Spatial variations in water quality
Spatial variations of water quality parameters were shown in Fig. 2. Overall, the pH, DO, TN and TP values showed no notable variation trends from the upstream to the midstream. The pH values ranged from 6.00 to 8.20, with a mean pH of 6.69. The DO, TN and TP concentrations were 5.90–11.20, 0.54–0.94 and 0.025–0.044 mg/L, respectively, and their mean concentrations were 7.55, 0.72 and 0.03 mg/L, respectively. In addition, CODMn, BOD5 and DOC concentrations displayed significant decreasing trends from the upstream to the midstream. Their concentrations varied in the range of 6.10–14.80, 1.80–4.20, and 8.50-15.66 mg/L, respectively, and their mean values were 8.77, 2.79, and 10.78 mg/L, respectively. However, the NH3-N concentration ranged from 0.19–0.64 mg/L with a mean concentration of 0.42 mg/L, and it exhibited a fluctuating decreasing trend from the upstream to the midstream.
Based on the above data, the classifications of water quality parameters were evaluated. Overall, DO, BOD5 concentrations and pH values met the level Ⅰ water standard. Both TP and NH3-N concentrations met the level Ⅱ water standard. However, TN and CODMn concentrations met the level Ⅲ and Ⅳ water standards, respectively. Significantly, CODMn concentrations were unusually high and did not satisfy the level Ⅲ water standard for potable water. Therefore, TN and CODMn were the key pollution indices in this study. Meanwhile, CODMn, BOD5 and DOC concentrations were the highest at the H1 site. CODMn and DOC concentrations of sites H1-H5 were much greater than those of sites H6-H13, indicating that the sites in the background area had higher CODMn, DOC loadings and the relative content of DOM than those in the anthropogenic disturbance area. The results showed that the HRB during the wet season had relatively high CODMn concentrations, and headwaters presented a high background value.
Figure 2
UV-visible spectra
Figure S1 showed several spectral parameters variations from the upstream to the midstream. UV254 could indicate the relative content of natural humic-like macromolecular organic matter and aromatic organic matter containing C = C and C = O. SUVA254 was positively correlated with the aromatic carbon content in DOM (Weishaar et al., 2003). For this study, UV254 and SUVA254 values varied from 0.196 to 0.473 cm− 1, and from 5.264 to 6.956 Lmg·C− 1m− 1, respectively. The SUVA254 values were above 4 Lmg·C− 1m− 1, indicating that DOM in the HRB during the wet season predominantly consisted of aromatic and high-molecular-weight natural organic matter (Ates et al., 2007). Meanwhile, UV254 and SUVA254 values of sites H1-H5 were more than those of sites H6-H13, suggesting that the aromaticity and molecular weight of DOM in the background area were higher than those in the anthropogenic disturbance area. Besides, S275 − 295, S350 − 400, and SR values were 0.014–0.016 nm− 1, 0.018–0.021 nm− 1, and 0.743–0.838, respectively. We observed that S275 − 295 < S350−400 at each site in this study (Figure S1c; Figure S1d), suggesting that DOM in the HRB was dominated by terrestrially derived substances. Helms et al. (2008) illustrated that lower SR values were indicative of a growth in DOM molecular weight. Sites H1-H5 had lower SR values than sites H6-H13 (Figure S1e), suggesting that DOM in the background area had higher molecular weight than that in the anthropogenic disturbance area. In a word, the results were mainly attributed to the introduction of natural organic matter into the HRB through rainfall and runoff during the wet season (Oh and Choi, 2022).
