4.1 Composition of DOM fluorescent components in pore water
C1 was recognized as fulvic acid-like substances with the maximum Ex/Em of 254/425 nm (Santín et al., 2009). C2 displayed a single Ex/Em peak of 270/315 nm and was assigned to protein-like structures similar to tyrosine (Murphy et al., 2011). C3 exhibited two peaks at 235 nm/ 354 nm (Ex/Em) and 275 nm/354 nm (Ex/Em), which were classified as tryptophan-like substances (Yang et al., 2015). C4 components were identified at Ex/Em wavelengths of 225/300–400 nm, which resembled aromatic proteins (Huang et al., 2022a).
C1b (Ex/Em = 235/412 nm) was similar to C1 and was categorized as a fulvic acid-like substance. C2b (Ex/Em = 225,275/320 nm) were tyrosine-like substances (Cheng et al., 2018). C3b comprised of two peaks at Ex/Em wavelength pairs of 264/445 nm and 355/445 nm, implying humic-like compounds with relatively high molecular weights (Wheeler et al., 2017). C4b had a primary fluorescence peak at an Ex/Em wavelength pair of 247/390 nm and a secondary peak at 298/390 nm, indicating microbial humic-like substances, which were relatively aliphatic and with low molecular weight (Murphy et al., 2011).
4.2 Influence of drought-rewetting process and salinity variation on DOM fluorescent components
Rewetting after drought resulted in lower HIX in pore water (Fig. 2), indicating that this process decreased humification of DOM, which was associated with increasing proportions of tyrosine-like components on the first day of period B (Fig. 3). Firstly, the drought period inhibited bacterial activities and utilization of protein-like substances and increased production of extracellular polymeric substances, including protein-like DOM (Gionchetta et al., 2019). Secondly, the drought-rewetting process resulted in higher preferential adsorption of carboxyl over amide, and fewer tyrosine-like components were adsorbed on the clay surfaces (Olshansky et al., 2018). Rewetting was reported to increase bioavailability of DOM and bacterial activities because occluded carbon sources were liberated for fast microbial reactivation (Ylla et al., 2011; Gionchetta et al., 2019), thus, higher BIX values were observed on the first day of period B than on the last day of period A (Fig. 2). The decreasing trend of BIX in period B may because that labile DOM was gradually exhausted. Notably, values of SR and E2/E3 decreased after the drought-rewetting process (Fig. 2), indicating an increase in the aromaticity and molecular weight of DOM, which was linked with the significantly increased Fmax of humic-like components (Fig. 3). Huang et al. (2022b) found that cycles of drought and wetting were conducive to DOM components becoming more aromatic, hydrophobic, and humified because the labile fractions were utilized and/or transformed to aromatic structures by the bacterial communities. Yamashita and Tanoue (2003) also considered that newly produced humic-like components may represent degradation products of amino acids and protein-like components by microbial metabolism. Large sized polysaccharides secreted by bacteria during drought period that may also account for increase in DOM molecular weight (von Schiller et al., 2015). In addition, direct solar radiation on sediments and relevant heat during drought conditions accelerated the humification of DOM by oxidative polymerization reactions (del Campo et al., 2019). The Fmax values of DOM components were observed to greatly increased after the drought-rewetting process in group S1. It was reported that moisture of sediment led to formation of cracks during drought conditions, which increased surface area and oxygen availability of sediments and promoted aerobic destruction of organic matter, thus leading to more DOM release (Schiebel et al., 2019). However, there was little influence of the drought-rewetting process on Fmax values of DOM components in groups S2 and S3, suggesting that higher salinity inhibited the release processes or mobilized the DOM from pore water to surface water.
The HIX values of S1 were lower than those of S2 and S3 during period A, and SR and E2/E3 had the opposite regularities, indicating that higher salinity of lake water caused DOM in pore water to become more aromatic and humified with a higher molecular weight. Previous studies reported that an increase in salinity caused DOM to show lower aromaticity in surface water (Xu et al., 2020a; Douglas et al., 2021), which was different from the results of this experiment, probably because the higher salinity may have facilitated photodegradation of aromatic components in surface water, while these components in pore water were not influenced by light. Significantly lower BIX values were observed in S3 (P < 0.05), indicating that increase in salinity decreased autochthonous DOM and inhibited effect of bacteria on DOM composition. The protein-like fractions were considered to be preferentially consumed during biodegradation (Zhou et al., 2019), as shown in Fig. 3. These processes were accelerated at high salinity, especially biodegradation of tyrosine-like components and aromatic proteins.
