**1. Introduction**

In the environmental sciences, eutrophication is considered problematic under excessive anthropogenic nitrogen (N) and phosphorus (P) enrichment in aquatic ecosystems [1–4]. The eutrophication of coastal marine waters is driven by N as the nutrient that limits biomass production. Instead, P is the major driver of primary production in rivers, lakes, and reservoirs [5,6]. A surplus of N and P triggers plant growth, mainly of algae [7–9]. Such algae blooms can drastically decrease water quality in terms of turbidity, foul odor/taste, and (brown-) green color [4]. Some species (e.g., cyanobacteria) even produce toxic substances that make water consumption harmful to human and animal health [1,2,10]. Moreover, dead algae sink to the ground and are decayed by microorganisms under the consumption of oxygen. Due to the large number of dead algae during algae blooms, decay increasingly depletes dissolved oxygen (DO) in the water until the ecosystem finally becomes hypoxic (low concentrations of DO) or even anoxic (depletion of DO) [1,11]. In the latter case, higher aquatic organisms depending on DO (e.g., fish) die and mass "fish kills" may result [3,6,12]. Such extermination of certain species undermines the structure and functioning of the aquatic ecosystem until it finally collapses [2,4].

Due to global warming, the frequency and geographical extent of harmful algae blooms has already increased in recent decades [1]. Climate change is expected to further intensify eutrophication [5,13]. Increasing air temperatures will lead to increasing water temperatures of surface waters [2,14,15]. Warmth fosters most biochemical processes [5,16]. It will likely trigger algae growth, especially of cyanobacteria at temperatures ≥25 ◦C [13,14]. Due to global warming, algae blooms are expected to start earlier and persist longer in the year [16,17]. Higher water temperatures also enhance the decay of dead organic matter at the bottom of water bodies, thus exacerbating the depletion of DO, and, in turn, P dissolution from sediment and fish kills [2,13]. Eutrophication risk is assumed to increase substantially in the summer, when high temperatures and low precipitation lead to low water levels. As a consequence, the dilution of nutrients is low [3,5,14]. Moreover, long residence times and a slow velocity of water are expected in warmer summers [16,18]. Such conditions foster the build-up of algae blooms. Additionally, climate change is predicted to increase the frequency and magnitude of extreme summer storms [2,13,18]. Such events trigger soil erosion and overland flow, transporting P-containing sediment into water bodies [5,14,19].

Climate change is projected to decrease surface water quality significantly [1,2,16]. This might become a challenge for the global community, because freshwater ecosystems are suppliers of drinking water [2,4]. As eutrophication reduces water quality, the production of drinking water from affected water bodies becomes more cost-intensive and insecure [3,5]. Moreover, due to lacking aesthetics or health risks, affected water bodies often cannot be used for tourism and recreational activities anymore [3,9]. There might even be considerable loss of value of waterfront real estate [4]. Furthermore, harmful algae blooms and fish kills result in lower profits for fisheries [9,12].

To assess the effects of climate change on eutrophication, a case study was conducted in the catchment of the Antrift reservoir (Hesse, Germany). Because 2018 and 2019 were uncommonly warm and dry years in Germany, our water quality assessment gives a preview of eutrophication under climate change conditions. We consider our data empirically illustrative of the assumptions made in modelling-based studies on climate change impacts on water quality. The goals of our study were (1) to empirically depict the temporal development of eutrophication in an uncommonly warm year, (2) to elucidate the effects of sediment P loads on eutrophication, and (3) to seek an indication of diffuse P sources beside erosion and surface runoff. A better understanding of eutrophication drivers and their relations with climate change is a prerequisite for targeted management to secure the availability of water in a warmer world [2,5].

#### **2. Study Area**

In Hesse (Germany), 79% of surface waters were classified as of poor ecological status in 2015, because they overstepped the total P (TP) threshold value of 0.1 mg/l [20]. According to modelling estimations, 1100 t of P are anthropogenically introduced into the Hessian surface waters every year; 65% resulting from communal point sources (wastewater treatment plants), 15% from erosion and surface runoff, 17% from other diffuse sources (e.g., rainwater channels), and 3% from industrial point sources [20]. On the basis of these results, the target values for effluents from wastewater treatment plants were drastically lowered from 1.0 to 0.2 mg TP/l in 2015–2021, which required the technical improvement of many sewage plants [21].

Improved wastewater treatment is estimated to reduce P inputs by only 43% [21]. Further technical improvement is currently considered technically impossible or not economically feasible. The Hessian State Ministry is skeptical that point-source-management would suffice to transfer the Hessian surface waters to a good ecological state [20]. When the proportion of point sources decreases, the relative contribution of diffuse sources to P inputs in water bodies increases. Hence, improved knowledge on diffuse P sources is required to develop targeted and efficient management strategies, to prevent or mitigate future P transfer to water bodies [3,5,14].

