Marine diatoms (e.g., Skeletonema spp.) are widely distributed in offshore China, especially, S. dohrnii have been found and isolated in China’s Yellow Sea coastal waters, and subsequently preserved in our laboratory. S. dohrnii cells were pre-cultivated and transferred in artificial seawater (ASW) medium. The cultivation was conducted at 25°C with a light intensity of 100 μmol photons m^–2 s^–1 and a 14:10 h light:dark cycle. The cells were grown at least for three generations before the initiation of the experiment.

Epiphytic bacteria was isolated from the degradation growth stage of microalgae by gradient dilution method. The 2216E agar plates were used to isolate epiphytic bacteria, which cultured in transparent conical flasks (500 mL) in a shaking incubator (26°C, 150 rpm). To identify the epiphytic bacteria, the genomic DNA of bacteria was extracted by TIANamp Bacteria DNA Kit (Tiangen-Biotech, Beijing, China). For polymerase chain reaction (PCR) amplification of the 16S rDNA V3 region, a universal bacterial primer was used. The phylogenetic tree (Neighbor-Joining tree, N-J tree) was constructed using the bacteria sequences and closest related sequences from GenBank, and the genetic distances were calculated. Based on the results of analysis, the isolated bacterial strains (CA-35) shared 66.67% sequence identity to the valid species (Bacillus pumilus) from GenBank.

The isolated positive colonies were identified by Gram staining reaction and biochemical tests (including arabinose test, glucose test, maltose test, urease test, Vogues Proskauer test, nitrate reduction test, starch hydrolysis test, aesculin test, 7% NaCl test, and pH 5.7 test).

The strains B. pumilus was cultured in 500 mL transparent conical flasks with pre-combusted (450°C, 5 h). As described in previous studies (Moens et al., 2016), the abundance of B. pumilus cells was measured by Accuri C6 flow cytometer (BD Biosciences, Erembodegem, Belgium). 0.01% SYBR Green I was applied to the sample to stain it for 30 min in the dark at 37°C. As an internal standard, 1 μm fluorescent beads (Polyscience, Warrington, PA, United States) were injected into each sample. Then, the samples were measured at a flow rate of 0.25 μL s^–1 for 1 min.

To avoid any carbon contamination, all glass materials were acid washed, rinsed with ultra-pure water, and precombusted (450°C for 5 h). A total organic carbon analyzer (TOC-3100, Germany) was used to detect the dissolved organic carbon (DOC). All DOC samples were gravity filtered using the GF/F glass fiber filters (0.7 μm pore size, 47 mm diameter, Whatman). GF/F glass fiber filters can be cleaned by high temperature combustion and can filter sufficient sample volumes without clogging, thus reducing potential sources of contamination. The pH variation of culture was measured using pH meter (Lab 850, SCHOTT Instruments).

Skeletonema dohrnii cells were cultivated in ASW medium, and after reaching the degradation growth phase [algal concentration was around (4.07 ± 0.02) × 10⁷ cells L^–1]. Microalgae liquid was filtered through a 0.2 μm polycarbonate membrane (Millipore, United States) to remove the particles. Then, the filtrate was regarded as the DOM fraction placed in 1 L conical flask with pre-combusted (450°C, 5 h).

To assess the effects of warming and acidification on the transformation of algae-derived DOM by epiphytic bacteria. Two different temperature (27 and 31°C) and pCO₂ (400 and 1,000 ppm) conditions were used in this work. The following four treatments were set up: (i) 27°C and 400 ppm (low temperature and low carbon, LL), (ii) 27°C and 1,000 ppm (low temperature and high carbon, LH), (iii) 31°C and 400 ppm (high temperature and low carbon, HL), and (iv) 31°C and 1,000 ppm (high temperature and high carbon, HH). B. pumilus were grown in above different growth conditions, which were gently bubbled with CO₂, and the gas-flow rate (0.5 L/min) was controlled using a Bronkhorst mass flow controller. In addition, the initial abundance of B. pumilus in each treatment group was approximately (8.51 ± 0.02) × 10⁹ cells L^–1. All experiments were carried out in triplicate and under dark conditions.

The fluorescence spectrophotometer (Hitachi F-7100, Tokyo, Japan) was used to take three-dimensional fluorescence spectroscopy measurements. The photomultiplier tube’s voltage was set at 700 volts. Excitation (Ex) from 200 to 400 nm and emission (Em) from 250 to 550 nm were identified in successive scanning of fluorescence spectra. The scanning speed was adjusted at 8,000 nm min-1 and the Ex and Em slits were kept at 5 nm. Instrument adjustments were carried out the procedure recommended by the Hitachi F-7100 instruction manual. To eliminate the majority of the Raman scatter, each samples spectroscopy was blank subtracted using ultra-pure water. After that, Raman calibration was performed based on literature (Lawaetz and Stedmon, 2009), and Rayleigh scatter (1st and 2nd order) effects were removed using the manufacturer correction procedure. A PARAFAC analysis was conducted in Matlab 2018b (Mathworks, United States) with the DOMFluor toolbox (Stedmon and Bro, 2008).

