3.1. N and P Dynamic in the Lake Paranoá Catchment
Average flows and nutrients concentrations of the four investigated tributaries, the two wastewater treatment plants (WWTP), the downstream dam and average concentrations of the nutrients in the lake for the wet and dry seasons are shown in Table 1
The nonparametric Mann-Whitney test showed that there were no significant differences (p < 0.05) between dry and wet season concentrations for the majority of the investigated nutrients, considering all tributaries. The exceptions were:
Significantly higher mean concentration (p < 0.05) of PO43−-P in the Riacho Fundo stream in the dry season,
Significantly higher mean concentration (p < 0.05) of TOP in the Gama stream in the wet season.
These results suggest different sources for these two chemical forms of P in the streams in the southern part of the lake (Figure 1
The higher concentration of PO43−-P in the Riacho Fundo Stream during the dry period demonstrates the contribution of the wastewater treatment plant that discharges its effluent in this tributary.
The downstream dam water samples showed significantly lower concentration (p < 0.05) only for ammonium in the wet season. This can be attributed to dilution by the rainwater and the lower concentration of the discharge (p < 0.05) of the wastewater treatment plant (WWTP) located on the shore of the lake in its northern portion.
Differences in the ammonium and total organic phosphorus (TOP) dynamics were found between the two WWTP located on the shore of the lake. As mentioned above, only the one located at the northern portion shows a decrease in ammonium concentration during the wet season. On the other hand, the WWTP located at the southern portion of the lake was the only one to present a significant increase of TOP concentration (p < 0.05) in its discharges into the lake. These differences can be attributed to their different operating conditions. The WWTP located at the northern portion was operating at half of its maximum capacity; the one located at the southern portion, according to data from the Sanitation Company (CAESB) itself, was close to its operating limit resulting in an inefficient treatment for phosphorus removal.
The lake waters showed significant differences (p
< 0.05) between seasons only for ammonium and nitrate. These two nutrients showed an inverse behavior, registering a decrease in the nitrate concentration and an increase in the ammonium concentration in the dry season. This can be attributed, at least in part, to the biogeochemical process of nitrate reduction or denitrification in the dry period. The data related to the lake are presented (Table 1
) in terms of general average, including the nine points of sampling and the two depths, for each season used for the calculation of the nutrient stock in the water column.
3.2. N and P Load and Storage
shows the results of N and P loads imported to the LP from external (Torto, Bananal, Riacho Fundo, Gama, WWTP North and WWTP South) and internal sources during the wet and dry seasons as well as the loads discharged from the LP (downstream dam).
The Torto and Gama streams presented the greatest variations between the dry and the wet season loads. The load of the Torto stream during the wet season was 13 times higher than that in the dry season for DIN and 12 times for TON. Increase of this magnitude was not observed for P in this tributary. Gama also presented wide input variation between wet and dry seasons. The load of this stream in the wet season was: 8, 6, 3.5 and 10 times higher than in dry season for DIN, TON, PO43− and TOP, respectively.
The Riacho Fundo e Gama tributaries, which discharge into the southern portion of the lake, presented the highest contributions of both N and P in the two periods.
The tributaries located in the northern portion contributed considerably less in the two periods, drought and rainfall. Among all, the Riacho Fundo stream was by far the one that contributed the most N and P to the lake.
The annual DIN load imported to the lake (Table 2
) shows the dominance of WWTP on input in the lake. The two wastewater treatment plants, the WWTP North and WWTP South, provide the major annual contribution of DIN (275 Mg, 66.8%).
The Southern tributaries (Riacho Fundo and Gama) provided the second major annual contribution of DIN (132 Mg, 32%). The Northern tributaries (Torto and Bananal) and internal load provided 1.2% (5.0 Mg) and 0.01% (0.03 Mg), respectively.
With regard to the data from the tributaries, these results can be related to the use and occupation of the soil in the hydrographic basin since the southern tributaries run through more urbanized areas, with a greater level of soil sealing and poorer sanitation system coverage. The northern tributaries, in turn, are less affected by urban expansion since they are located in an environmental preservation area and have greater sanitation system coverage.
A load of 62 Mg of DIN was discharged through the dam, which is equivalent to 15% of the input load.
Considering only the contributions of the tributaries, as there is no estimate of organic nitrogen from the wastewater treatment plants, the streams of the southern region provided the major contribution of TON (146 Mg, 70%). Therefore, 30% of TON (62 Mg) was supplied by tributaries of the northern region (Torto and Bananal).
A total load of 347 Mg of TON was exported through the dam, which is equivalent to 67% of the input load.