Characteristics of EEMs-PARAFAC components
Four fluorescent components (C1-C4) of the 13 samples in the HRB were recognized using PARAFAC (Fig. 3; Table 2). Based on previous studies, a lot of peaks matching the selected fluorescent components were also found. The fluorescent components summarized in this study were similar to those described in previous studies, indicating that common attributes could be identified. C1 exhibited Ex/Em maxima at (275, 365)/470 nm, corresponding to UVC-humic-like substances with high molecular weights (Gao et al., 2017). Prior studies showed that C1 was usually abundant in the allochthonous terrestrially-derived DOM (Stedmon et al., 2003). While C2 had Ex/Em maxima at (250, 330)/440 nm, resembling UVA-humic-like substances with low molecular weight (Harun et al., 2016). Besides, C3 had Ex/Em maxima at (230, 360)/440 nm and resembled marine humic-like substances. The fluorescence maxima of C3 were shifted towards lower excitation wavelengths (blue shift) relative to the peak M. Parlanti et al. (2000) demonstrated that biological activity was closely related to the blue-shifted fluorescence phenomenon. Finally, C4 showed Ex/Em maxima at (220, 280)/400 nm, similar to the tryptophan-like fluorophore. Zhao et al. (2017) found that C4 originating from autochthonous sources was closely associated with microbial activities and autochthonous DOM production. In summary, C1, C2 and C3 were assigned as allochthonous humic-like substances, whereas C4 was characterized as autochthonous protein-like substances.
Table 2
Spectral characteristics of excitation and emission maxima of four fluorescent components identified by PARAFAC modeling compared with previously identified sources
Component no. | Peak max position Ex/Em | Type | Tradition peak Coble (1996) | Comparison with other studies using PARAFAC |
C1 | 275 nm (365 nm)/470 nm | UVC humic-like | Peak A: 260/380–460 Peak C: 350/420–480 | C3: 270 (360)/478a C2: 240 (370)/480b |
C2 | 250 nm (330 nm)/440 nm | UVA humic-like | Peak A: 260/380–460 | C4: <250 (295)/358c C2(HLC2): 250 (310)/410d |
C3 | 230 nm (360 nm)/440 nm | Marine humic-like | Peak M: 312/380–420 | C5: 290/338e C2: 300/402f |
C4 | 220 nm (280 nm)/400 nm | Tryptophan-like | Peak T: 275/340 | C4: <240 (279)/327g C3: <250 (290)/365h |
a Stedmon et al. (2003) |
b Kowalczuk et al. (2013) |
c Yamashita et al. (2011) |
d Mendoza and Zika (2014) |
e Harun et al. (2016) |
f Pitta et al. (2017) |
g Yang et al. (2018) |
h Lee et al. (2015) |
Figure 3
Table 2
To better research the variability of DOM from the upstream to the midstream in the HRB, percentages and fluorescence intensities of four fluorescent components in our samples were evaluated. C1, C2, C3 and C4 accounted for 21.0%, 33.9%, 26.5% and 18.6%, respectively (Fig. 4a). Specifically, C1, C2 and C3 accounted for 81.4%, while C4 only accounted for 18.6%. Furthermore, the proportion of humic-like components displayed a decreasing trend along the HRB. In contrast, the proportion of protein-like component increased from 15–20% along the HRB. The results showed that DOM was dominated by terrestrially derived substances. The total fluorescence intensity of four components displayed a decreasing trend along the HRB (Fig. 4b). Because headwaters in the HRB were affected by humic substances in the forest, the sites in the background area had stronger fluorescence intensity than those in the anthropogenic disturbance area.
Figure 4
Fluorescence parameters
Three fluorescence parameters of DOM were calculated based on their corresponding definitions (Fig. S2). FI values varied from 1.42 to 1.56, all of which were between 1.4 and 1.9 (Fig. S2a). It implied that DOM originated from terrestrial and microbial sources. In addition, BIX values were from 0.54 to 0.62 (Fig. S2b). Huguet et al. (2009) reported that lower values of BIX (< 0.7) corresponded to lower DOM production. Thus, this revealed that DOM in the HRB had fewer authigenic components. On the whole, BIX values exhibited an increasing trend from site 1 to site 13, indicating that there were more and more freshly produced autochthonous DOM from the upstream to the midstream. This was consistent with our study describing variations in the relative distribution of C4 in the HRB (Fig. S2b; Fig. 4a). Finally, HIX values ranged from 6.55 to 13.33, all of which were above 6 (Fig. S2c), indicating that DOM in the HRB displayed important humic character (Table 1). The average HIX values of sites H1-H5 and sites H6-H13 were 11.24 and 8.855, respectively. Thus, the background area had a higher humification degree than the anthropogenic disturbance area.