We conducted a case study in the catchment of the Antrift reservoir 5 km to the northwest of Alsfeld (Hesse, Germany) (Figure 1) [22]. The climate is warm and temperate [22,23], with an average annual temperature of 8.4 ◦C and with annual mean precipitation of 714 mm [24]. For 2035, regional climate projections depict a significant increase in the annual mean temperature (+0.8 ◦C; high confidence), an increase in heat extremes and droughts, an increase in events with intensive precipitation (>20 mm precipitation), and a slight increase in total annual precipitation [25,26].

**Figure 1.** Location of the Antrift catchment with main water courses and sampling/measurement stations. (Data source: CORINE land use classification by Umweltbundesamt, 2019, DEM 1 m by Hessian Administration for Soil Management and Geoinformation, and overview map according to Weihrauch et al. 2020).

The catchment of the Antrift reservoir has a total area of 61.7 km2 and an altitudinal difference of 152 m from the higher southern parts to the lower northern area. The population ranges between 49–69 persons/km<sup>2</sup> (Hessian average: 269 persons/km2). The catchment is a rural region [27] composed of ca. 7.9% settlement area, 45.6% forest, and 45.2% agricultural areas (53.0% cropland and 47.0% grassland) [27,28]. Its landscape is shaped by geological and pedological features (i.e., suitability for land use; Table 1). Forests are mainly found on wooded hills and ridges, where Cambisols and Leptosols (WRB classification) developed from Tertiary alkali basalts or red Mesozoic sandstones. Agricultural lands are largely situated in moderately sloping terrain on Luvisols, Stagnic Luvisols, and Anthrosols developed from Pleistocene loess. Grassland can largely be found in the floodplain areas and valley floors on Gleysols, Gleyic Fluvisols, and Anthrosols developed from Quaternary alluvial sediments [29–33]. Wooded areas are used for forestry, and grassland for grazing [28]. The farms mainly produce wheat, winter wheat, and barley. Out of 67 agricultural holdings, only five practice organic farming [27].

The Antrift catchment comprises the main river Antrift, with a length of 16.5 km, and the headwaters Göringer Bach (11.0 km) and Ocherbach (10.0 km) [34]. Both headwaters join with the main river before it flows into the Antrift reservoir (Figure 1). North of the reservoir, the Antrift river continues to drain towards the Weser River, which finally flows into the North Sea.

The Antrift reservoir has an average water surface of 0.31 km<sup>2</sup> and a maximum depth of 10.0 m [35]. It has been characterized as carbonatic and polymictic, with an unstratified water column throughout the year (except for May 2015) [35]. While construction began in 1971, the reservoir was first used for water retention in 1981 [34]. Besides flood protection, it was initially planned to be used for recreational activities and swimming. However, this has never been realized due to strong eutrophication, frequent algae blooms, and fish kills since the 1980s [34]. In 2015, the Antrift reservoir was classified as being in a poor ecological state (polytrophic) according to the European Water Framework Directive [35,36]. High TP concentrations of 0.14 mg TP/l were already measured at the reservoir's inlet [35], indicating that the local eutrophication issue was catchment-based instead of solely reservoir-based.


**Table 1.** Environmental characterization of the Antrift catchment.

The poor ecological status of the Antrift reservoir has remained unexplained to the present day. Several official assessments were carried out to develop effective strategies for water quality improvement. In line with the research history, the eutrophication of the Antrift reservoir was first attributed to P inputs from point sources [1,34,37,38]. However, eutrophication recurred after the only wastewater treatment facility of the catchment had been closed in 2002 (personal communication, F. Reissig, Regional Administrative Council Giessen). Since then, erosion and surface runoff have been considered the major sources of P inputs [37,39–41], even though Grimm et al. (2000) concluded in their assessment that only few agricultural lands in the Antrift catchment were severely threatened by erosion [42]. Moreover, erosion has been alleviated further due to financial support and far-reaching advice to farmers, with regard to conservation tillage, intercrop cultivation (especially in winter), conversion of arable lands into permanent grassland, as well as the establishment of vegetated buffer strips along surface waters [20,21]. Since the implementation of erosion conservation measures, the ongoing eutrophication of the Antrift reservoir has been increasingly attributed to P-containing sediment in the catchment water bodies, from which P can be desorbed [37,43,44]. As the impacts of erosion are rather small in this catchment [42], there might not be much sediment in the water courses from which P could be desorbed continuously in an extent large enough to lead to poor ecological state. With regard to legacy P [43], we presume a relatively quick equilibration between

P-containing sediments and the water column took place after sediment deposition. This is suggested by Cornel et al. (2015), who showed in a desorption experiment that 20% of the mobilizable P from solids within the effluents of waste-water treatment plants (with Fe and Al precipitation to eliminate P) were set free within a few minutes, and 50% were mobilized within two days [45]. Hence, as sediment deposition has been largely controlled in the Antrift catchment since the 1990s, we do not assume P release from prior sediments to be still relevant today. Thus, theoretically, it is difficult to attribute the present eutrophication of the Antrift reservoir to the P sources commonly discussed in the scientific literature.

#### **3. Material and Methods**
