4.4.1. General Sediment Features

The sediments from the Antrift catchment were largely silty-loamy (average: 63.9 mass-% silt, 18.1 mass-% sand, 18.0 mass-% clay; Table 2). However, the clay content decreased and the sand content increased from the headwaters (25.7 mass-% clay, 5.5 mass-% sand; averages) to the reservoir (16 mass-% clay, 27.3 mass-% sand; averages). The silt content decreased slightly from the headwaters/main river (mean: 68.7 mass-%) to the reservoir (mean: 56.7 mass-%). There was also temporal variation in the sediments' textures from the first to the second sampling (average deviations: +6.2 mass-% silt; +9.3 mass-% sand, −15.5 mass-% clay).


**Table 2.** General sediment features.

**<sup>a</sup>** Organic matter; **<sup>b</sup>** not determined.

With an overall mean pH of 6.8, the sediments depict slightly acidic to neutral conditions. However, over our study period, the catchment pH shifted between slightly alkaline conditions (first sampling) and slightly acidic conditions (second sampling). The average pH increased from the headwaters (slightly acidic), to the main river, to the reservoir (slightly alkaline).

The sediments in the Antrift catchment were relatively high in organic matter (OM; overall mean: 8.6 mass-%). There was a slight tendency for OM contents to decrease from the headwaters to the main river to the reservoir at the first sampling (17.1 vs. 7.9 vs. 4.7 mass-%, respectively). However, OM contents were relatively similar at the second sampling (8.3 vs. 8.8 vs. 7.4 mass-%, respectively).

#### 4.4.2. Sediment Phosphorus Contents

The sediments in the Antrift catchment are relatively high in each of the determined P fractions (Table 3). We found spatial and temporal variability in the sediment P data. The average PdHCl contents were significantly larger at the second sampling (+30.3%). Besides, we found relatively large ranges of the PdHCl means of the hydrological units. Through our study period, the largest mean PdHCl occurred in the main river stations. However, for the first sampling, the smallest mean PdHCl was in the reservoir. On the second date, the minimum average PdHCl was instead observed in the headwaters.

The Pml contents changed insignificantly from the first to the second sampling (+26%) and among the hydrological units. However, there was an increase in Pml in the headwaters and the reservoir at the second sampling. The mean Prc contents also increased in the reservoir at the second sampling, but decreased in the headwaters. The largest mean Prc contents were generally found in the main river.

We found relatively large DPM2 and DPM3 (i.e., >0.4). From the first to the second sampling, DPM2 remained constant in the main river. It was extremely small in the reservoir at the first sampling, but almost doubled to the second sampling. By contrast, DPM2 decreased by about 33% in the headwaters from Sampling 1 to Sampling 2. DPM3 increased slightly to the second sampling (+19%). However, we found no significant changes between the sampling dates and hydrological units.


**Table 3.** Differential P fractions and degrees of P mobilization of sediments (means).

<sup>a</sup> for explanation, see 3.5; \* different letters indicate significant differences between means (*p* ≤ 0.05); lower-case letters refer to comparison by line (sampling dates per P fraction), upper-case letters refer to comparison by row (hydrological units per P fraction and sampling date).

#### 4.4.3. Correlation Analyses

We conducted Spearman correlation analyses for all our data, as well as separately for the data of each sampling date. Different correlation coefficients result in these three versions of data aggregation. The specific results for the correlation of all data are shown in Figure 9. However, we also found overarching correlations (i.e., present in all three versions of data aggregation).

**Figure 9.** Spearman correlation of sediment and water data for both samplings. *p* ≤ 0.05 (green), *p* ≤ 0.01 (blue) and *p* ≤ 0.001 (brown), not significant (no color).

First, there are regular correlations between the moderately labile to recalcitrant P fractions and the acid metal cations, as well as the base cations. PdHCl, Pox, and PAR are correlated positively and clearly to strongly with Alox, AlAR, FeAR, NaAR, MgAR, CaAR. For all data, there are furthermore positive correlations of PdHCl, Pox and PAR with Feox (weak) and KAR (weak to clear; Figure 9).

