3.2 Seasonal Dynamics in Aqueous Geochemistry and Microbiome Membership
Using a principal component analysis, we identified which geochemical indicators were most representative of connectivity and generated a reduced dimensional space with uncorrelated components. The primary principal component (PC1) corresponded to bulk ionic strength and explained 62.2% of the variance in water chemistry and the secondary principal component (PC2) explained 17.8% of variance and was strongly driven by SO42- concentrations (Figure 1c, Table S2). All ion concentrations were negatively related to PC1 with Na+, Ca2+, Cl-, Mg2+, and K+having moderate loadings (between -0.39 to -0.48). SO42- had a strong positive loading on PC2 (0.84) while Ca2+and Mg2 had moderate negative loadings (-0.36 and -0.30, respectively).
Seasonal geochemical patterns at Inflow followed a snowmelt dilution pattern where geochemical ion concentrations (e.g., Na+) were lowest during peak flows (Figure 1, S1 & S2). This geochemical pattern propagated strongly to sites with surface connections to the river during high flows resulting in high geochemical connectivity strength (σg) (Figures 2 & 3). As stage declined, geochemical composition diverged between Inflow and most floodplain target sites, resulting in lower σg values (Figure 2 & 3). Floodplain target sites crossed the 0.5 σg value across a wide range of Inflow stages (Igeo) from 351 to 650 mm, demonstrating substantial heterogeneity in connectivity dynamics across the river-floodplain system. Connectivity strength at Main-Mid and Outflow never declined below 0.5 while Pond-Iso had distinct geochemistry from Inflow throughout the season resulting in low σg across the study period that never exceeded 0.5. A sharp decline in σg at Pond-Iso was observed at low Inflow stages during the period that the pond was going dry, which may be the result of evapo-concentration (Figure 3),
Using a principal coordinates analysis (PCOA), we explored seasonal dynamics in microbiome membership and examined the potential to utilize microbiome membership as an indicator of connectivity. The PCOA of microbiome membership identified a major axis PCOA-1 that explained 32.5% of the variance in microbiome membership and a secondary axis that explained 15.9% of the variance (Figure 1c). All additional axes explained less than 10% of the variance. Microbial membership at Inflow was relatively stable between the rising limb and falling limb, with more observable shifts in membership at Inflow observed during the recession period (Figure 1c, S3 & S4). During peak flows, sites with structural surface connections to the river had microbiomes similar to Inflow, resulting in high microbial connectivity strength (σm) values (Figure 2 & 3). Major channel sites maintained their similarity to Inflow for most of the study period with some divergence at the lowest flows later in hydrograph recession. At side channels and connected pond sites, microbiome membership started diverging from the seasonal pattern at Inflow in either the falling limb or recession hydro-period, resulting in lower σm values later in the season (Figure 2 & 3). Floodplain target sites crossed the 0.5 σm value at Inflow stages (Imicro) ranging from 308 to 577 mm. Like with the geochemical metric, the model fit for σm at Main-Mid and Outflow never declined below 0.5 while Pond-Iso had distinct microbial membership from Inflow throughout the season with low σm across the study period that never exceeded 0.5.