (floating) algae and debris/waste accumulation is usually visible on stagnant water, between the floating building and the margin, on most of the visited locations. Additionally, different birds were observed at the study locations, which are backed by the droppings of birds that are frequently visible on the wooden terraces that surround the floating houses.

It was observed in the images that a lively and diverse habitat is created in the vicinity of these structures. The presence of the floating structures provides surfaces that allow the attachment of vegetation and organisms, and/or provide food and shelter for smaller fish. If designed correctly, floating constructions have the potential to stimulate aquatic life and biodiversity. Different studies acknowledge the potential of floating solutions for environmental restoration, enhancing habitats or providing wave protection (Kyozuka et al. 2001; Dafforn et al. 2015), such as integration with floating wetlands or salt marshes. Dafforn et al. (2015) discusses engineering-based solutions to improve water quality by reducing pollutant contamination at boating infrastructure sites (e.g. physical containment). Improvements in water quality could also be achieved by using seaweed to absorb inorganic contaminants or stimulate the growth of bivalves to improve water quality through filtration (Forrest et al. 2009). Lu et al. (2015) conducted water quality measurements near a floating island prototype equipped with a mechanical system for continuous aeration. It was observed that the island generated multiple benefits such as homogenization of water temperature, provision of habitat and food for different organisms and animals, increased dissolved oxygen levels, lower differences in pH values between water layers, as well as contributing to the nitrification and microbial environment. The ecological implications from placing these structures in the water are described by Härtwich (2016), also addressing large-scale scenarios of floating cities. However, floating structures may also serve as habitat for invasive species or create ecological traps, and therefore require careful analysis to target appropriate native species and avoid negative ecological outcomes (Dafforn et al. 2015).

Some aspects remain unclear after this monitoring campaign, and additional studies are necessary to further understand (i) the cause and processes that led to the occurrence of lower dissolved oxygen levels underneath the floating house is still unclear, as it can be attributed to (a combination of) different factors, such as a reduced photosynthesis or reduced aeration. As this study mostly focused on data collected at a limited water depth immediately underneath the floating building, the impact at other parts of the water column remains unstudied. Lenz (2018) indicates that water quality at the bottom layers may be more affected than at the surface; (ii) quantification of heat transfers between the water and the floating houses, as well as the effect of the diminished air–water interaction; (iii) the effect of wind in mixing and aeration processes could not be evaluated due to the lack of storms during the monitoring period; (iv) the volume of water affected by this type of projects is also not yet known, nor how it relates with local hydrodynamics (e.g. flow velocity/currents/tides, stagnation) and water renewal rate/residence time underneath the floating houses. In The Netherlands, frequently floating houses are installed in small ponds with little water circulation, which is a considerably different environment than, for example, projects in coastal waters; (v) how sedimentation rates are affected by the presence of floating buildings, and its impact on ecosystems; (vi) further understand potential changes in biodiversity (possible dominance of certain species that are able to survive in the different water quality conditions); (vii) investigate potential relations between water quality or ecology and the characteristics of the water body, the space available underneath the structures and the materials used (de Lima et al. 2015).

It is recommended to continue monitoring this type of project as their scale increases, to improve the monitoring campaign and to conduct baseline measurements before new structures are installed. To characterize the changes in the water system after the installation of floating platforms, it is advised to conduct longer monitoring campaigns, including measurement in all the different seasons, collect depth profiles of water quality, analyse sediment samples to study changes in the accumulation of bio-deposits and the change in macroinvertebrate species composition, map the distribution, density and diversity of benthic flora and fauna, measure the sedimentation rate, measure the light intensity underneath the structures and in shaded areas, monitor hydrodynamic variables (e.g. flow velocity and direction of currents) and measure other water quality parameters (e.g. nutrients, organic matter redox potential turbidity and phytoplankton). Additionally, considering the complexity of the interactions involved, multi-variate analysis and numerical modelling of water quality, ecology or light penetration are needed for a better understanding of the processes affected by the floating structures and contribute to the optimization of design of this type of constructions (e.g. size, shape and orientation) to minimize undesired impacts.

New projects have the opportunity to implement design choices that take into consideration the health of ecosystems. Guidelines and recommendations are needed to ensure that the design and management/maintenance of floating platforms take into consideration the minimization of undesired effects and the implementation of mitigation and ecological restoration measures. Aspects such as platform shape, size or orientation can influence water quality and can therefore be optimized to reduce negative impacts. For example, the distance to the edges of the platform where sunlight is available is smaller on an elongated platform than from the centre of square platforms. Placing the platforms perpendicularly to water flow/currents creates more disturbances to water flow and increases potential impacts. Different functions can be combined using floating structures, such as water treatment, habitat enhancement, wave protection or green landscaping. Ensuring adequate maintenance of the floating buildings by performing tasks such as harvesting of the mussels and macrophytes that colonize the bottom of the platforms can help reducing the accumulation of bio-deposits and removing nutrients from the aquatic system.