Every component model may be tested using a split-half analysis residual analysis, and loadings to provide correct information about fluorescence components. All fluorescence components were described using water Raman units (RU). Fluorescence intensity arbitrary units (a.u.) were utilized to fluorescence peak intensity. Spectroscopic indices were further derived from the EEMs: fluorescence index (FI) is often used to indicate the origin of the DOM. Biological index (BIX) is a measure the degree of autochthonous pollution. The β/α (a ratio of two known fluorescing components, i.e., freshness index), and humification index (HIX) showed the humification degree (Parlanti et al., 2000; McKnight et al., 2001).

The correlation analysis and DOM associated parameters were obtained using the “corrplot” package in RStudio. To rigorously define the significance, the difference was determined significant at the levels of p < 0.05, p < 0.01, and p < 0.001. The measured value of the data is represented by the mean ± standard deviation (SD).

The abundance of B. pumilus strain growth in different treatment groups is shown in Figure 1. The initial abundance of B. pumilus was around (8.51 ± 0.02) × 10⁹ cells L^–1 in every treatment groups. The B. pumilus abundance changed at 30 days in all conditions (Figure 1A). However, there were several differences for the abundance of these treatments. LH and HH achieved higher abundance of up to (13.52 ± 0.07) × 10⁹ and (13.08 ± 0.05) × 10⁹ cells L^–1, respectively. LL and HL had moderate abundance of (11.48 ± 0.04) × 10⁹ and (11.94 ± 0.08) × 10⁹ cells L^–1, respectively. This suggests that the significant increase of B. pumilus abundance was caused by elevated temperature and pCO₂. The initial DOC concentration was (0.52 ± 0.03) mg L^–1 for the DOM treatment groups and increased at the end of the 30-day period (Figure 1B). The largest DOC increase occurred in the HH group, which rose by (0.82 ± 0.03) mg L^–1 DOC over the 30-day period. There were significant differences between each treatment group. The initial pH value of the sample was 8.07 ± 0.02 (Figure 1C). At the end of the experiment, the pH increased in LL and HL groups, conversely, decreased in LH and HH groups as a result of acidification.

The culture labeled as CA-35 was identified as Bacillus pumilus (accession number SRR17041859) based on nucleotide homology and phylogenetic analysis. It was found that the strain CA-35 was Gram-positive rod. Results of various biochemical characteristics of CA-35 was shown in Table 1. The result shows that arabinose, glucose, 7% NaCl, malonate, starch hydrolysis, aesculin, pH 5.7 were positive at the initial phase of the experiment. The individual treatment groups showed different results under different incubation conditions. By contrast, warming and acidification caused changes in some characteristics such as arabinose, glucose, maltose, Voges-Prosk test.

All DOM samples collected from the four different treatments were modeled and analyzed with EEM-PARAFAC. In this study, all of the components’ spectral properties were compared to those previously reported in PARAFAC components (Table 2). As shown in Figure 2, one individual component (peak T, tryptophan-like) was identified by the PARAFAC analysis at initial stage. All components (protein-, tryptophan- marine humic-, and humic-like components) in this study have been successfully matched in the OpenFluor database. Component 1 (including A1, B1, C1, and D1; Figure 2) has two fluorescence peaks, located at the Ex/Em wavelength pairs of 225/300–400 and 275/300–400 nm, respectively (Coble, 1996; Coble et al., 1998). Similarly, double peaks have been found in other components. Component 2 (including A2, B2, C2, and D2; Figure 2) was distinguished as marine humic-like components (Ex/Em = 235 and 340/400 nm) (Yao et al., 2011). A humic-like component (peak A, Ex/Em = 250 and 325/425 nm) (Stedmon et al., 2003; Yamashita et al., 2011) associated to terrestrial substances was given to Component 3 (including B3, C3, and D3; Figure 2). Component 4 (including D4; Figure 2) corresponded to humic-like fluorescence (peak C, Ex/Em = 260 and 350/458 nm) (Yamashita and Tanoue, 2008).

