Insight into the evolution characteristics on molecular weight of compost dissolved organic matters using high-performance size exclusion chromatography combined with a two-dimensional correlation analysis

The information on molecular weight (MW) characteristics of DOM and relevant evolution behaviors during composting are limited. In this study, DOM extracted from co-composting of chicken manure and rice husks were comprehensively analyzed by using high-performance size exclusion chromatography (HPSEC) combined with a two-dimensional correlation spectroscopy (2D COS) to explore the evolution characteristics of MW of compost DOM. The HPSEC detected at UV of 254 nm and at fluorescence (FL) Ex/Em wavelengths (315/410, 270/455 nm) all showed a gradual increase in both weight-average and number-average MW for DOM, suggesting that the large MW fractions were continuously generated and polymerized during composting. The 2D COS applied on HPSEC-UV and -FL further identified the key active MW chromophoric (i.e., 0.5, 7.2. 9.5, 26.3, 30.7, and 83.9 kDa) and fluorophoric (i.e., 0.55 and 3.5 kDa) molecules that mainly participated in the transformation processes of compost DOM. Moreover, these active MW species were preferentially formed by the order of small to large molecules. A hetero-2D COS analysis disclosed the change sequence in the order of 0.5 and 7.2 kDa chromophores → 3.5 kDa fluorophores, and the 0.55 and 3.5 kDa fluorophores → 26.3 and 83.9 kDa chromophores.


Introduction
Composting is an environmentally friendly treatment technology for organic wastes from livestock manure, agricultural residues, and municipal solid wastes (Chen et al. 2017;Huang et al. 2021). During composting, the strong degradation and humification of organic matters are generally happened in aqueous phase due to intense microbial activity (Che et al. 2021;Niu et al. 2021;Zhu et al. 2020). The dissolved organic matter (DOM), as the most active component, is then widely recognized to be a useful indicator for assessing stability and maturity of compost (Jiang et al. 2019;Xiao et al. 2019). Consequently, the essential information on physicochemical properties and compositional structures of compost DOM is helpful for monitoring the composting process and even for evaluating the quality of compost product.
Compost DOM is a heterogeneous mixture, including organic acids, carbohydrates, proteins, humic substances, and other bio-macromolecules, with a broad molecular weight (MW) range from ~ 10 2 to 10 6 Da (He et al. 2015;Liu et al. 2020). Different DOM fractions generally exhibit different degradation and transformation characteristic during Responsible Editor: Thomas D. Bucheli composting. Many studies focusing on MW fractionated DOM fractions suggest that the low MW fractions are susceptible to degradation by microorganisms, while high MW ones remain stable and can even be formed by the transformation of low MW fractions Liu et al. 2020). Subsequently, the variations of MW fractions can have crucial impacts on physicochemical properties (e.g., molecular size, humification degree), the fate and even environmental effects of compost DOM. Till now, the knowledge on MW distributions of compost DOM and its relevant transformation behaviors during composting are still limited.
High-performance size exclusion chromatography (HPSEC) has been widely used to separate, quantify, and characterize the MW-dependent DOM species (He et al. 2015. The application of UV absorbance and fluorescence (FL) detection techniques followed HPSEC further allow us better understanding of the MW-dependent compositional characteristics of compost DOM in terms of chromophores and fluorophores therein (Castan et al. 2020;Trubetskaya et al. 2020). In addition, a two-dimensional correlation spectroscopy (2D COS) analysis is a powerful technique to reveal the susceptible and preferential sequences of the active fractions and structures of DOM in response to an external perturbation (Guo et al. 2018;Yu et al. 2019). Consequently, the combination of HPSEC-UV and -FL with 2D COS can not only gain deep insight into the MW characteristics of compost DOM but also the relevant evolutionary behaviors on DOM MW species during the composting process.
In this study, a conventional co-composting of chicken manure and rice husks was conducted, and composting timedependent DOM were extracted and analyzed. The objectives of the study were to investigate (1) the variations on MW distributions of DOM during composting using HPSEC-UV and HPSEC-FL analysis and (2) the evolutionary behavior of MW DOM species using a 2D COS analysis applied on HPSEC chromatograms. The results enhance our understanding of the compositional characteristics of DOM and their transformation behaviors during composting.

