Planktonic microbiomes comprise both free-living organisms and those attached to particles, which is typical and visible in water columns of estuaries. The planktonic microbiomes in estuaries are one of the most active microbial communities [1] and they contribute primarily to the most productive environments on the planet [2]. For example, estuaries have a high CO2 flux (~ 0.25 Pg C y− 1) between water and air, which is largely supported by the process of microbial decomposition and carbon fixation [3]. These microbiomes drive the estuarine biogeochemical processes of the elements for life [4]. They are powering the cycles of nutrients (e.g., carbon, nitrogen, phosphorus, and sulfur), with important impact on the composition of greenhouse gases in the atmosphere, the formation of algal blooms, and the integrity of estuarine ecosystems [5–7]. In fact, microbiomes are the foundation for the estuarine food chains and food webs [8–10], and their composition/distribution are important in the balance and stability of the entire estuary ecosystem.
Estuaries harbor a tremendous diversity of microbes. Due to the strong temporal and spatial gradients and surrounding land uses, the composition and distribution of the estuarine microbiomes are largely affected by human activities and climate/environmental changes [11–13]. Many studies have shown that spatiotemporal environmental variations enrich certain microbial taxa to dominate in estuaries, including the Chesapeake Bay [14], Delaware Bay [15], Sacramento-San Joaquin River Delta [16], Pearl Estuary [17], and Columbia River estuary [18]. In response to strong physical, chemical and biological gradients, estuarine microbiomes exhibit pronounced fluctuations in production, biomass [19, 20], and community composition [11, 21, 22]. It has also been shown that different microbial groups respond differently to spatiotemporal variations [11, 23]. These earlier studies provide important insights into the impacts of spatiotemporal variations and other disturbances (such as anthropogenic pressures) on estuarine microbial community structure and dynamics. More interestingly, the spatial and temporal succession patterns are repeatable and predictable on an annual base [24–28], suggesting the microbial population dynamics are closely interrelating to their ambient aquatic environments. Annual selection pressure driven by environmental forcing has, through the induction of recurrent patterns in resource availability, predator–prey dynamics and microbial interactions, allowed for the assemblage of largely stable and resilient microbial communities [28]. Annually reoccurring patterns indicate that different microbiomes indeed have distinct niches with limited redundancy, otherwise different combinations of microbiomes would appear under the same conditions and prediction can be difficult [26]. These annually reoccurring assemblages and structure of microbiomes suggest their potential ecosystem functions (e.g., autotrophic or heterotrophic microbial production) are specialized and annually repeatable as well [24].
Functional traits of microbiomes are products of multiple populations within communities rather than those of a single population [29], and therefore interactions among microbial taxa are critical to maintain ecosystem integrity and microbiome functionality. Different species or populations interact with each other to form complicated networks through various types of interactions, such as predation, competition and mutualism [30, 31]. Theoretical studies showed that communities in which a large proportion of members connected through positive links (i.e. positive correlations) are deemed to be unstable; in such communities, members may respond in tandem to environmental fluctuations, resulting in positive feedback and co-oscillation [31]. In contrast, ecological networks with compartmentalization and presence of negative interactions could increase the stability of networks under disturbances [31–33]. For instance, high proportion of negative links could better balance the asynchronous dynamics and therefore stabilize co-oscillation in communities and promote stability of networks [31]. Further, modeling studies show that increasing strength of a few key interactions within a food web can destabilize trophic cascades, such as the “gatekeepers” [34], as removal of influencers causes a network to fragmentation [35]. In general, microbial interactions (networks) play critical roles in maintaining community states, ecological niches and function distribution in the context of the microbiome [36].
Co-occurrence networks can reveal information on associations within microbiomes and stability of whole communities [31, 36, 37]. It has been increasingly used to infer microbial interactions [30, 38] in soils [39, 40], oceans [41, 42], coastal waters [43], lakes [44, 45], rivers [46], and even in metabolic modeling [47] and genomic surveys [48]. These correlation-based networks show important details of community interaction rules reflecting ecological processes such as cooperation and habitat partitioning and could represent mathematical associations among different microbial groups [30, 37]. Despite co-occurrence networks have been researched in plenty of earlier studies, they rarely focused on the co-exist microbial networks and how these networks respond to spatiotemporal variations in estuarine environments such as the Chesapeake Bay (CB), the biggest estuary in North America with long residence time (up to 9 months) [49]. Taking into account the dynamic environmental gradients in CB such as temporal variations, freshwater runoffs and ocean water intrusion, microbial interactions and their stability/resilience to environmental changes are critical in further understanding the CB ecosystem. As expected, not only the spatiotemporal variations have potentials to reorganize networks of interactions between co-existing estuarine microbial taxa, the characteristics of these networks themselves can also determine the resilience to environmental disturbances (such as agriculture and urban development). Nevertheless, there are still few detailed studies on microbial networks and their responses to environmental changes in typical estuarine environments [50–53], and we need to address this important knowledge gap to deepen our understanding of estuarine microbiome ecology.
Recently, we have characterized the microbial community structures across both temporal and spatial scales in the CB, where planktonic microbiomes were collected across seven sampling stations (along spatial and salinity gradients) in four seasons over three consecutive years [54]. In this study, we constructed co-occurrence networks based on the detailed 16S rRNA gene high-throughput sequences. Significant and strong correlations (including both positive and negative correlations) were included in co-occurrence network analysis [55]. Due to previously described predominant seasonal and spatial variations in prokaryotic microbiome structure, networks from different seasons (temporal) and locations (spatial) were also constructed. In order to test the susceptibility and stability of co-occurrence networks, we assessed responsiveness of network fragmentation to removal of significant nodes (i.e., with highest betweenness centrality). Further, primary environmental drivers for microbial interactions in the estuary were tested with multivariate regression tree analysis (MRT, [56]). Lastly, pathways that may explain how environmental gradients contribute to shaping estuarine microbiomes and their interactions were identified and quantified with piecewise structural equation model (SEM, [57]). Based on all these analyses, quantitative interactions of environmental drivers and estuarine microbiome structures (phylum level) were proposed. This study provides the first snapshot on CB microbiome networks, the stability and resilience, as well as their quantitative responses to environmental gradients, which are critical in improving our knowledge and understanding of the ecology of estuarine microbiomes.