In relation to P, the WWTPs are responsible for 99.8% (258 Mg) of the PO43− input with WWTP South alone adding 84.5% (218 Mg) to the lake’s total annual external load. The tributaries (Torto, Bananal, Riacho Fundo and Gama) can contribute up to 0.16% (0.41 Mg) to the total annual load imported to the lake. The internal load was much less significant, reaching only 0.04% (0.12 Mg).
A load of 0.48 Mg of PO43−-P was discharged through the dam, which is equivalent to 0.20% of the input load.
The WWTPs were responsible for 84.8% (14 Mg) of the TOP input with WWTP South alone contributing 61.8% (11 Mg) to the lake’s total annual, external load. The tributaries imported 15.2% (2.6 Mg) to the total annual.
In general, these results suggesting that external loads can be major sources of N and P, contributing to eutrophication in LP. On an annual basis, the N input to Lake Paranoá is mainly controlled by the southern tributaries and the WWTPs, while the WWTPs provide by far the majority of the external loads of P. Therefore, the internal load provides a secondary role in the eutrophication of LP.
The N and P stock in each compartment (water column and sediment) of Lake Paranoá is shown in Figure 4
. Surface sediment is the major storage compartment for both N and P. As expected, phosphorus is usually transported in the particulate form attached to sediment [38
], whereas nitrogen, can be largely fixed in eutrophic lakes [39
] and transported from the water column to the sediment.
Total stocks of N and P in the water column of LP were approximately three and two times higher respectively than those reported by Ramírez-Zierold et al. [9
] for the Valle de Bravo (VB) reservoir, a deep urban lake in Mexico. The total stocks of N and P in the VB were 375 Mg and 27.4 Mg respectively. N and P stocks in the sediment were not determined for the Valle de Bravo. LP and VB had comparable results, probably due to the high hydraulic residence time (>2 years) and high inputs recorded in both reservoirs.
Water column nutrient stocks in shallow lakes located in the USA, Japan, China and New Zealand are in the range of 185–7115 Mg for TN and the 3–464 Mg for TP [4
]. The data presented here (Figure 4
) fit within these values suggesting that average depth and climate are not determinant factors for nutrient storage in lacustrine environments.
Comparable published data, especially for N storage calculations in sediment, were difficult to find. To date, most of the N and P stock calculation studies have been conducted in shallow lakes in temperate climate zones [4
], the data from which cannot be easily applied to tropical deep lakes. Besides this, many studies have focused on a single lake compartment (e.g., water column) or a single nutrient (e.g., phosphorus).
Total P stock in the sediment of LP (Figure 4
) was approximately four and nine times higher than stocks reported by Waters and Webster-Brown [4
], respectively, for Lake Forsyth in New Zealand, and by Ul Solim and Wanganeo [12
], for Dal lake in India, both shallow lakes located in temperate zones. The total stock of P in Lake Forsyth was reported as 182 Mg and Dal lake, 79 Mg. Total N stock in the sediment was not determined for either lake.
These results may be related to the high P inputs and the high hydraulic residence time of Lake Paranoá. Inputs and hydraulic residence time are key factors for P storage [7
]. LP had its hydraulic residence time increased due to the water crisis registered between 2016 and 2017. Besides this, most (>98%) of the P total load in Lake Paranoá originates from the two WWTPs (WWTP North and WWTP South) located on the shores of the lake. Most lakes and reservoir studies that have been reported in the specialized literature are not impacted by WWTP effluents.
3.3. N and P Budget
The results of the TN and TP mass budget in the hydrological period of 2016–2017 indicated that Lake Paranoá acts as a sink for both nutrients (Table 3
). Different forms of N show different behavior: inorganic forms are retained (NH4+
-N and NO3−
-N) in the lake; while the organic form is exported (TON), regardless of the period considered. In the case of P, all chemical forms exhibited the same behavior, both the organic form (TOP) and the inorganic one (PO43−
-P) were retained during the entire study period (wet and dry season).
-N retention tended to be lower during the dry season than in wet season, while NO3−
-N retention tended to be higher. TON exportation tended to be lower during the dry season more than in the wet season. NH4+
-P and TOP retention was similar in both seasons (Table 3
). Despite the export of TON, there was a net balance in retention and export over the study period owing to the higher retention of N inorganic forms.
TON was the only exported form of N throughout the period, with a high export percentage (67%) of the total incoming load (Figure 5
). This information suggests that there are other important sources that were not part of the study, such as N fixation and discharge by WWTPs, the latter being the most probable.
Nutrient budgets have been estimated for a series of lakes and reservoirs around the world (Table 4
). In some cases, both inorganic and total phosphorus and nitrogen [11
] were analyzed; while in others, only the inorganic and total phosphorus and nitrogen and the total Kjeldahl nitrogen were determined in the nutrient budget [41
]. Since many studies have considered only the nutrient budget for TN and TP, a data comparison with regard to organic forms was difficult.
