Hydrodynamics drive pelagic communities and food web structure in a tidal environment

Hydrodynamic processes can lead to the accumulation and/or dispersal of water column constituents, including sediment, phytoplankton, and particulate detritus. Using a combination of field observations and stable isotope tracing tools, we identified how hydrodynamic processes influenced physical habitat, pelagic communities, and food web structure in a freshwater tidal system. The pelagic habitat of a terminal channel differed spatially, likely aligning with differences in hydrodynamics. Three zones that we classified by exchange with downstream habitat had distinct water quality characteristics, supported different densities of zooplankton and nekton, and exhibited disparate support from benthic and pelagic trophic pathways to pelagic consumers. Hydrodynamically driven zones and their emergent characteristics appeared sensitive to hydrology, as elevated runoff was correlated with a shift in hydrodynamic habitat and organismal distributions. The results of our study highlight the relationship between hydrodynamic processes, biological responses, and climate, and suggest that understanding the physical process can improve understanding of pelagic habitats and communities.

Phytoplankton, detritus, benthic algae, and other carbon sources are important to estuarine consumers (Deegan & Garritt, 1997;Peterson et al., 1985), but the relative value of each can be highly variable within and among estuaries (Chanton & Lewis, 2002;Deegan & Garritt, 1997). Large-scale estuarine hydrodynamics demonstrably affect these productivity pathways and food web structure, particularly with respect to ETMs. For example, in many systems, the abundance of mixotrophic and heterotrophic organisms is elevated in ETMs relative to other habitats (Chikugo River estuary, Islam et al., 2005; Columbia River estuary, Crump & Baross, 1996; Chesapeake Bay, Lee et al., 2012), suggesting variability in the organic carbon sources available to consumers. In the Chikugo River estuary, Suzuki et al. (2012) identified the differential distribution of phytoplankton and plant detritus with respect to the ETM, whereby phytoplankton was more abundant up-and downstream of the ETM while plant detritus was more abundant in the ETM. This matches the findings of Suzuki et al. (2009), who noted elevated abundance of the mysid Hyperacanthomysis longirostris in the Chikugo River estuary ETM, a species tied to detrital food sources (Schroeter et al., 2015).
Similarly, the copepod Acartia tonsa consumed detritus within the Río de la Plata ETM and phytoplankton outside of it (Derisio et al., 2014), and in the Columbia River estuary detritus was the dominant contributor to food webs in the ETM (Simenstad et al., 1990).
Physical processes other than gravitational circulation and tidal pumping, such as tidal asymmetries, can also influence the distribution of suspended particles and organic material in estuaries. Tidal asymmetries (i.e., flood-ebb bias in velocity) can elevate suspended sediment concentration at or near the extent of the tidal excursion (Dronkers, 1986) and be associated with elevated organic matter concentrations (Jago et al., 2006). Tidal asymmetries can be found in peripheral and terminal channels in both saline and freshwater portions of estuaries, and it occurs as the tidal wave transitions from progressive wave to standing wave and is reflected off the end of the terminal channel. Tidal asymmetry and associated dissipation of tidal energy along terminal channels causes incomplete exchange with surrounding habitats, and thus retention of water across daily, spring/neap, and seasonal time scales (Fagherazzi et al., 2008). Incomplete exchange creates gradients in physical properties such as residence time and variable within-channel exchange that act as a mechanism to either accumulate or disperse water constituents (P. R. Stumpner, Burau, et al., 2020), including water column or marshderived organic matter. Gradients in physical properties can create significant variability in chemical and biological properties and thus establish habitat gradients along the longitudinal axis of a slough or channel (McLusky & Elliott, 2004). Thus, localized processes driven by interactions between tidal currents and local topography can lead to high suspended sediment concentrations, elevated productivity, and/or accumulated organic matter.
As part of dendritic tidal marshes, terminal sloughs, and channels dominate peripheral habitats in many estuaries and can provide a crucial ecological function (Kneib, 1997). In many estuaries, anthropogenic impacts have altered or destroyed these peripheral habitats (Kennish, 2002), eliminating their ecological benefits and obscuring relationships between underlying physical processes and ecological responses. One such estuary, the San Francisco Estuary (SFE), has been severely altered in a variety of ways, including the loss of 98% of its historical tidal wetland (Nichols et al., 1986;Whipple et al., 2012). Despite the disruption of habitat loss, recent research in the SFE has identified regions where the hydrodynamic processes associated with terminal channels can still impact pelagic communities (Feyrer et al., 2017;Montgomery, 2017). The distribution of both plankton and nekton in a long terminal channel were associated with a local turbidity maximum caused by flooddominated tidal currents . This nekton includes pelagic fish, such as the endemic, endangered Delta Smelt Hypomesus transpacificus, which is present in the channel and similar habitats year-round (Hobbs et al., 2019;Moyle et al., 2016;Sommer & Mejia, 2013).