Composting and sampling
The composting was carried out in an industrial-scale organic fertilizer plant in Fuyang, Anhui Province, China. The raw composting materials including chicken manure and rice husks were both obtained from a breeding company in Fuyang. The physico-chemical characteristics of the raw materials are presented in Table S1 of the Supporting Information. In the composting process, about 8 tons of chicken manure and 2 tons of rice husks were well mixed and packed into the fermentation compartment using a windrow of 7.0 × 2.5 × 1.1 m (length × width × height). The C/N ratios and moisture content of the initial compost were around 25 and 62%, respectively. The entire composting process lasted 30 days, and the windrow was turned over every 3 days using a fertilizer turner for aeration. The homogeneous and representative compost samples were collected after 0, 3, 6, 12, 18, 24, and 30 days using multipoint sampling at a depth of 50-100 cm of the windrow. The collected samples were manually mixed and stored at − 20 °C prior to analysis.

Extraction of DOM
Composting samples were freeze-dried, ground, and then passed through a 100-mesh sieve. The dried compost samples were immersed in ultrapure water at a solid to water ratio of 1:10 (w/v) and shaken at 250 rpm for 24 h at 25 °C in a thermostatic oscillator. The suspensions were centrifuged at 8000 rpm for 30 min at room temperature, and passed through a 0.45-μm filtration membrane to obtain DOM. In this study, the DOM samples were prepared and analyzed in four replicates for each compost sample unless otherwise specified. The dissolved organic carbon (DOC) concentrations of DOM were measured using a total organic carbon analyzer (TOC-VCPN, Shimadzu, Japan).

Spectral analysis
Prior to UV-vis and fluorescence measurements, the DOC contents of DOM samples were diluted below 10 mg L −1 to avoid inner filter effects (Chen et al. 2017;Li et al. 2019). The UV-vis absorption spectra of DOM were determined using a UV-2600 spectrophotometer (Shimadzu, Japan) with a wavelength range of 200-700 nm. The EEM spectra were recorded by an F-4600 fluorescence spectrophotometer (Hitachi, Japan). The scanning wavelengths of Ex and Em were set to 200-400 nm and 290-520 nm with an increment of 5 nm, respectively. The scanning speed was fixed at 12,000 nm min −1 . The spectra parameters, including specific UV absorbance (SUVA 254 ), absorptive ratio (E 2 /E 3 ), fluorescence index (FI), humification index (HIX), and biological index (BIX) were calculated to assess the chemical properties and sources of the DOM (Guo et al. 2018;Musadji et al. 2020;Yuan et al. 2018a). The details were provided in Text S1 of the Supporting Information.
EEM coupled with parallel factor (PARAFAC) analysis was an effective tool to quantitatively and qualitatively characterize the fluorescence components in compost DOM (Xiao et al. 2019;Yu et al. 2019). In this study, EEM-PARAFAC analysis was performed for assessing the transformation of fluorescence components within DOM during composting, which were conducted by using MATLAB with the DOMFluor toolbox (Stedmon and Bro 2008). The modeling was constrained by non-negative values, and the identified components were validated by a residual analysis, split-half analysis, and visual inspection. The maximum fluorescence intensity (F max ) values were used to indicate the concentration of the corresponding components (Xiao et al. 2019;Yan et al. 2021). It should be noted that the F max could reflect a guide rather than real concentration of the fluorescence components. Subsequently, the relative content (%) of each component was calculated by dividing the F max value of a certain component by the sum of the total fluorescent components (Huang et al. 2021;Xiao et al. 2019).