Second, we mostly found weak positive correlations of PdHCl, Pox, and PAR with the sediments' OM content.

#### **5. Discussion**

#### *5.1. Evaluation of Climate Data: Discharge and Precipitation*

The trends of air temperature and monthly precipitation showed uncommonly warm and dry conditions in the Antrift catchment in 2018 and 2019 (Figure 2). These years could be termed 'extreme', regardless of the general trend for precipitation to decrease on average by 68 mm per decade in Hesse [22,72]. Despite the long-term drought, increased heat and low flow conditions, individual higher to heavy precipitation events occurred during the investigation period, whose rainfall intensity might have enabled soil erosion [68,74]. However, a high potential for causing soil erosion can only be

attributed to the storm event on 20 May 2019, which exceeded a daily precipitation sum of 70 mm, and could hence be termed an extreme rainfall event [68,74,75]. Such extreme weather conditions are assumed to increase in frequency under climate change [16].

Climate forecasts (e.g., IPCC, HLNUG) predict an increase of annual mean temperature, heat extremes, drought periods, and heavy precipitation events [25,26,76]. We documented all these factors during our investigation period. Therefore, our data are likely to draw a picture of water quality impairment under climate change.

#### *5.2. Water Quality Assessment*

#### 5.2.1. Water Temperature

In general, water temperature is correlated with air temperature. An increase in air temperature through climate change leads to increasing water temperatures [16]. In addition, water temperature is a main factor controlling chemical and physical properties of aquatic ecosystems (e.g., pH, diffusion rates) [2,14]. The water temperature within the Antrift reservoir and its tributaries ranged between the lower and upper thresholds for good ecological status during our investigation period. Although a moderate increase of air temperature from winter to summer was detected (see Supplementary Materials), the influence of climate change and a long-term increase of water temperature could not be investigated, because this study considers a 10-month period only. However, a long-term increase in water temperature is documented for rivers and lakes by other studies and can be assumed for the Antrift catchment under aggravated climate change conditions [5,77]. Deviations from the moderate annual increase of water temperatures from spring to summer occurred only due to heavy rainfall events (e.g., a thunderstorm in May 2019). Moreover, minor differences were observed due to spatial variation between lower water temperatures at sampling sites shaded by riparian vegetation (e.g., headwaters, the main river) and higher temperatures in the unshaded Antrift reservoir.

Water temperature and sufficient nutrient concentrations stimulate algal growth [1]. To do so, temperatures were too low (<20 ◦C to 23.0 ◦C) at the beginning of our measurement period [77,78]. Climate change is projected to increase the persistence of high water temperatures—especially in shallow lakes—and thus lead to extended eutrophication periods and algal blooms [2,16,17]. However, Richardson et al. (2019) observed factors limiting total phytoplankton growth and fostering the decrease of harmful cyanobacteria in nutrient-rich environments. Thus, climate change effects on algae blooms might be more complex than commonly assumed [79].

In our study, temperatures have been favorable for algal growth since June 2019. On 5 June 2019, an incipient bloom of blue-green algae was observed, leading to a widespread green-brown algal carpet on the water surface of the Antrift reservoir in August 2019 (see Supplementary Materials). At the Antrift main river and the headwaters, no such visible water quality impairment was observed, probably due to more riparian shading and higher flow velocity [80,81].

#### 5.2.2. pH

On 5 November 2018, pH in the Antrift catchment was relatively low (Figure 5b), probably as a result of the bacterial degradation of the 2018 algae bloom [1,15]. During the winter period, pH oscillated around the lower threshold (7.5 pH) for good ecological quality. Beginning in April 2019, pH rose to a more alkaline level [2]. The high fluctuation (i.e., range) of pH is typical for increasing eutrophication appearances [1,82,83]. At that time, we observed increased biomass formation in the reservoir. A second decrease of pH in connection with biomass increase was not observed, because our investigation period ended in August 2019.