The fluorescence index was further used as a parameter indicating changes in fluorescence characteristics (Figure 3). The origin of DOM is identified via FI. The three treatment groups had similar values for FI, which had no obvious change under warming and acidifying conditions. The freshness index was also calculated, with the β peak representing newly produced DOM and the α peak representing more decomposed DOM. All of the samples had β/α values between 1.12 and 1.16, whereas LL had maximum values of 1.16. For BIX, the values are almost appropriate (i.e., 1.28–1.32). The HIX is an indicator of how much organic matter has degraded. There are varying degrees of alteration in the degree of humification, especially in HH group.

The fluorescence intensity of the peaks was altered in all treatment groups after 30 days of incubation (Figure 4). Peak B (protein-like) and peak T (protein-like) maintained a high fluorescence intensity (3,000–4,000 a.u.) in the initial phases, with a similar degree of variation for treatment groups. There was minimal fluorescence intensity in the LL group, suggesting that warming and acidification slowed bacterial consumption of protein-like substances. Additionally, peak N, an unknown component, followed a similar trend to peaks B and T, with generally lower values for all treatment groups. For peak A, acidification resulted in a reduction in the fluorescence intensity of the like-humic substances (i.e., LH and HH groups). However, the fluorescence intensity of peak M (like-humic) decreased and the warming slows down the utilization efficiency of peak M (i.e., HL and HH groups). The only difference is that the fluorescence intensity of peak C (like-humic) is generally increased in all groups.

According to Boyce et al. (2010) reported, total phytoplankton abundance is rapidly declining as a result of warming, with diatoms being the most impacted category (Toseland et al., 2013). Under stressed conditions like as warming, acidification, and nutrient scarcity, phytoplankton releases large amounts of DOM. Bacterial abundance was inversely proportional to phytoplankton abundance under these circumstances. Warming has been demonstrated to favor bacterial growing, which might be owing to two factors: (1) the favorable effect of temperature on bacteria, with more bacterial cells at higher temperatures (Apple et al., 2006), and (2) the availability of more enriched DOM for bacteria. Put simply, phytoplankton produces labile DOC, which heterotrophic bacteria can convert to recalcitrant DOC (Stoderegger and Herndl, 1998). Previous investigations have shown that the addition of pCO₂ had considerable impact on overall bacterial abundance (Baragi and Anil, 2016). However, in this study, algae-DOM was employed as the only carbon source and the LH group showed a relative increase in bacterial abundance. This is likely because high pCO₂ only slowed the rate of increase in bacterial abundance, but did not limit/inhibit it. Besides, the increase in B. pumilus abundance might be attributed to the effect of decreased pH, which increases respiration, as a result, the metabolic and energy cost of bacteria (Siu et al., 2014).

In the present study, a fixed amount of S. dohrnii-derived DOM was added at the initial phase. A considerable portion of the bioavailable fraction of DOM was reduced by bacterial respiration and absorption. However, the DOC concentration showed a distinct increase, suggesting that DOM components were transformed by bacteria to produce new DOC (i.e., production was higher than consumption) under warming and acidification. The DOM can be used as a substrate for remineralization of carbon by heterotrophic microorganisms. Microorganisms use enzymes to catalyze DOC into smaller molecules that can be transported to the environment through bacterial cell membranes (Arnosti, 2011). The organic carbon is subsequently absorbed into the biomass or expelled as DOC as metabolic products (Arnosti, 2011). This provides a new perspective for observing the transformation process of diatoms-derived DOM by epiphytic bacteria.

To further understand the characteristics of DOM fluorophores, we built models with components and used a split-half analysis to validate them all. Four fluorescent components, including typically occurring protein- and humic-like components, were detected in S. dohrnii-derived DOM samples (Figure 2). At the initial phase, protein-like components (peak B and T) were detected. For example, peak T is typically connected to autochthonous tryptophan-like compounds, amino acids, and proteins. In aquatic ecosystems, peak T usually represents the proteinaceous material produced by phytoplankton and bacteria. Moreover, the presence of visible components (peak T) could be attributed by energy transfer to tryptophan and quenching by adjacent groups suppressing tyrosine fluorescence (Lakowicz, 2013). Some of the components have previously been related to high molecular weight and aromatic humic material characterized as peaks A, M, and C in the literature (Wünsch et al., 2015).

All treatment groups essentially produced two-peak patter dominated by discrete S. dohrnii-derived DOM fluorophores, indicating that different compounds, such as protein- and humic-like have been described in the algae-DOM. Several studies have shown that protein peaks (e.g., tryptophan-like) are dominant in phytoplankton growth (Stedmon and Markager, 2005; Jørgensen et al., 2011; Liu et al., 2021b). In previous studies (Kinsey et al., 2018; Liu et al., 2021a), DOM produced in phytoplankton showed similar fluorescence patterns to presented here. As reported by Fukuzaki et al. (2014), the DOM spectra of algal cultures have also been studied. Although the DOM fluorophores are different, the DOM patterns produced are generally similar. The current results show that bacteria use and transform DOM to obtain different fluorescent components. Under elevated temperature and acidified conditions, more fluorescent components were identified, indicating that certain physicochemical characteristics of the bacteria were affected (Table 1), further leading to altered DOM utilization by the bacteria and more humic substances being produced.