The HPSEC method
The MW properties, including the size and distribution of compost DOM, were analyzed by HPSEC. Before the HPSEC analysis, 1.5 mL of DOM samples were prefiltered through 0.22 μm PTFE filters. For HPSEC analysis, an aqueous gel filtration column (Polysep-GFC-P 3000, Phenomenex) and a guard column (Polysep-GFC-P, Phenomenex) were used. A water/methanol (9:1, v/v) with 25 mM ammonium acetate was used as mobile phase; the sample injection volume was 100 μL with a flow rate of 1 mL min-1. The MW separated DOM was detected using a refractive index detector (RID, Shimadzu) connected to a diode array detector (SPD-6A, Shimadzu) and fluorescence detector (RF-10A, Shimadzu), respectively. The apparent MW of the DOM was calibrated with polyethylene glycol (PEG) standards (Sigma Aldrich, USA) of 238, 601, 1020, 3450, 4080, 11,100, 17,900, and 41,300 Da. The PEG had been widely used to calibrate the apparent MW of DOM Lv et al. 2013;Zhang et al. 2018). The void volume and exclusion volume were tested with polystyrene sulfonate (MW 210 kDa) and acetone (MW 58 Da), respectively. The Ex/Em wavelength pairs of the fluorescence detector were selected according to the positions of the fluorescent components (C1 and C2) derived from the PARAFAC analysis. The calibration curves for chromophores and fluorophores are shown in Fig. S1, which shows good linear correlations between the retention time and logMW (R 2 = 0.9987).
The weight-averaged MW (Mw), number-averaged MW (Mn), and polydispersivity (ρ) were calculated based on SEC-absorption/fluorescence chromatograms. They were determined according to the following equations (Ignatev and Tuhkanen 2019;Song et al. 2010): where h i and MW i are the height of the chromatogram and the apparent MW of DOM samples corresponding to the ith retention time, respectively. The Mw and Mn of compostderived DOM were estimated by HPSEC and were dependent on MW standards, which were calculated based on PEG calibration in this study.

The 2D COS analysis
To address the molecular size variations of DOM during composting, 2D COS of HPSEC chromatograms detected at UV254 (HPSEC-UV) and fluorescence Ex/Em pair wavelengths of 270/455 nm (HPSEC-FL), respectively, were carried out using composting time as the external perturbation. The 2D COS analysis were performed by using the "2D shige" software (Kwansei-Gakuin University, Japan) (Noda and Ozaki 2004), though which two 2D COS maps, including synchronous and asynchronous maps, were generated. The 2D COS maps were furtherly interpreted based on so-called Noda's rule, which had been widely used in many previous studies Guo et al. 2018;Noda 2016). The detailed interpretation of 2D COS maps were also given in Text 2 of the Supporting Information.

Statistical analysis
Statistical analysis was conducted using the IBM SPSS Statistics 26.0. One-way analysis variance (ANOVA) was used to assess the significant differences of spectra parameters, Mw and Mn of different DOM during composting, using Tukey's honest significance test (HSD). The significant differences were set at p < 0.05 in this study.

General characteristics of compost DOM
In this study, the variation of the DOC content of DOM was shown in Fig. S2. It was clear that the DOC continuously decreased from day 1 to day 24, followed by a slight increase until to day 30. The DOC exhibited a final reduction of 36% (Figs. S2). The results indicated that the DOM contained amounts of labile DOC that were easily degraded due to intense microbial activity, which was a typical feature of DOM during composting (Abid et al. 2020;Che et al. 2021;Wang et al. 2021b). The UV-vis spectra normalized by the DOC content of compost DOM as a function of time are shown in Fig. 1. The spectra generally showed a decrease in absorbance with increasing wavelength, which matched (3) =

Mw Mn
the typical absorption characteristics of compost DOM reported in previous studies (Abid et al. 2020;Wang et al. 2021a;Yang et al. 2020). Besides, the spectra were gradually increased with the composting time, suggesting the aromaticity of the DOM was enhanced during composting ( Fig. 1). The results were well consistent with the observations for other composting previously reported (Che et al. 2021;Wang et al. 2021a;Yang et al. 2020).
To investigate the changes of chemical composition and properties of DOM during composting, some useful spectra parameters (i.e., SUVA 254 and E 2 /E 3 ) were measured. As shown in Table 1, the SUVA 254 of DOM increased substantially over the entire experiment, and was ~ 4.6 times higher than the initial ones. The E 2 /E 3 values followed the opposite trend, with a continuous decline from 3.93 on day 1 to 2.61 on day 30. It has been previously shown that SUVA 254 and E 2 /E 3 are positively and negatively related to the aromaticity and MW of natural DOM, respectively (Wang et al. 2021a;Yang et al. 2020). This suggested that during the composting process, the DOM was likely to be humified, with the degradation of labile hydrophilic compounds, as well as the polycondensation of aromatic species and an enlargement of molecular size (Abid et al. 2020;He et al. 2011;Wang et al. 2021a;Yang et al. 2020).

