Dissolved organic matter (DOM) plays a crucial role in driving biogeochemical processes and determining water quality in shallow groundwater systems, where DOM could be susceptible to dynamic influences of surface water influx. This study employed fluorescence excitation-emission matrix (EEM) spectroscopy combined with principal component coefficients, parallel factor analysis (PARAFAC), co-occurrence network analysis and structural equation modeling (SEM) to examine changes of DOM fractions from surface water to shallow groundwater in a mesoscale lowland river basin. Combining stable isotope and hydrochemical parameters, except for surface water (SW), two groups of groundwater samples were defined, namely, deeply influenced by surface water (IGW) and groundwater nearly non-influenced by surface water (UGW), which were 50.34 % and 19.39 % recharged by surface water, respectively. According to principal component coefficients, reassembled EEM data of these categories highlighted variations of the tyrosine-like peak in DOM. EEMs coupled with PARAFAC extracted five components (C1-C5), i.e. C1, protein-like substances, C2 and C4, humic-like substances, and C3 and C5, microbial-related substances. The abundance of the protein-like was SW > IGW > UGW, while the order of the humic-like was opposite. The bacterial communities exhibited an obvious cluster across three regions, which hinted their sensitivity to variations in environmental conditions. Based on co-occurrence, SW represented the highest connectivity between bacterial OTUs and DOM fractions, followed by IGW and UGW. SEM revealed that microbial activities increased bioavailability of the humic-like in the SW and IGW, whereas microbial compositions promoted the evolution of humic-like substances in the UGW. Generally, these results could be conducive to discern dissimilarity in DOM fractions across surface water and shallow groundwater, and further trace their interactions in the river watershed.