#### 5.2.3. Dissolved Oxygen

Despite the three-week interval between measurements, two trends in DO were observed. First, both algae blooms (end phase in November 2018, build-up in June 2019) were correlated with low DO

concentrations. The minimum DO concentrations were measured during the end of the 2018 algae bloom, probably resulting from oxygen depletion through the intensified bacterial degradation of biomass [1,84,85]. Hence, after the winter increase, DO concentrations decreased again after August 2019, when the 2019 algae bloom had already built up for at least two months. Second, increases in DO coincided with increases in precipitation and discharge after the long drought phase in the second half of 2018. Increasing flow velocity leads to a higher level of DO in the surface waters [13,86]. Hence, during winter and spring 2019, DO stayed above the threshold for good ecological status.

The spatial differentiation of DO concentrations reflects the water temperature differences between the reservoir, the main river, and the headwaters [87]. Higher DO concentrations occurred at the headwater stations on single dates, probably as a result of higher flow velocity and shading by riparian vegetation (see Supplementary Materials). Still, during the 2019 algae bloom, the lowest DO concentrations were measured in the headwaters, most likely because of low flow rates and low velocity [88,89]. This indicates a higher vulnerability of the headwaters for water quality impairment under climate change (e.g., due to prolonged low flow conditions).

#### 5.2.4. Total Phosphorus

Our data confirm that the Antrift reservoir is eutrophicated according to the EU Water Framework Directive. High TP concentrations occurred throughout our investigation period (Figure 5). However, TP concentrations indicating poor ecological quality were measured only from November to December 2018, and sporadically (event-driven) in February and May 2019. On the one hand, the high TP concentrations between November and December 2018 can be explained by low flow rates and a low nutrient dilution in the water column [16]. On the other hand, this period marks the end phase of the 2018 algae bloom with low DO concentrations and pH.

The average TP concentrations decreased continuously until January 2019. This can be attributed to the increase in precipitation and flow rates within the Antrift catchment. Even with the relaxation of the eutrophication situation during winter 2018, the median TP concentrations stayed close to the lower threshold of eutrophic conditions (except for single outliers; Figure 5d). After the autumnal decrease, TP concentrations varied slightly for the rest of the studied period.

Since July 2019, there was a strong negative correlation of TP concentrations with water temperature (rSP = −0.66; *p* = 0.01). On 5 June 2019, the water temperature reached the 20 ◦C-threshold favorable for algae growth. As a consequence, the TP concentrations decreased, suggesting a consumption of P. This decrease of TP is corroborated by our field observations of increasing algal growth, which was also documented in other case studies [67,90].

With regard to our sampling stations, some conclusions can be drawn regarding potential P sources. We found no increase in TP concentrations with increasing percentage of agricultural land use or settlements [65,91,92]. Interestingly, TP concentrations are somewhat lower in the headwater stations than in the main river and reservoir, but they are still within the range of eutrophic nutritional status.

The reservoir and both stream classes are clearly different in hydrology and morphology. However, no significant difference in stream morphology is assumed between the headwaters and the main river [67]. Generally, our TP data mostly depict no clear differences between the three types of surface water bodies. Thus, not only is the Antrift reservoir eutrophicated, but all the investigated water bodies in the catchment, even the headwaters, are. This indicates that agriculture (erosion, surface runoff) and settlements (point sources) should not be termed the major factors for causing eutrophication in the Antrift catchment. Because the headwater station GOE is situated in a forest region, high TP concentrations also occur in areas without erosion risk and point sources (see Supplementary Materials). Hence, there seem to be further, so far unnoticed, sources for P losses from soil to water.