Fluorescence spectra
The EEM spectra The changes in the EEM spectra of the DOM during composting are shown in Fig. S3. In general, five peaks were observed at Ex/Em of 270/305, 215/305, 220/375, 280/405, and 310/405 nm, respectively. Peak B at 270/305 was associated to tyrosine-like materials (Coble 1996;Mostofa et al. 2013), and was also linked to microbial byproduct-like compounds (Chen et al. 2003). Peak T at 215/305 nm was likely tryptophan-like substances in the UV region (Mostofa et al. 2013) and aromatic protein-like compounds (Chen et al. 2003). Peak A at 220/375 nm was mainly associated with humic-like substances in the UV region and was more likely fulvic-like compounds (Chen et al. 2003;Guo et al. 2018), and peaks C at 280/405 was mainly concerned with visible humic-like substances as well (Chen et al. 2003;Coble 1996;Mostofa et al. 2013). Peak M at 310/405 nm was related to microbial humic-like substances (Lv et al. 2013;Mostofa et al. 2013). This indicated that the compost DOM were mainly comprised of protein-like and humic-like species. It was noted that the peaks B and T obviously showed in EEM of day 1 were declined with composting time and had fully disappeared on day 18 (Fig. S3). This finding implied that the labile organic matter was more likely associated with the protein-like substances that were substantially biodegraded during composting. Moreover, the peaks A and M were gradually transformed to be the dominant peaks for compost DOM (Fig. S3), especial on day 30, suggesting the predominance of humic-like substances. This finding implied the strong degree of humification or maturation for DOM when the composting finished. These results were in good agreement with those for compost DOM reported in previous studies (He et al. 2015;Lv et al. 2013;Wang et al. 2021a;Yang et al. 2020 As seen in Table 1, FI and BIX were generally in the range of 1.60-1.84 and 0.75-0.97, respectively. The FI > 1.5 and BIX > 0.6 both suggested strong contributions of microbial sources to compost DOM Yuan et al. 2018a). In addition, both FI and BIX were gradually decreased with composting proceed, implying the weakening of microbial activity and also enhancement of aromatization and humification during composting (Musadji et al. 2020). The HIX value is a useful indicator for characterizing the degree of maturation and humification of the compost DOM (Guo et al. 2018;Yan et al. 2021;Yu et al. 2019;Yuan et al. 2018b). The HIX increased sharply over the composting time, with values increasing from 1.20 on day 1 to 11.45 on day 30. This finding confirmed again that the transformation of DOM during composting resulted in a high degree of humification.