Although quite unusual among the systems already studied, the export of N was also recorded by Cook et al. [41
] in two shallow reservoirs in a semi-arid region of Australia and by Garnier et al. [13
] in three deep reservoirs in France. Cook et al. [41
], for example, found an average retention of 7% of total nitrogen and an average export of 6% of organic nitrogen. According to the authors, this export of N may have been related to high rates of biological N fixation, which is more common in subtropical and semi-arid climatic reservoirs, leading to a reduction in the retention capacity of this nutrient. The N exportation found by Garnier et al. [13
] was very high (100%–350%) of the NH4+
According to Matson et al. [42
], anthropogenic activities have more than doubled the N input into terrestrial ecosystems. Environments initially limited in N have retained this element in vegetation and microbial growth, accumulation in biomass, and possibly also in soil organic matter. However, at some point, the input of N can exceed the biological demands and the system will begin to lose its retention capacity for this nutrient, which may explain the exportation to surrounding aquatic environments, as in the case of Lake Paranoá.
The annual N budget for dissolved inorganic nitrogen (DIN), however, showed retention of this form of N, equivalent to 85% of the total incoming load (Figure 6
). The total load retained in LP can be attributed to the increase of the hydraulic retention time (>2 years) and high inputs, especially from WWTPs (WWTP North and WWTP South). It is noteworthy that WWTP North at 49% of its capacity utilization, while WWTP South at 88%.
The retention of P in Lake Paranoá (Figure 7
and Figure 8
) fit within other values reported in the literature (Table 4
). The retention of P in LP was similar to what has been reported for the Valle de Bravo [9
], but higher when compared to Swiss lakes [8
]. Despite the high hydraulic residence time of Swiss lakes, low P retention may be associated with low inputs imported to the lakes (Table 4
Although it is difficult to compare different lakes, Brazilian shallow lakes with lower hydraulic residence time (<1 year) showed retention of phosphorus ranging from 15% to 67%. The highest input, in Lake Broa (3260 Mg), resulted in the lowest retention percentage (Table 4
Amongst the factors that can affect nutrient retention (particularly phosphorus), the parameter hydraulic retention time predominates [7
The hydraulic retention time of LP was calculated for the study period concurrent with the period of increased water retention rate in the reservoir in the function of the lower flushing rate from the dam. This was done to provide an emergency catchment of 700 L s−1 water for human consumption as regulated by Company of Environmental Sanitation of the Federal District (CAESB) in October 2017 to meet the water demand during the period of Brasilia’s water crisis, 2016/2017. At that time, the volume of L. P’s main reservoirs was used to supply the population and the Descoberto and Santa Maria reservoirs reached the lowest levels in history. As this location has been chosen for permanent water collection by CAESB, this higher TR will likely become normal.
With high hydraulic retention times, the flushing rate from Lake Paranoá was lower and the retention of TP in the lake system was very high (99% of inflows). This phosphorus retention coefficient (RP
) of 0.99 is at the higher end of what has been recorded in the published data. Kõiv et al. [7
] reported RP values of between 0 and 0.93 with a mean of 0.5 for 54 lakes and reservoirs worldwide. According to these authors, the standard P retention in natural lakes corresponds to 47% ± 28% and in reservoirs to 42% ± 22% of total external input of phosphorus, therefore, there are no meaningful differences among lakes and reservoirs. It should be noted, however, that the majority of the data originated from temperate zones in Europe and North America.
Another important factor in phosphorus retention is the relative depth (ZR
). According to Nõges et al. [53
] and Kõiv et al. [7
], phosphorus retention in lakes and reservoirs is directly proportional to relative depth. This was validated by the latter only for great lakes (>25 km2
) and with shorter hydraulic retention time (TR
< 0.3 years). Considering the relative depth of Lake Paranoá (ZR
= 0.55), our results confirmed the previous hypothesis that phosphorus retention capacity is higher in deeper lakes with higher ZR
than in shallow ones with deeper ZR
. On the order hand, our results indicated that this relationship was also valid for a reservoir with a higher hydraulic retention time, as in the case of Lake Paranoá.
Although the work of Kõiv et al. [7
] exhibits significant considerations concerning the P retention process in lakes and reservoirs, most of the data was derived from climate regions different from LP, that is a tropical reservoir in a savannah climate. Furthermore, it is important to point out that while nutrient cycling is a common process in all ecosystems, the amount of element retention in the water column and in the sediment (N and P storage), as well as the diffusive fluxes that occur among these compartments, are relative to each ecosystem.
This study, then, introduces an important contribution to the knowledge of the behavior of these nutrients in deep urban reservoirs in a tropical savannah climate, where the retention and export mechanisms are still not well known.