The TP trend, in combination with the trend of water temperature, pH, DO, and field observations, indicates that, under climate change, the relaxation of water quality impairment during winter could be shortened. Our data depict an extended duration of the 2018 algae bloom and a new worsening of water quality in April 2019. P stocks conserved from the prior algae bloom could accelerate the next

algal bloom in spring. This is corroborated by Richardson et al. (2019), who showed that an increase in water temperature, sufficient nutrient levels, and hydrological conditions (e.g., intensive precipitation after drought events) might trigger cyanobacteria growth and complex algal bloom dynamics [79].

#### *5.3. Phosphorus Sources in the Antrift Catchment*

After the only wastewater treatment plant closed in 2002, no major point sources for P inputs remained in the Antrift catchment [93]. Erosion and surface runoff are thus expected to be the major P sources due to the large percentage of agricultural lands (>45%) [27]. However, the local erosion risk has been judged low in an official assessment [18,68]. Besides, vegetated buffer strips have been established along the water courses in the catchment, even along smaller drainage channels. Moreover, the influence of direct surface runoff and drainage must be considered minor, as a result of the very limited proportion of sealed surfaces.

We tested the influence of possible P inputs by erosion and surface runoff, with a correlation analysis between TP concentrations and their occurrence (i.e., number of days) after the last precipitation event (see Section 3.5. Statistical Analyses). The cross-correlation between the time series data of precipitation and discharge shows that precipitation events influenced the discharge trend significantly (see Supplementary Materials). This correlation is significant for a range between 0 and +2 days after rainfall events. Precipitation events thus had a significant impact on runoff in water bodies for a maximum of two days after the respective event.

The Spearman correlation analyses resulted in a significant negative correlation between average TP concentrations and the number of days since different kinds of precipitation events (Figure 8). Thus, a larger number of days since the last precipitation event coincided with lower TP concentrations in the water bodies. This indicates an influence of precipitation events and intensity on the TP concentrations (e.g., erosion and surface runoff) [71,93]. However, both processes are short-term processes, with a potential direct impact on water quality within a few minutes to hours, maximally in <2 days [68].

During our investigation period, no precipitation event occurred so shortly before a measurement date (i.e., <2 days) that it could be attributed plausibly to erosion or surface runoff [68]. In particular, all the potentially highly erosive higher and heavy precipitation events happened more than two days before our measurements [68–70]. Therefore, neither point sources nor extensive erosion (with surface runoff) can be considered as the major reasons for the all-season high P concentrations in the Antrift catchment. Instead, our results might tentatively point to an effect of underground P translocation (interflow-P), with a delay after precipitation events due to water infiltration and percolation [94,95]. However, our approach is limited by the relatively coarse temporal resolution of our measurements (three weeks) [68,69].

#### *5.4. Sediment Analyses*

#### 5.4.1. General Sediment Features

The texture of the sediments results from the typical bedrocks and substrata in the Antrift catchment. The dominance of silt probably results from the large spatial extension of loess [29,96]. The somewhat higher clay content in the headwater area could be explained by the abundance of basalt in this section of the catchment. By contrast, the increasing percentage of sand in the reservoir sediments could be due to the occurrence of sandstones in the reservoir's surroundings. The variation we found between sediment textures of Samplings 1 and 2 generally demonstrates that sediments are dynamic components of the soil-water-interface, which undergo constant changes by transport, selection, and biochemical reactions [97,98]. Those processes could affect the textural composition of the sediments.

Over our investigation period, the sediments oscillated between slightly acidic and slightly alkaline conditions (Table 3). This general pH range can also be attributed to the dominance of alkali basalt and loess in the Antrift catchment [96]. That pH was lower in the headwater area than in the reservoir is probably geochemically determined. The headwaters are surrounded by basalts, which contain relatively many base cations (Ca, Mg, K), but even more Al/Fe (among other elements). Instead, base cations are probably more abundant in the loess areas (main river, reservoir) [96]. Moreover, pH could increase towards the reservoir due to the increasing concentration of leached basic cations, with increasing flow distance through the catchment.

