3.1 Eutrophication potential of drinking water reservoirs by CTSI assessment
The results of eutrophication potential by using trophic state index assessment are shown in Fig. 1. In Taiwan, there are 5 reservoirs that have a eutrophication potential of more than 50%, namely Feng-Shan, Ming-Te, Ching-Mien, Cheng-Ching-Hu, and Pai-Ho reservoirs. Feng-Shan reservoir is the most severely eutrophicated reservoir in Taiwan and classified as hypereutrophic based on CTSI value, as shown in Fig. S1. It has the highest potential than other reservoirs, approximately 100% followed by Ming-Te, Ching-Mien, Cheng-Ching-Hu, and Pai-Ho Reservoirs, respectively. This present finding is consistent with a previous trophic state classification study [21]. High eutrophication levels affected by nonpoint source pollution from fertilizers and farm drainage cause excessive algal growth [32]. Feng-Shan reservoir is fed by a river tributary that is downstream from hundreds of pig farms. Farmers discharged their pig manure massively into the river [33], rich in organic matter, nitrogen, and phosphorus [34]. In Taiwan, the main watersheds are mostly fragile slate, and poor soil conservation induces sand and gravel upstream are easily flushed down to the reservoir by high precipitation [21]. The adverse effects of eutrophication induce loss of dissolved oxygen in the reservoir [35] and the rapid growth of harmful phytoplankton [36]. Eutrophication assessment is still necessary in order to prevent the detrimental impact on reservoirs, especially in the anthropogenic influences [37]. The results have shown a significant divergence of eutrophication level for 21 drinking water reservoirs based on CTSI evaluation.
3.2 Deviations of three TSI variables in CTSI application
The analysis of trophic state index deviations was followed by a previous study [29]. It describes that approximately 78% of the studied reservoirs were categorized into a small particulate predominance based on the relations between "TSI (Chl-a) – TSI (SD)" and "TSI (Chl-a) – TSI (TP)" during all the seasons as shown in Fig. 2. The influence of larger particulates and the effect of algae growth are negligible. It indicates that non-algal turbidity occurred in most reservoirs, as evidenced in Fig. S2. Non-algal particulates such as color or turbidity, which are derived from external or internal sources, produce sufficient light extinction to limit algal productivity [17]. These findings also suggest that phosphorus is not the dominant factor limiting the trophic state for non-algal impact reservoirs. This outcome is contrary to a recent study which has found that most studied reservoirs show a dominance of larger particles and phosphorus limitation occurs at a high algal growth condition [38]. This discrepancy could be attributed to the dominant nutrient regulating the trophic state. When there is available surplus phosphorus in reservoirs, nitrogen is known to be the dominant factor regulating the trophic state or some other factors that co-varied with nitrogen [29, 39, 40]. In contrast, the small particulates predominance condition indicates that a low water transparency level is dominantly formed by inorganic sediment than Chl-a [41].
For further understanding in TSI deviations at various seasons, this study identified the effect of seasonal variations on the trophic state index values of 21 drinking water reservoirs in Taiwan. Figure 3 shows that non-algal turbidity (Region III) dominantly occurred during all seasons. However, small particulates predominance and P limitation were markedly formed in the summer and autumn season. It could be attributed to the algae growth is likely increased in high temperature. This finding is also supported by some studies which have confirmed that high temperature deteriorates water quality and increases cyanobacteria occurrence and diatom blooms in reservoirs to cause significant eutrophication [42, 43, 44]. Another study also has proven that wind speed in summer has a substantial impact on alga growth. In addition, wind speed in different seasons could gradually suspend the phosphorus and promote algal growth [45]. It can be concluded that seasonal variations affect the level of eutrophication for drinking water reservoirs, especially in the summer and autumn seasons.
Although Taiwan is categorized as subtropical/tropical area, which has a productive season, most drinking water reservoirs in Taiwan were classified as mesotrophic, approximately 57% as shown in Fig. 4a. It could be associated with the residence time in reservoirs, which is very low compared to the lakes. Higher residence time can affect the trophic state index more significantly [12]. However, approximately 30% of trophic state index classification was categorized as eutrophic (Fig. 4a) based on the averaged CTSI values higher than 50 occurs in one-year evaluation, where the highest ratio occurs in the summer season (Fig. 4b). It could be associated with the intense rainfall in subtropical/tropical areas that significantly suspend sediment (i.e., inorganic particles). In addition, the increased temperature in summer also improves Chl-a level attributed to the increasing algae growth. Additionally, the trophic state level is much higher in the summer season.
3.3 Empirical relationship of TSI variables
The correlation of trophic state index variables was quantified to understand the relationship between TSI variables. Figure 5 shows that there is an insignificant correlation among three TSI variables. Only some reservoirs have a medium correlation between TSI (SD) and TSI (TP). It is similar to a previous study in Taiwan, which has shown that the correlation between trophic state index was insignificant [15]. The outliers can be extremely sensitive to the Pearson correlation values cause the strength of a relationship can be exaggerated or diminished by extreme values indicate that the variables are not normally distributed. Therefore, it is inappropriate to use trophic state index variables when the correlations are not normally distributed [46]. This finding has indicated that the CTSI application could not show the actual trophic state of drinking water reservoirs without strong algal impact in Taiwan.