This turbidity maximum and associated habitat gradient represent variability in the accumulation and/or dispersal of water quality constituents, which could also include variation in productivity available to local consumers, specifically in situ primary production and local detrital accumulation. The term detritus here is meant to incorporate the actual detrital organic material and associated bacteria and microzooplankton essential for detrital nutrient cycling (Azam et al., 1983). Zooplankton and pelagic nekton in the San Francisco Estuary are ultimately fueled by multiple carbon sources contingent on availability, including phytoplankton (Grimaldo et al., 2009;Müller-Solger et al., 2002) and detritalderived sources (Harfmann et al., 2019;Howe & Simenstad, 2011;Young et al., 2020), and so further understanding of hydrodynamic mechanisms that drive the availability of these two different trophic pathways can inform habitat management for Delta Smelt and species-specific habitat management.
In our study, we sought to identify how gradients in hydrodynamic exchange can structure pelagic habitat and influence community and food web structure by addressing the following: (1) the relationship between hydrodynamic conditions and pelagic habitat; (2) the distribution of pelagic zooplankton and nekton among habitats that vary in hydrodynamic exchange; and (3) the importance of trophic pathways to consumers across hydrodynamically driven pelagic habitats. Understanding the role of hydrodynamics with respect to local food webs provides important insight into the causes of food web spatiotemporal variability.
This spatiotemporal variability can stabilize complex food webs (Polis et al., 1997;Winemiller, 1996) and maintain consistent support for higher trophic levels (McCann et al., 2005). Anthropogenic impacts have led to the loss of many geomorphic features, such as intertidal floodplain and tidal marsh with complex dendritic channels, and generally reduced heterogeneity (Whipple et al., 2012). Physical alteration has been concomitant with changes to the pelagic community and declines in native species (Feyrer et al., 2007;Kimmerer et al., 1994). These changes have resulted in localized hotspots of native fish abundance and/or diversity, typically associated with some remnant element of the natural landscape or ecological process (Moyle et al., 2012). Sampling was conducted within one such hotspot, the Sacramento Deep Water Shipping Channel (channel), with a documented residence of the endangered Delta Smelt (Moyle et al., 2016;Sommer & Mejia, 2013). The channel is a unique, man-made channel built in 1963 to allow large, oceangoing vessel traffic to access the Port of Sacramento. The channel is long (~42 km), has a uniform, consistent bathymetry (a central lanẽ 11 m deep with narrow benches~2 m deep), and is terminal, meaning that it has no hydrologic inputs except at the downstream mouth. The upstream end is disconnected from the Sacramento River by artificial channel gates and accumulated sediment. Municipal stormwater flows into the upstream end, and agricultural inputs are small but present throughout the length of the channel.
We selected sampling sites downstream, within, and upstream of the turbidity maximum observed by Feyrer et al. (2017). Each of these sites was contextualized by the underlying hydrodynamic processes (advection and dispersion) that define exchange zones (P. R. Stumpner, Burau, et al., 2020). Each site corresponded to specific U.S. Coast Guard navigation markers within the channel (see Figure 1; Moderate Turbidity, High-Exchange-CM 56; Turbidity Maximum, and Turbidity Minimum,. Exchange zones are defined based on the range of tidal excursions, or the distance a water parcel travels upstream from the mouth of the channel over a spring-neap period, and the average concentrations from numerical simulations (P. R. Stumpner, Burau, et al., 2020). The high-exchange zone is located from the mouth of the channel to the minimum tidal excursion, that is, the distance a water parcel moved on the weakest flood tide in a spring-neap cycle. In this zone, exchange is due to advection and occurs every tidal period (~12.5 h). The moderate-exchange zone (not sampled) is located between the range of minimum and maximum spring-neap tidal excursions; in this zone, exchange is due to advection and occurs over a period of 1-7 days during the transition from neap to spring tides. The low-exchange zone is located from the maximum tidal excursion upstream to the location where dispersive exchange with the downstream channel is negligible; exchange in this zone is due to dispersion and occurs on timescales >7 days. The no-exchange zone is defined as the location where dispersive exchange with the downstream channel is negligible, to the end of the channel; exchange in this zone is due to dispersion, largely insignificant with the downstream channel, and occurs on the order of weeks or longer.