We found the highest variance of pH between our sampling dates in the main river stations. This seems plausible because the main river represents the largest part of the catchment area and hence the most hydrologically diverse conditions [99]. Instead, the headwaters and the reservoir depict relatively small sections of the catchment with rather controlled hydrological conditions [97,99].

That we found large OM contents is not surprising for fluvial and lacustrine sediments, especially with eutrophication-enhanced primary productivity and the increased accumulation of dead OM [1,5]. Our OM contents decrease from the headwaters to the reservoir, which might be due to increased OM mineralization during summerly eutrophication and algae blooms [3]. Moreover, this spatial differentiation of sediment OM might be related to the spatial distribution of land uses in the Antrift catchment. The headwater areas are largely composed of forests, which produce and transfer more OM to the adjacent water courses. The main river and reservoir stations are instead located in agricultural (conventional farming) and grassland areas, where less OM is produced [96].

#### 5.4.2. Sediment Phosphorus Contents

The large sediment P contents might in part result from the bedrocks in the Antrift catchment (e.g., basalt). Besides, they could be the legacy of prior land management (especially in the agricultural areas around the main river stations). However, the build-up of sediment is generally relatively small and also well-controlled in the Antrift catchment since the 1990s [42]. Instead, sediment pH might be a relevant and overarching driver of sediment P in this catchment. The mean pH levels of the sediments were between 6.0 and 6.5. P mobility is large in this pH range [100,101] because the solubility of P-sorbing metal oxides is relatively low (lability increases with pH), while at the same time, P-bearing minerals precipitating with base cations are also relatively easily soluble (stability increases with pH [96,102]). Hence, any shift of pH would foster P mobilization: Under more acidic conditions, the P-bearing minerals of base cations would be dissolved [103,104]; under more alkaline conditions, P would increasingly be desorbed from metal oxides [105,106]. It is a particularity of the Antrift catchment that sediment P retention is governed by sorption to metal oxides and by precipitation with base cations, instead of either of both. Thus, sediment P contents could be large due to the large number of potential reaction partners. However, a decrease in pH might be most favorable for P retention in the Antrift sediments, because metal cations are much more abundant in the sediments than the base cations.

We found larger sediment PdHCl contents at the second sampling. This might result from an increased tendency to Pml mobilization (i.e., conversion to PdHCl) under high flow conditions (high dilution). Such conditions would cause a disequilibrium between bound P in the sediment and dissolved P in the water column [102,107]. As equilibrium would be shifted towards bound P, P mobilization would be enhanced to restore the equilibrium state [108,109].

The differences between the average PdHCl contents at both samplings suggest that the three hydrological units might be subject to different P dynamics, especially under an acute algae bloom. In the reservoir, PdHCl was very low at the first sampling, possibly due to enhanced P mobilization (i.e., loss of PdHCl) during the 2018 algae bloom and severe DO depletion (Figure 5c). Without an acute algae bloom, PdHCl increased significantly at the second sampling. By contrast, PdHCl decreased in the headwaters at the second sampling. This could be the consequence of increased P mobilization under high flow conditions.

The Pml contents increased slightly but non-significantly on the second date. This might suggest a tendency to P mobilization under high flow conditions, which could successively convert recalcitrant P forms into Pml [6]. Even though not statistically significant, Pml increased in the headwaters and in the reservoir. Most likely, in the reservoir, the small Pml content at Sampling 1 resulted from the acute algae bloom and enhanced P mobilization from the sediment. In the headwaters, the increase in Pml might instead have resulted from successive equilibrium-driven P mobilization under high flow conditions [6,107]. This might also explain the decrease in Prc in the headwaters at the second sampling. However, the increase in Prc in the reservoir might also have resulted from a relatively small Prc at Sampling 1 due to the acute algae bloom with severe DO depletion.