The PARAFAC modeling
The PARAFAC modeling was conducted to characterize and distinguish the fluorescence components of compost DOM. As shown in Fig. 2a, four independent components were identified by the PARAFAC analysis. C1 exhibited a maximum peak at an Ex/Em of 230 (315)/410 nm, and was generally attributed to a fulvic-like substances Yan et al. 2021). C2 had a maximum peak at an Ex/ Em of 270 (365)/455 nm, which is a typical characteristic of humic-like fluorophores Lee et al. 2020). It was noted that the peaks of C2 was obviously redshifted compared to that of C1, which suggested that more condensed structures and/or larger sized aromatic molecules were abundant in C2 (Fellman et al. 2010;Huang et al. 2018;Ignatev and Tuhkanen 2019). C3 was located at an Ex/Em of 215(270)/305 nm, which could be assigned to typical tyrosine-like substances Yu et al. 2019). C4 had a peak at Ex/Em of 280(220)/355 nm, which was a typical tryptophan-like and humic-like fluorescence region Yan et al. 2021), and was most likely the result of mixed microbial by-products and humic-like fluorophores (Wei et al. 2016;Yu et al. 2019).
The relative distributions of C1-C4 of DOM over the composting time are shown in Fig. 2b. It was clear that the proportion of C3 decreased (52 to 0%) over the whole duration of the composting process, but there were increases in the proportion of C1 (20 to 47%) and C2 (10 to 38%). This demonstrated that the degradation of tyrosine-like substances was accompanied by the formation or accumulation of humic materials, which confirmed the transformation of protein-like to humic-like substances due to intense microbial activities during composting (Chen et al. 2017;Guo et al. 2018;He et al. 2011;Lv et al. 2013;Yu et al. 2019). As revealed in many previous studies, the enhancement of humic-like materials in compost DOM was strongly associated with the degree of stabilization of the compost Huang et al. 2021;Lv et al. 2013;Zhang et al. 2016). Therefore, the result obtained here indicated that the final compost DOM had a strong degree of stability and maturation.

The MW properties of chromophores within DOM
The multi-wavelength HPSEC chromatograms of DOM were investigated to characterize the changes in their composition and MW profiles during composting (Fig. 3). Based on the peak locations in the MW ranges, three MW fractions with distinct apparent MW regions were identified (100-1000 Da, 1000-50,000 Da, and 50,000-200,000 Da), which were termed as Fractions I-III (Fig. 3). It was obvious that the HPSEC profiles were dominated by Fractions II for all DOM, and Fractions III were more enhanced as composting proceed. This result indicated that more condensed chromophores with a larger MW (> 1000 Da) were formed with composting. This directly confirmed that the DOM were changed to become more mature and humified during composting, which agreed well with the results obtained from spectra parameters (such as E 2 /E 3 , SUVA 254 , and HIX)  Zhu et al. 2020). Although the > 1000 Da fractions were gradually transformed to be the dominant chromophores within DOM during composting, Fractions I were also important chromophores within the DOM. This might be attributed to the intermolecular interactions between small and large MW fractions that incorporated low MW molecules into the high MW fractions. This preserved the low MW fractions and prevented them from being microbially consumed and degraded (Ignatev and Tuhkanen 2019;Jokubauskaite et al. 2015).
In this study, the HPSEC-UV chromatograms recorded at an absorption wavelength of 254 nm for DOM at different composting times were specifically analyzed and compared (Fig. 4). The chromatograms generally displayed three apparent peaks with two sharp peaks and a broad shoulder for all DOM samples. The two sharp peaks were centered at ~ 0.5 and ~ 85.1 kDa (Fig. 4), respectively, and the broad shoulder were distributed between 1.0 to 50.0 Da (Fig. 4). It should be noted that the primary peak within the shoulder was shifted from small MW (~ 3.5 kDa) on day 1 to a larger one (17.2 kDa) on day 30 during composting. This implied that the predominant MW fractions within DOM were transformed to become larger ones during composting. This was further evidence confirming the enlargement of the MW of DOM during composting, which had been indirectly confirmed in previous studies (Chen et al. 2017;Guo et al. 2018;Jiang et al. 2019;Zhu et al. 2020). Table 2 shows the Mw, Mn, and ρ of compost DOM based on the chromatograms within the 0.1-200.0 kDa region recorded at UV254. The Mw and Mn of DOM gradually increased during the composting. In general, the Mw of DOM increased from 10.7 kDa on day 1 to 21.3 kDa on  day 30, and the Mn of DOM continuously increased from 0.7 kDa on day 1 to 1.5 kDa on day 30. This suggested that the MW of DOM became larger during composting, which matched the results discussed above. This reflected the accumulation of small aromatic molecules into DOM during composting (Chen et al. 2017;Guo et al. 2018;Jiang et al. 2019;Liu et al. 2020;Niu et al. 2021).

