4.2 Connectivity Regimes: Scaling from Site to System Scales
Aggregating target site specific dynamics to river-floodplain system scale behavior is critical for understanding how river connectivity in floodplains impacts broader landscapes processes. While our analyses here are limited due to our relatively small sample size of sites, aggregation of site behavior did reveal important distinctions between mean system behavior and spatially distributed behavior. As conceptualized in the flood and flow pulse concepts (Junk et al., 1989; Tockner et al., 2000), mean connectivity across the river-floodplain system rose as streamflow increased (Figure 7c). Thus, while threshold-like behavior was observed at many individual sites, mean system behavior followed a continuous gradient because the connectivity thresholds were highly variable among sites. However, it is also clear that the mean is a poor descriptor of the spatially aggregated behavior, particularly at lower river flows when connectivity strength values across the floodplain had a bimodal distribution with some sites maintaining relatively high connectivity while the majority were disconnected. This has important implications for scaling many spatially distributed biogeochemical and ecologic processes impacted by connectivity such as carbon production and storage, nutrient retention, and methane fluxes (Lynch et al., 2019; Roley et al., 2012; Samaritani et al., 2011; Sutfin et al., 2016).
Aggregating site-specific behavior also demonstrated that river-floodplain system-wide variance in connectivity was maximized at intermediate river flows. The low variance in connectivity observed at high river flows (Figure 7c) is consistent with the flood homogenization theory that physical and chemical states across floodplains are more similar at high flows (Thomaz et al., 2007). Our results also support the idea that physio-chemical condition at individual sites in floodplains are most different from the river at the lowest flows (low σ values) due to isolation. However, our findings diverge from the homogenization theory in that peak variability in connectivity dynamics was observed at intermediate flows rather than low flows as the theory suggests. Thus, while individual sites might be most different from each other at lowest flows due to isolation, the distribution of connectivity dynamics across the floodplain was most variable when river stage was intermediate and some sites were isolated while others remained strongly or moderately connected to the source.
4.3 Inter-annual Variability in Connectivity at Site-Specific and River-Floodplain System Scales
In watersheds characterized by a single large snowmelt event, hydrologic variability is often driven by inter-annual variation in snowpack accumulation and melting that regulates the timing and magnitude of streamflow (Hammond et al., 2018). In floodplains within these watersheds, whether connectivity regimes are sensitive to this inter-annual variability will depend on the interactions between streamflow hydrographs and the physical structure of river-floodplain connections and corresponding thresholds. As future climate predictions indicate lower snowpack, earlier snowmelt and drier late season conditions (Barnett et al., 2005; Stewart et al., 2005), assessing the degree of sensitivity is important for understanding how future hydro-climatic regimes may change connectivity in river-floodplain systems. Our connectivity predictions for five years at intermittent sites highlight that site-specific and river-floodplain system scale connectivity regimes are sensitive to streamflow variability with substantial year to year shifts in the duration of high and low connectivity. However, within a given river-floodplain system, there will be spatial variation in sensitivity that will be driven by the river-floodplain physical structure and corresponding stage thresholds, but also by the manner of changes in streamflow hydrographs. This can observed in our dataset by comparing floodplain connectivity in two low flow years: 2018 and 2020. In 2018, we observed the lowest peak flows at Inflow in the five year dataset but 2018 had a longer duration of medium to high flows than was observed in 2020 (Figures 8 & S5). As a result, a majority of sites remained highly connected for longer in 2018 than 2020, while durations of intermediate connectivity were highest in 2020. As such, efforts to understand how climate change will alter floodplain function in snowmelt watersheds will need to consider both changes to flow magnitudes and to flow durations generated by changing climatic conditions.