The largest Prc contents were generally found in the main river. With regard to the local bedrocks/substrata (loess), this is possibly due to the high abundance of base cations in the water, which can precipitate with dissolved P, especially under low flow conditions (Sampling 1) [96,109]. Moreover, the main river stations are not affected by algae-bloom–enhanced DO depletion and resulting P mobilization from sediments. Hence, more Prc might accumulate

The relatively large DPM2 and DPM3 suggest that the sediments in the Antrift catchment are generally prone for P mobilization due to their large share of readily and moderately soluble P forms. We found slightly larger DPM2 (+19%) and DPM3 (+19%) at the second sampling, which might point to increased P mobilization under high flow conditions. By contrast, the small DPM2 in the reservoir at Sampling 1 might have resulted from algae-bloom-mediated P mobilization.

To the second sampling, DPM2 decreased significantly in the headwaters. Because the lowest pH were observed for the headwaters, there might be more reactions between P and Al/Fe, which are favored under acidic conditions [103,104]. P associated with Al/Fe would raise sediment Pml (based on Pox) [105,106] and thus result in a smaller DPM2. Furthermore, Pml might have increased due to the above-mentioned successive mobilization of Prc.

For DPM3, we found no significant changes between the sampling dates and hydrological units. However, the smallest DPM3 was found in the reservoir on the first date. This might also have resulted from the above-stated tendency for sediment P mobilization under severe DO depletion.

#### *5.5. Spearman Correlation Analyses of Sediments and Water Samples*

Our correlation analyses resulted in different relationships between the sediment and water parameters when all our data were combined (Figure 9) or grouped according to sampling dates (Figures A1 and A2). Hence, temporal dynamics of rather short-term processes like acute eutrophication with an algae bloom are probably depicted more adequately as snapshots according to sampling dates instead of sums or averages of longer timespans (e.g., yearly averages [3]).

#### 5.5.1. Correlations with Acid and Base Metal Cations

We found overarching positive correlations of PdHCl, Pox, and PAR with Alox, AlAR, FeAR, NaAR, MgAR, and CaAR. This indicates that P bonding in the sediments is strongly governed by the acid metal cations (especially Al), probably largely via sorption, which the strong correlations with Alox point to [102,103]. The correlations of PdHCl, Pox, and PAR with AlAR and FeAR might furthermore indicate that the precipitation of Al/Fe and P-containing minerals could play a role. However, the correlations of PdHCl, Pox, and PAR with NaAR, MgAR, and CaAR indicate that mineral precipitation, either with instantaneous (e.g., Ca-P-minerals) or with subsequent P bonding (e.g., adsorption and surface precipitation [110,111]), might also be relevant for sediment P retention.

#### 5.5.2. Correlations with Sediment Organic Matter

We found weak positive correlations of PdHCl, Pox, and PAR with the sediments' OM content. Because OM contains P, an increasing amount of OM also means an increase of the bound P fractions. Furthermore, the mineralization of organic matter leads to P mobilization to the water column and might result in an increase of TP. This is indicated by the weak negative correlation between sediment OM and TP in the water (Figures 9, A1 and A2). However, this correlation is visible only at the second sampling. On the first date, other eutrophication-related processes might instead have controlled TP.

Generally, the relatively large sediment OM contents (average: 8.61 mass-%) might hamper precipitation reactions and favor the formation of more weakly crystalline mineral phases [104,112]. This might explain why we found correlations of the P fractions with sand and silt but not with clay.