The joint effects of salinity increase and drought-rewetting process were to make the more humified, more aromatic, less bioavailable DOM remain in pore water, and to preserve DOM concentrations at lower levels, especially when the salinity was at the concentration of 6,000 mg/L, which influenced cycling and utilization of carbon in brackish-water lake, and further affected the lacustrine ecosystem.
4.3 Response of bacterial relative abundances in sediments to DOM components, salinity, and drought-rewetting process
It has been reported that drought-rewetting process caused rupture of many cells by osmotic shock, especially bacterial communities dominated by Proteobacteria (Schimel et al., 2007; Marxsen et al., 2010). In this study, decrease of Shannon indexes and relative abundances of Proteobacteria after drought-rewetting process were observed (Fig. S5), the values had not been recovered to levels before this process by the end of the experiment. Different from the previous work, relative abundances of Proteobacteria in group S3 were similar before and after the drought-rewetting process, likely due to its resistance to salinity (Chen et al., 2017) and stress of salinity on many other phyla. Moreover, the drought-rewetting process enhanced the differences of in bacterial communities among groups. As shown in Fig. S5, there were no significant differences of Shannon indexes among groups before drought-rewetting process, but Shannon indexes in S1 were significantly lower than those in S2 (P < 0.05) after this process, indicating that this process magnified the influence of salinity on bacterial communities, may result in greater differences of DOM at different salinity.
According to Fig. 5a, Hydrogenophilaceae showed positive correlations with salinity and Fmax of C1 and C3 and negative correlations with Fmax of C2 and C4 in period A. Thiobacillus was the dominant genus of Hydrogenophilaceae, which could adapt to high salinity environment (Xu et al., 2020b). Thiobacillus could degrade organic matter containing sulfur (Wang et al., 2018), probably producing fulvic acid-like substances and contributing to higher HIX and lower SR and E2/E3 under higher salinity. Thiobacillus decreased pH by sulfur oxidation, inducing release of DOM indirectly (Zhu et al., 2022) and secreted extracellular polymeric substances containing proteins (Ye et al., 2021), accounting for the relationships between relative abundances of Hydrogenophilaceae and Fmax of protein-like components. The relative abundances of Oxalobacteraceae were negatively correlated with salinity, indicating that this family was not tolerant to high salinity. Moreover, the relative abundances of Oxalobacteraceae were positively correlated with Fmax of C2 and C4 and negatively correlated with Fmax of C1 and C3. Massilia was the dominant genus of Oxalobacteraceae and could biodegrade refractory DOM to compounds with low complexity and aromaticity (Qiao et al., 2021), suggesting that this genus degraded fulvic acid-like components. Massilia was reported to produce indole derivatives by utilizing tryptophan (Agematu et al., 2011), which was one of reasons for negative correlation between relative abundances of Oxalobacteraceae and Fmax of C3. The correlations between Flavobacteriaceae and salinity and Fmax of four DOM components were similar to those of Oxalobacteraceae because the dominant genus of Flavobacteriaceae was Flavobacterium, which could utilize amounts of dissolved proteins and degrade macromolecular organic matter (Ye et al., 2020). Anaerolineaceae, Comamonadaceae, Desulfobacteraceae, Pseudomonadaceae, and Xanthomonadaceae had relatively little influence on DOM components during period A.
After drought-rewetting process, interaction patterns between dominant bacteria and DOM components were greatly changed. As illustrated in Fig. 5b, Hydrogenophilaceae showed a negative correlation with four DOM components, which was different from that before drought-rewetting process, because of drought stress and alteration of sediment environments. Correlations between sulfate-reducing bacterium Desulfobacteraceae and salinity and Fmax of DOM components were similar to those of Hydrogenophilaceae, likely because of its positive correlation with Thiobacillus. Anaerolineaceae with strong resistance to hypersaline conditions due to protection of extracellular polymeric substances (He et al., 2020) and acted as a primary degrader for refractory DOM, such as humic-like substances (Zhang et al., 2018). Thus, their relative abundances were positively correlated with salinity and negatively correlated with Fmax of C1b, C3b, and C4b. Planococcaceae was positively correlated with Fmax of C1b and C2b, suggesting that Planococcaceae participated in the production of fulvic acid-like and tyrosine-like substances. Clostridiaceae is a proteolytic bacterium that utilizes protein as its preferred substrate (Kim et al., 2020) and produces fulvic-like, humic-like, and tyrosine-like components during the biodegradation of proteins, resulting in positive correlations with Fmax of these components. Oxalobacteraceae, Caulobacteraceae, and Chitinophagaceae had relatively little influence on DOM components during period B.