The MW properties of fluorophores within DOM
As discussed in Sect. 3.3, the fulvic-like (C1) and humiclike (C2) substances were the dominant fluorophores within compost DOM, especially for the final compost products. In this study, the MW characteristics of C1 and C2 were analyzed using HPSEC, with the detection of fluorescence Ex/Em based on the primary peaks identified by EEM-PAR-AFAC. The resulting HPSEC-FL chromatograms detected at Ex/Em of 315/410 nm (C1) and 270/455 nm (C2) for DOM at different composting times are shown in Fig. 5.

The 2DCOS analysis on HPSEC-UV chromatograms
To visibly characterize the evolutionary behavior of chromophores within DOM with respect to molecular size during composting, a 2D COS analysis was applied to the series of HPSEC-UV chromatograms, with composting time as the external perturbation. The resulting synchronous and asynchronous maps were shown in Fig. 6a, b. In the synchronous maps, three positive autopeaks were located on the diagonal line (i.e., 0.5, 20.4, and 83.9 kDa), with peak intensities decreasing in the order of 20.4 > 83.9 > 0.5 kDa (Fig. 6a). According to the Noda's rule, the higher the intensity of auto-peaks, the stronger the susceptibility of variations for corresponding MW species Guo et al. 2018;Lee et al. 2020). The spectra and HPSEC results had revealed that the DOM evolved to be stronger oxidation and humification as composting proceeding. Thus, the higher intensity of auto-peak at 20.4 kDa suggested that the ~ 20.4 kDa MW fractions were more susceptible to oxidation during the composting process than other MW species. In addition, all cross-peaks in the synchronous maps were positive, suggesting that the three dominant MW fractions within DOM were changed in the same direction (Guo et al. 2018;Lee et al. 2020). According to the HPSEC and UV-vis spectra analysis, it was speculated that three MW fractions (i.e., 0.5, 20.4, and 83.9 kDa) were more likely formed over the composting time, with the dominant formation of a ~ 20.4-kDa fraction. It agreed well with the MW distribution variations of absorption intensity observed in the HPSEC-UV chromatogram (Fig. 4). The asynchronous map, on the other hand, disclosed the sequential or successive changes of the MW absorption intensities in response to composting time. As shown in Fig. 6b, four negative cross-peaks were observed at 26.3/0.5, 83.4/0.5, 30.7/7.2, and 83.4/9.5 kDa below the diagonal of the asynchronous map. This finding suggested that DOM possessed many active sites relevant to different MW species, which participated in the humification and maturation of DOM during composting. All of the cross-peaks had the opposite signs in the synchronous and asynchronous maps. According to Noda's rule (Guo et al. 2018;Lee et al. 2020), the sequence of the MW variation in response to times followed the order of 0.5 → 7.2 → 9.5 → 26.3 → 30.7 → 83.9 kDa (" → " means "prior to", unless other specified). Interestingly, the sequential order exactly followed the trend of the MW size. This suggested that the low MW fractions were preferentially bio-oxidized, resulting in the formation of intermediate sized MW fractions compared to the high MW fractions during composting. This sequential order was considered to be reasonable because the lower MW species were more susceptible to biodegradation, and the larger ones were more likely to be condensed due to humification and aromatization during the composting process.

The 2DCOS analysis on HPSEC-FL chromatograms
A 2D-COS analysis of the HPSEC-FL chromatograms of C1 and C2 was conducted. It was found that the 2D COS spectra of C1 and C2 were similar, and the latter one was further analyzed in this study due to its rapid formation during composting (Fig. 4b). The corresponding synchronous and asynchronous maps are displayed in Fig. 6c, d. Two obvious auto-peaks located at 0.55 and 3.5 kDa were observed in the synchronous maps (Fig. 6c), with both peaks having similar intensities. In addition, one cross-peak at 3.5/0.55 kDa with a positive sign was also found. These findings suggested that both of the MW fluorophores (0.55 and 3.5 kDa) participated strongly in the composting process in the same direction.
In an asynchronous map, only one apparent cross-peak located at 3.5/0.55 kDa with a negative sign was observed (Fig. 6d). According to Noda's rule (Guo et al. 2018;Lee et al. 2020), it was concluded that the low MW fluorophores (0.55 kDa) were the first to be changed during composting, followed by the high MW fluorophores (3.5 kDa). The sequential order of this reaction coincided with that of the different MW chromophores discussed in the previous section, and the findings reported for the sequence of MW species within DOM from municipal solid waste composting in a previous study (He et al. 2015). These findings suggest that low MW species, whether for chromophores or fluorophores, participated in complex reactions (i.e., degradation, polymerization, and aromatization) more rapidly than large ones.