Generally, P bonding increases with clay content because most of the major P sorbing particles belong to the clay fraction (e.g., Fe/Al oxyhydroxides, OM [100,113]). However, Weihrauch (2018) found correlations of soil P fractions with sand and silt instead of clay in a study area characterized by weakly acidic to weakly alkaline soils [59]. The author explained this finding with the precipitation of carbonates adsorbing P weakly in a monolayer (at low P concentration) and with the precipitation of rather amorphous Ca-P-minerals (at higher P concentration) of larger diameter, due to the interference with OM [112]. Moreover, P might become occluded in a recalcitrant form within Mn and Fe concretions, which could grow with time [96,114,115]. This might explain the correlations between PAR and sand (Figures 9, A1 and A2). The correlations between sand and PdHCl might instead result from P sorption and surface precipitation on calcite, the formation of easily soluble Ca-P-minerals, and/or amorphous metal oxides [102,116]. Such a P retention in larger particles might also explain the clear negative correlation between the sediments' silt content and TP in the water (e.g., apatite often forms crystals in the silt fraction [96]). Hence, sediment OM is indicated to affect sediment P contents directly as well as indirectly by influencing the potential P bonding sites.

The above-mentioned correlations show for only the second sampling, probably due to strong effects of eutrophication on P dynamics on the first date. However, we found a strong negative correlation of sediment OM and pH in the water on the first date, which could have resulted from decreasing pH with increasing OM abundance and related mineralization [88,117].

#### **6. Conclusions**

We investigated the eutrophication of the Antrift reservoir in Hesse (Germany) in the uncommonly warm and dry years 2018/2019. Our results give an empirical preview on the development of local water quality under climate change. They furthermore enabled us to answer crucial questions regarding the poor ecological status of the Antrift reservoir.

Our results clarify that not only the Antrift reservoir is affected by eutrophication due to high TP concentrations, but so is the entire catchment. The catchment TP concentrations are high throughout the year, but apparently not due to the P sources commonly stated in eutrophication literature (e.g., agriculture). Possibly, natural pH-driven P mobilization from catchment soils and sediments (depending on soil moisture and flow conditions) might foster the eutrophication of the local surface waters. This hypothesis of an "autogenous eutrophication" should be investigated further.

Conceptually, we differentiate between two kinds of constraints on TP concentrations (graphical abstract): biological and geochemical constraints. Biological constraints are largely determined by water temperature, pH, and nutrient availability (especially P). In the Antrift catchment, pH and TP concentrations are favorable for algal growth throughout the year. Hence, temperature might be the major driver of algal growth, DO depletion, and resulting sediment P mobilization in summer. During the winter, water temperatures <20 ◦C prevent algal growth and DO depletion. Thus, only in summer can biological constraints lead to algae blooms with acute DO depletion. Under such conditions, they outweigh the geochemical constraints.

The geochemical constraints are determined by flow conditions (elemental dilution), water and sediment pH, redox status, as well as, to a lesser degree, by the local bedrocks/substrata (supply of P, Al, Fe, Mn, Ca, Mg, K). At low flow conditions and low dilution (summer), P might increasingly be retained in sediments. By contrast, P is likely to be increasingly mobilized from sediments at high flow conditions and high dilution (winter). This should be studied further.

In the literature, P dissolution/remobilization from sediment mostly describes P desorption/dissolution from sediment that has been transferred to water courses due to land use (especially agriculture) [43,118]. For our argument, it is not relevant where the sediments come from. Sediment P equilibrates relatively rapidly with P in the water column [6]. Hence, any P desorption/dissolution from P-enriched sediment from agricultural lands would likely happen quickly after deposition [45]. Then, the sediments would be subject to the geochemical constraints in the Antrift catchment, which generally foster a high abundance of labile sediment P and temporary P mobilization. This would apply to any sediment, regardless of its original P load and origin regarding land use. Hence, we are skeptical about the scientific theories of a "hysteresis" of P dissolution from sediment, which is sometimes evoked to explain why best-management to improve water quality practices do not succeed right away, but might need decades [43,119].

With regard to climate change, our data from the recent uncommonly warm years corroborate the assumptions of others (e.g., LeMoal et al. 2019; Moss et al. 2011; Whitehead et al. 2009) that global warming will likely foster eutrophication, as well as prolonged and spatially extended algae blooms. Climate change might hence aggravate biological constraints in summer. In winter, however, lowered flow conditions might instead lead to reduced P mobilization, or, in other words, to increased P retention in the catchment sediments. However, combined effects between cyanobacteria and phytoplankton, short-term hydrological changes, and nutrient availability in freshwater ecosystems and sediments might complicate a clear prediction of the interaction of constraints on P mobilization/retention under climate change [79,120,121].