Salinity variations and the drought-rewetting process greatly altered structures of bacterial community, probably due to stress and lack of labile carbon sources. Bacterial community structure was correlated with its functions (Riah-Anglet et al., 2015), thus the alteration of bacterial communities, especially dominant bacteria associated with DOM components, was one of crucial reasons for variation of DOM compositions in pore water.
4.4 Correlation among the dominant families under drought-rewetting process and salinity gradient
As shown in Fig. 6, there were 24 significant positive correlations and 25 significant negative correlations (P < 0.05) among dominant families in period A, there were 38 significant positive correlations and 25 significant negative correlations (P < 0.05) among dominant families in period B. The drought-rewetting process increased the positive correlations among the bacteria. The results of this study support Stress Gradient Hypothesis: competitive interactions decrease and facilitative interactions increase along the stress gradient (Bertness and Callaway, 1994). Previous studies have demonstrated that environmental stress increased facilitation among bacteria (Piccardi et al., 2019; Hernandez et al., 2021). In this study, bacteria may conduct metabolic cross-feeding and interactions among them switch from competition to facilitation when labile DOM is depleted (Goldford et al., 2018), which can be supported by increase of positive correlations between Hydrogenophilaceae and other bacteria after drought-rewetting process. It was worth noting that Hydrogenophilaceae was dominant family and was correlated with DOM components. Romdhane et al. (2021) found that modified bacterial interactions greatly changed cycling of nitrogen and carbon in soil. The alternation of the interactions between Hydrogenophilaceae and other bacteria may affect its role on DOM transformation and cycling. Moreover, decrease in ratio of negative and positive correlations among bacteria results in bacterial communities becoming more precarious, because the bacteria support each other and perturbations on any of them can easily influence the whole system (Coyte et al., 2015). Therefore, drought-rewetting process destabilized the bacterial communities in sediments, which may further threaten the stability of DOM cycling.
As shown in Fig. S6, significant positive correlations (P < 0.05) among dominant families were 27 in group S1, which were 31 and 35 in groups S2 and S3, respectively. Significant negative correlations (P < 0.05) were 19, 35 and 25 in groups S1, S2 and S3, respectively. Increase of facilitative interactions were observed along the salinity gradient, however, numbers of competitive interactions did not show a linear change under increase of salinity. It was interesting that highest negative:positive correlation was observed in group S2, the value of negative:positive correlation in group S1 was similar to that in group S3, indicating that bacterial communities was more stable under salinity of 3,600 mg/L, higher or lower salinity probably both destabilized bacterial communities in sediments of Shahu Lake. Fluctuations of salinity also affected stability of bacterial communities linked with DOM transformation. For example, significant positive correlations (P < 0.05) between Hydrogenophilaceae and other bacteria were 6, 6, and 4 in groups S1, S2 and S3, respectively. Significant negative correlations (P < 0.05) were 4, 2, and 0 in groups S1, S2 and S3, respectively. Hydrogenophilaceae was resistant to salinity and raise of salinity increase its relative abundances in this study, however, salinity increase diminished its interrelationships with other bacteria and destabilize its communities, may indirectly threaten bacterial diversity and DOM cycling.
4.5 Water recharge scheme and management measures for brackish-water lakes
Exorbitant salinity and drought-rewetting process resulted in less protein-like substances and more humic-like components in pore water, thus reducing pool of bioavailable DOM, destabilizing bacterial communities in brackish-water lake, which was harmful to the utilization of carbon or nutrients and may deteriorate water quality. It was worth noting that excessive low salinity is also not conducive to the stability of bacterial communities in brackish-water lakes. Water recharge is essential measure to maintain water balance of inland lakes, take Shahu Lake as an example, annal volumes of water recharge should exceed 22.09 million m3 based on average evaporation and precipitation, but exorbitant water recharge also destabilizes the bacterial system. To maintain the stability of salinity and water level, continuous recharge of water with lower salinity to brackish-water lakes were appropriate instead of intermittent recharge. Important problems are that it is difficult to determine volumes of water recharge, and water recharge only alleviate raise of salinity concentrations in the short term, but does increase salinity loading. Thus, it is necessary to monitor salinity, dynamically change water recharge volumes based on salinity, and strengthen water exchange between brackish-water lakes and other water bodies. In addition, buffer strips construction, saline-alkali soil remediation, and prevention of agricultural return flow draining into lake or water recharge canals were also effective measures to avoid excessive increase of salinity and frequent drought-rewetting process in brackish-water lakes.