FT-IR spectra
The structure of DOM was characterized by comparing absorption peaks of FT-IR spectra. Due to the complexity of DOM, there will be an overlap among the absorption bands. FT-IR spectra of all samples showed a similar trend (Fig. S3). The strong absorption band around 3411 cm− 1 could originate from the overlap of O-H stretching of alcohol, phenol, carbohydrate and carboxylic acid compounds (Abdulla et al., 2010). The result showed that DOM contained lots of hydroxyl and carboxyl groups. The band around 1723 cm− 1 because of C = O stretching was a weak peak or a shoulder peak (Yang et al., 2015). The band around 1619 cm− 1 originated from the stretching of C = C and C = O in the amide Ⅰ band, while the band near 1416 cm− 1 was associated with N-H bending and C-N stretching of the amide Ⅱ band (Wang et al., 2017). The result indicated that DOM in the HRB had protein-like substances. In addition, the band around 1482 cm− 1 represented bonds with C-H bending in aliphatic alkane structures (Lin et al., 2021). There was a strong absorption band around 1103 cm− 1, which could result from the vibrational coupling of C-O stretching of carbohydrate, ether and ester compounds (Abdulla et al., 2010). Hence, the functional groups showed that DOM in the HRB contained aromatic, aliphatic, carbohydrate and protein compounds.
PCA and correlation coefficient matrix analysis
Statistical analysis was performed to further study DOM characteristics. The data matrix constructed in PCA was based on four PARAFAC components (C1-C4) from all 13 sampling sites. According to our sampling design, the sampling sites were divided into two groups in this study. Specifically, group A, including sites H1-H5, was in the background area, and group B, including sites H6-H13, was in the anthropogenic disturbance area. The first two principal components explained 99.9% of the total variance. In detail, the first principal component (PC1) and the second principal component (PC2) were 93.8% and 6.1%, respectively. The biplot, including loading values and scores, was presented in Fig. 5a. C1, C2 and C3 were clustered in the lower quadrant, which had positive PC1 loadings and negative PC2 loadings. However, C4 was alone in the upper quadrant with positive PC1 loadings and PC2 loadings. Previous studies found that the loading values of principal components could clearly distinguish samples originating from allochthonous and autochthonous sources (Yamashita et al., 2013). Hence, it could be concluded that DOM was mainly derived from terrestrial humic-like substances represented by PC1, followed by autogenic protein-like substances represented by PC2. Additionally, PC1 and PC2 scores from 13 sampling sites displayed the spatial differences in the HRB (Fig. 5a). Positive PC1 scores were found at sites H1-H5, while negative PC1 scores were found at sites H6-H13. Specifically, in the background area, except for H4 and H5 sites, the PC1 scores of the others were greater than their PC2 scores. On the contrary, in the anthropogenic disturbance area, except for H8 and H11 sites, the PC1 scores of the others were less than their PC2 scores. The distribution of PC1 and PC2 scores showed that the content of humic-like components in the background area was greater than that in the anthropogenic disturbance area, but the content of protein-like component exhibited an opposite phenomenon. Significantly, PC2 was only responsible for the 6.1% variance. Thus, PC2 may be less valuable than PC1 with respect to assessing their contributions. Consequently, PC1 distributions revealed that humic-like components were dominant in the HRB, which was consistent with that in 3.3.
The Pearson correlation coefficients (r) among selected parameters were presented in Fig. 5b. There were obviously positive relationships among the fluorescence intensities of C1, C2 and C3 (r > 0.99, p < 0.05), suggesting that C1, C2 and C3 might have a common source and a similar transport process (Wu et al., 2017). Meanwhile, the first three fluorescence intensities exhibited strongly positive relationships with DOC and CODMn concentrations (r > 0.92, p < 0.05), which suggested that DOC and CODMn were mainly derived from humic-like substances. Besides, the first three fluorescence intensities were positively correlated with HIX values, whereas they were negatively related to BIX values. And HIX values were also negatively related to BIX values (r=-0.80, p < 0.05). Therefore, the humification degree was further more than the biodegradation degree in the HRB.
Figure 5