To mitigate the eutrophication of the Antrift catchment, it is important to register that TP concentrations are naturally high throughout the year. Hence, any artificial reduction of TP concentrations (e.g., by common best-management practices) would foster geochemical P mobilization from the sediment. As mentioned above, the Antrift catchment could theoretically self-regulate its ecological status under natural conditions (e.g., by flushing winterly nutrient excesses downstream). The Antrift reservoir undermines this self-regulation and creates an environment of all-year P accumulation in the water column, which artificially fosters algae blooms in summer.

We conclude that the eutrophication (i.e., TP concentrations of 0.03–<0.1 mg TP/l) of the Antrift catchment cannot be prevented. Still, best-management practices should be kept active (e.g., erosion conservation) to prevent further P inputs that could aggravate eutrophication and lead to a poor ecological state (TP concentrations ≥0.1 mg/l). To restrict the summertime P mobilization from the sediment, DO concentrations should be started to be managed besides TP concentrations. Hence, algae blooms (and resulting DO depletion) might be controlled with algicides or the introduction of certain key species into the water bodies [1], bearing in mind the possible ecological consequences. Furthermore, artificial aeration of the Antrift reservoir could significantly relax summerly hypoxia and anoxia. Without hypoxia, the geochemical constraints would probably be dominant in summer and trigger P retention in the sediments. A certain monitoring and the eventual removal of the catchment sediments should also be considered, especially regarding sediments loaded with P in summer. In addition, the respective material would have beneficial features for application as fertilizer (e.g., high in carbonate and OM, pH 6–7, much labile P), if it was free of toxins, etc.

Our study indicates that—beside the known sources of P inputs—there might be currently unknown and unregulated diffuse P sources which contribute to the high P concentrations of the surface waters in the Antrift catchment. These sources seem to depend on the flow conditions, but are activated >2 days after precipitation events. Hence, we hypothesize an underground P translocation with the soil water (interflow-P) [94]. Such P sources should be investigated further so that they could be effectively managed and restricted. Hence, our study demonstrates that freshwater eutrophication is not yet conclusively understood despite the successes in research and practice. Challenges remain for science and should possibly be tackled apart from the disciplinary and conceptual mainstream.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2571-8789/4/2/29/s1, Figure S1: Average monthly air temperature for the climate station Alsfeld-Eifa, Figure S2: Pictures of algal bloom in the Antrift reservoir, Figure S3: Stations of water and sediment sampling in the Antrift catchment, Figure S4: Cross-correlation matrix of precipitation and discharge data, Table S5: Days since precipitation/discharge events according to measurement periods.

**Author Contributions:** Conceptualization, C.J.W., C.W.; Methodology, C.J.W., C.W.; Software, C.J.W., C.W.; Validation, C.J.W., C.W.; Formal Analysis, C.J.W., C.W.; Investigation, C.J.W., C.W.; Resources, C.J.W., C.W.; Data Curation, C.J.W., C.W.; Writing—Original Draft Preparation, C.J.W., C.W.; Writing—Review & Editing, C.J.W., C.W.; Visualization, C.J.W., C.W.; Supervision, C.W.; Project Administration, C.W.; Funding Acquisition, C.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Regional Administrative Council Gießen (Regierungspräsidium Gießen).

**Acknowledgments:** The authors gratefully acknowledge supply with discharge data by the Wasserverband Schwalm e.V. We furthermore thank Alexander Santowski for support in statistical data evaluation and Christin Wedra for support in laboratory analyses. Finally, we thank our students who partook in the research-based seminar and laboratory courses on the eutrophication of the Antrift reservoir.

**Conflicts of Interest:** The authors declare no conflict of interest.