For further investigation in the contribution to CTSI at independent TSI, the calculations of trophic state index higher than 50 on 21 reservoirs in Taiwan are presented in Table 1. It has revealed that the proportion of trophic state index variables is dominated by TSI SD, approximately 63% over TSI (Chl-a) and TSI (TP). TSI (Chl-a) and TSI TP only contributed 24% and 20%, respectively to the CTSI level. In the specific case of CTSI higher than 50, TSI (SD) significantly contributed as much as 96%. Figure 6 illustrates that TSI (SD) values had a higher level compared with the other TSI variables. This result is similar to a recent study, which has shown that TSI (SD) plays a dominant role in the CTSI levels [16] due to the high rainfall rate, heavy storm runoff [23], and direct nutrient discharges [47], leading to increase the high turbidity level of reservoirs in Taiwan. The deviations of three TSI values have indicated that the CTSI has limited applicability on water quality management for Taiwan reservoirs.
Table 1
The proportion of TSI variables are greater than 50 for 21 reservoirs in Taiwan
CTSI variables
|
The proportion of TSI>50
|
The proportion of TSI>50 as CTSI is higher than 50
|
TSI(SD)
|
63
|
96
|
TSI(Chl-a)
|
24
|
60
|
TSI(TP)
|
20
|
58
|
Water transparency in the reservoir system can be controlled by suspended solid [48] and phytoplankton biomass [38]. The correlation between TSI (SD) and influencing water quality parameters such as turbidity and suspended solid were investigated further. For the purpose of comparison, the correlation between TSI (Chl-a) and suspended solid was observed. Figure 7 shows that the correlation of suspended solids and chlorophyll concentration was negligible. On the contrary, there was a significant correlation between turbidity, suspended solids, and TSI (SD). This result is similar to the previous studies in Taiwan, which have confirmed that suspended solid played the biggest TSI (SD) value contribution [15, 16]. These findings suggest that suspended solid strongly influences the water clarity because non-algal turbidity occurred in most reservoirs in Taiwan.
Figure 8a shows that when TSI (SD) values lesser than 50, 2 reservoirs have a significant correlation between TSI variables, namely Jen-Yi-Tan and Hu-Shan Reservoirs. While the other reservoirs have an insignificant correlation. On the contrary, Figure 8b shows that when TSI (SD) values higher than 50, it is highlighted that some reservoirs have a medium correlation between TSI (TP) and TSI (SD). It indicates that particulate phosphorus influenced the water transparency level. While there is an insignificant correlation of other TSI variables. These findings also suggest that putting TSI (SD) in CTSI calculation is unable to show the distribution level of each CTSI variable for non-algal impact reservoirs. It can be concluded that when non-algal turbidity occurs in reservoirs, the use of TSI (SD) data to infer algal biomass and trophic state is inappropriate where even moderate amounts of non-algal turbidity are present [49]. Despite the fact that non-algal turbidity occurs in most reservoirs in Taiwan, the CTSI can still be used for a specific condition. The CTSI application is more reasonable as the TSI (SD) level limitation is lower than 50 or at a low turbidity level (<5 NTU) in Taiwan. Thus, the algae-carried Chl-a is more suitable as a single trophic state index to evaluate the eutrophication potential for Taiwan reservoirs.
3.4 Relationship of eutrophication influencing factors
The result of Carlson's trophic state index based on two-dimensional analysis is shown in Table 2, the first component (C1) obtains 55,87% of the total variances. Based on the trophic state index value, this study has revealed that TP has the highest factor loading of the main component, while the factor loading of Chl-a, COD and NH3 are similar. The correlation between TSI (TP) and TSI (SD) or TSI (Chl-a) is improved for most cases where it has a medium Pearson correlation value when the TSI (SD) values are higher than 50 at high turbidity level (Fig. 8b). Furthermore, as seen in Fig. 9, the TSI (Chl-a) has the insignificant factor load of the second component, where the TSI (SD) has the highest factor load along with a close tendency of SS. In addition, there is a negative correlation between Chl-a and SD. It suggests that most drinking water reservoirs in Taiwan are operated under long-term turbid impact with limited algal particulates dominance. The critical finding in this study is total phosphorus is confirmed to be the most significant influencing factor on water quality in terms of COD, NH3 and Chl-a (as seen in Table 2). Based on PCA inferences, it indicates that CTSI approach could often leads to confusion because the assessment criteria for CTSI can be degraded by factors such as suspended solids, which does not actually reflect the eutrophication potential. Thus, the best strategy for water quality management is proposed to strengthen TP regulation in the upstream catchment area of drinking water reservoirs and minimize TP loading in reservoirs, which could significantly influence the variety of algal growth, ammonia, and organic matter loadings.
Table 2
Factor analysis results of the significant water quality parameters
Water Quality
|
Component 1
|
Component 2
|
SD
|
-0.103
|
-0.934
|
Chl-a
|
0.768**
|
0.259
|
TP
|
0.907**
|
0.127
|
SS
|
0.378
|
0.687*
|
COD
|
0.748**
|
0.329
|
NH3
|
0.810**
|
0.007
|
TSI(TP)
|
0.774**
|
0.433
|
TSI(SD)
|
0.213
|
0.949**
|
TSI(Chl-a)
|
0.625*
|
0.333
|
Initial Eigenvalues
|
55.871%
|
16.924%
|
Kaiser-Meyer-Olkin
|
0.78
|
p-value
|
<0.001
|
Double asterisks (**): a strong factor loading; a single asterisk (*): a medium factor loading |