The HPSEC-UV/HPSEC-FL hetero-2D COS analysis
The covariations of different MW chromophores and fluorophores were studied by a HPSEC-UV/HPSEC-FL hetero-2D COS analysis (Fig. 7). In the hetero synchronous maps, six cross-peaks were located at MWs of 0.5, 20.4, and 83.9 kDa for chromophores with corresponding fluorophore MWs of 0.55 and 3.5 kDa (Fig. 7a). All the cross-peaks displayed positive signs, indicating that all of the MW chromophores and fluorophores were changed in the same direction (Guo et al. 2018;Lee et al. 2020), and were most likely generated during the composting process. These results also revealed that humic-like fluorophores with MWs of 0.55 and 3.5 kDa might be the key basic units of the chromophores with 0.5, 20.4, and 83.9 kDa within compost DOM.
Five obvious cross-peaks were found in the asynchronous map, which were located at the chromophores/fluorophore MWs of 0.5/3.5 kDa, 7.2/3.5 kDa, 26.3/0.55 kDa, 83.9/0.55 kDa, and 83.9/3.5 kDa, respectively (Fig. 7b). The first two peaks had positive signs, while the latter three were negative. These findings implied that the changes of the 0.5 and 7.2 kDa chromophores occurred before the changes of the 3.5 kDa fluorophores. However, the variations of the 0.55 and 3.5 kDa fluorophore occurred before the variations of the 83.9 kDa chromophores. The 0.55 kDa fluorophores were also changed before the 26.3 kDa chromophores. The results allowed us to determine the correlations between different active sites in terms of MW for chromophores and humic-like fluorophores. Fig. 7 Hetero-synchronous (a) and hetero-asynchronous (b) maps generated from the HPSEC chromatograms detected at an UV254 and those at a fluorescence Ex/Em of 270/455 nm

Conclusions
In this study, the transformation of DOM in terms of physicochemical properties and MW species during composting was comprehensively investigated. Spectroscopic analysis revealed that the aromaticity and humification of DOM almost exhibited gradual increase with the composting proceeding. HPSEC analysis suggested that the MW distributions of chromophores (UV254) and PARAFAC-derived humic fluorophores (C1, C2) within compost DOM covered continuum MWs ranging from 0.1 to 20.0 kDa. Besides, the molecular size (i.e., Mw and Mn) of compost DOM, especial for chromophores therein, changed to be much larger on day 30 than on day 0. These findings suggested that the composting could enhanced the maturation of DOM accompanied with enlargement of their MW. During composting, distinct active MW chromophoric and fluorophoric molecules in the DOM were involved in transformation. 2D COS analysis indicated that the polymerization of chromophores and fluorophores mainly occurred gradually from small to high MW molecules. The hetero-2D COS analysis further correlated the changes in the active MW of molecules within the chromophores to those within the fluorophores of DOM during composting. Both 2D COS and hetero-2D COS offered insight into the transformation behaviors of DOM in terms of MW species during composting.
Author contribution Xufang Yu: writing-original draft, formal analysis, methodology. Ao Cheng: investigation, software, data curation. Dan Chen: investigation, data curation. Ting Li: investigation. Xingjun Fan: methodology, supervision, funding acquisition, writing-review and editing. Xiang Wang: data curation. Wenchao Ji: software, formal analysis. Jianfei Wang: methodology, writing-review and editing. Lantian Ren: writing-review and editing. Data availability All data generated or analyzed during this study in this article are available from the corresponding author on reasonable request.

Declarations
Ethics approval and consent to participate Not applicable.

Competing interests
The authors declare no competing interests.