In the current study, results showed that the fluorescence values of peak C increased on the whole, while peaks B, T, N, A, and M decreased comparably to the initial value. It suggested that the epiphytic bacteria utilize DOM, which leads to the increase of bacterial abundance, and then the DOM is transformed into recalcitrant DOM and gradual accumulation by epiphytic bacteria. It is noteworthy that on the time scale of deep ocean circulation, these humus-like substances have proven to be recalcitrant, whereas only a small fraction of humic-like substances were biodegradable (Catalá et al., 2015). Besides, the generally low fluorescence in the open ocean suggests that the degradation experiments were conducted over a longer period time than our and other studies (Gruber et al., 2006; Fukuzaki et al., 2014). Indeed, a prior study found that a single strains is incapable of entirely degrading high molecular weight DOM compounds, meaning that multiple bacterial species must work together (Horemans et al., 2013). Briefly, the utilization and transformation of DOM by a single bacterium take much longer.

The spectroscopic index provided a clear insight into the process of DOM-related substance changes. The values of FI were all greater than 1.8, demonstrating that bacteria were primarily responsible for the fluorescence component of DOM transformation and production (Lavonen et al., 2015). In aquatic ecosystems, BIX could be used as a measurement of DOM traceability, with larger values indicating more DOM degradation. The significant changes in β/α and BIX values over a short period of time indicated that endogenous carbon products are most likely produced through bacterial processing of DOM. The HIX indicates the degree of degradation of organic matter, with higher values characteristic of higher molecular weight, aromatic compounds. In other words, the humic concentration of DOM is roughly proportional to HIX (Huguet et al., 2009).

Elevated temperature and acidification caused a general increase in FI and β/α values. Interestingly, we found that β/α values decreased in the HH group, which may be due to the fact that the highly decomposed DOM was higher than recently derived DOM. Besides, both warming and acidification led to a decrease in HIX value. From the current results, the effect of acidification was more prominent. However, the BIX value of each group was not significantly different.

The overall DOM concentration increased by 9.23–26.15% in this study due to elevated temperature and acidification. This result indicates that warming promoted the accumulation of microbial-derived DOM and depressed the aromaticity of DOM. Previous studies have demonstrated that microbial-derived aliphatic carbon is the most labile DOM, which is more readily absorbed than plant-derived aromatic carbon (Yang et al., 2016). Changes in bacterial abundance and DOM properties such as pH, fluorophores, and fluorescence intensity could be explain the discrepancy. Acidification is linked to the decomposition of more refractory organic carbon (e.g., lignocelluloses), as compared to warming (Williamson et al., 1999). Specifically, pH increased the aromatic fractions in DOM (peak C), which are recognized as organic matter produced by bacteria. The bacterial abundance in our study was increased from 8.51 × 10⁹ to 13.52 × 10⁹ cells L^–1 under acidification conditions, besides, significant increase in DOC concentration. Thus, it is regarded as enhanced utilization of aromatic DOM by bacteria.

Pearson’s correlation analysis showed significant variability between warming and acidification for DOM-associated parameters. As shown in Figure 5, temperature and DOC concentration had a significant positive correlation (p < 0.001). Besides, FI and HIX were positively correlated with temperature (p < 0.05), respectively. These results indicating that temperature affects microbial activity and production, which was consistent with results from previous studies (Thangaraj and Sun, 2021). The pH and peak A had a significant positive correlation (p < 0.01), and bacteria abundance and FI were significantly negatively correlated with pH (p < 0.05). Moreover, the formation of aromatic DOM or humic-like is thought to be facilitated by acidification-driven microorganisms. This appears to be corroborated by the fact that when the temperature rises, and pH decreases (0.09–0.14 units). Bacterial abundance and microbial activity have both been shown to increase when pH is raised (Han et al., 2022). This is overall consistent with our results.

As discussed above, elevated temperature and acidification showed different impacts on DOM utilization and transform by bacteria. This work has important implications for better understanding how microbes in the ocean transform and sequester DOM. Our findings show that warming and acidification conditions will exacerbate the accumulation of organic matter, particularly phytoplankton-derived aromatic substances. Given the crucial impacts and consequences of bacteria on the use of phytoplankton-DOM under warming and acidification conditions, further investigations regarding carbon transform and release process in natural environments (especially microbial environment) under global climate change deserves serious attention.