The selection of adequate locations is also key to prevent negative impacts, as the extent of impacts depends on the characteristics and sensitivity of the ecosystem to the specific impacts. For example, an ecosystem with high turbidity and low habitat diversity might benefit from the presence of filter feeders, while ecosystems which already show high diversity and density are more likely to undergo undesired changes due to limited availability of light, high sedimentation rates and the accumulation of organic matter caused by the presence of floating platforms. Therefore, before the floating platforms are installed, the physical and biological properties of the location should be investigated, and the design of the platform should be adjusted according to the characteristics of each specific location.

During the construction phase, measures should be taken to avoid unnecessary contaminations and disturbances (e.g. plan construction works for periods when wildlife is least susceptible to stress; use unharmful substances/materials for the construction of the floating platforms and take measures to prevent spills).

This study discussed potential environmental impacts and benefits that can appear from introducing floating structures on water bodies such as lakes, ponds or canals. Water managers often face concerns of harming the environment with the implementation of this type of projects, resulting in the delay or hindering of new floating developments that could potentially provide cost-efficient and environmentally friendly solutions, when compared with alternatives such as land reclamation of land-based urban expansion.

The (in situ) water quality and ecology monitoring campaign consisted of 18 study locations, where different types of floating buildings of varying sizes are located (ranging from small floating houses to larger pavilions/greenhouses). The different water bodies have different characteristics (e.g. canals, rivers, lakes and ponds), and measurements with underwater drones were taken at different seasons and under different weather conditions. Results show that there are detectable differences in the measured parameters at the measured locations. The magnitudes of these differences seem to be linked with the characteristics and type of water system where it is anchored (e.g. hydrodynamic conditions, bathymetry and water depth under floating structures). Lower dissolved oxygen concentrations were detected underneath the floating buildings, which are likely caused by reduced oxygen production through photosynthesis in the shaded areas (based on the observed hourly variations of this parameter). The concentration of dissolved oxygen stayed within the range of 6–9 mg/l for most locations, which is higher than the threshold of 4.5 mg/l that safeguards healthy aquatic life in ponds and water bodies. The temperature measured underneath the floating houses was higher than in open water in some locations. Although it was expected that temperatures would be lower under structures considering the reduction in direct sunlight, results indicate that the floating buildings contribute to prevent cooling of the water, potentially by reducing losses of heat to the atmosphere or by transferring heat to the water. The long-term continuous measurements illustrated how these differences vary during the day, and how water temperature and dissolved oxygen content strongly depend on weather conditions. Different locations showed similar patterns and (daily) cycles.

Additionally, video recordings captured by the underwater drone provided underwater impressions of the colonization of the platforms by filter feeders/bivalves (on the walls and on the underside), confirmed by the presence of fish in the vicinity of floating structures, showed reduced growth/absence of macrophytes in the shaded areas and under the platforms and revealed the accumulation of (dead) shells under the platforms. The collected images showed evidence that lively ecosystems can emerge after the construction of floating building within relatively short periods of time: in some locations with 1–3 years of old floating projects, mussels and fish could already be found in its vicinity and submerged surfaces.

The use of underwater drones combined with water quality sensors and video cameras proved to be a resourceful tool to collect water quality and ecology data in difficult to access locations, such as underneath floating structures. The underwater images gave a good impression of the lively underwater environments and can also be used for data collection in a wide variety of studies (de Lima et al. 2020). When using underwater drones (ROV) connected with umbilical/tether cables care must be taken to prevent entanglement or other field challenges. When installing permanent in situ sensors for long periods in the field, one of the main concerns regarded the safety of (costly) equipment against risks such as theft or damage caused by passing boats. These aspects influenced the choice of the field study locations. Additionally, the installation of the equipment required the use of boats to reach the open water position and of underwater drones to transport cable across the underside of the floating platforms.

Given the wide variety and types of water bodies considered in this study, it is encouraging to observe that during the measurement periods, the water quality was not critically affected by the presence of the floating houses. This study has shown that while floating platforms can lead both to undesired and beneficial changes in ecosystems, understanding the mechanisms leading to these changes can open opportunities for new smart designs that minimize negative impacts and contribute to improve the health of water systems. The knowledge gained can be used to support the development of policies and regulations that facilitate and guide the construction of floating buildings, selection of suitable locations and requirement of adequate management/maintenance actions. Further research is still needed to increase the understanding of the complex interactions between chemical/biological/physical processes that take place under floating structures.

The authors thank the multiple researchers who contributed to this research, in particular Janko Lenz, Hannah Hartwich and Vladislav Sazonov. The authors appreciate the availability of owners/tenants of the visited properties that allowed the research team to access the platforms and provided additional insights about each location. Finally, the authors wish to acknowledge Blue21, Tauw, Deltares and Valorisatieprogramma Deltatechnologie en Water for providing the resources needed for the success of this research.

All relevant data are included in the paper or its Supplementary Information.