| Field data collection
We collected samples in three seasons, Spring (May-June), Summer  net tows a subsample of collected nekton species were frozen and retained for later stable isotope analysis.
Two zooplankton samples were collected at each trawl location via a vertical net tow. The net had an opening of 50 cm, mesh size of 125 μm, and was towed through the water column at approximately 0.3 m/s. One sample was preserved in 10% formalin for later enumeration, and one sample was frozen for stable isotope analysis.
Zooplankton sample enumeration was conducted by EcoAnalysts Inc.
following the methods of Beaver et al. (2014) and included biovolume measurement. Before sampling at each location, we collected samples of primary producers such as dominant aquatic vegetation and suspended particulate organic matter (POM), including phytoplankton. POM was collected as seston vacuum-filtered onto precombusted 47 mm GF/F filters. Primary producer and POM samples were frozen and kept for stable isotope analysis.
Physical habitat parameters (water quality) were collected at the surface using a handheld multiparameter sonde (YSI EXO2; Yellow Springs Instruments Inc.). The physical measurements taken included temperature (°C), turbidity (FNU), specific conductance (μS cm −1 ), chlorophyll α (μg L −1 ), fluorescent dissolved organic matter (fDOM; μg L −1 ), an indicator of bioavailable detrital material, and dissolved oxygen concentration (mg L −1 ). We also took vertical water quality profiles to ensure that the water column was well-mixed. Hydrodynamic (i.e., water velocity and discharge) and hydrologic (i.e., tidally filtered discharge) data were obtained from U.S. Geological Survey operated water quality monitoring stations located nearby (see Data Availability). All sondes were calibrated and checked before and after sampling bouts to ensure measurement accuracy according to USGS operating procedures (U.S. Geological Survey, 2018).

| Sample preparation
Primary producers were rinsed with deionized water to remove contamination. To obtain enough material for zooplankton stableisotope analysis, individuals were pooled, with up to 300 individual calanoid copepods or cladocerans for each sample. All fish were filleted, with left posterior dorsal muscle tissue removed. Fish stomachs were then removed and placed in 90% ethanol. Shrimp muscle was extracted from the tail. Plant and animal tissues for stableisotope analysis were oven-dried for 48-96 h at 60°C and then ground to a homogeneous powder. Glass fiber filters containing POM filtrate were oven-dried for 3-6 h at 60°C, desiccated for 18 h with a container of 5 ml 12 M hydrochloric acid to remove carbonates, and then oven-dried for another 3-6 h at 60°C before sample submission.

| Diet analysis
All invertebrates in fish stomach contents were identified to the lowest practical taxonomic level and enumerated. Fish diets were analyzed by EcoAnalysts Inc.

| Isotope analysis
We weighed the powdered samples on a microbalance and placed them in tin capsules for isotope analysis. All samples were analyzed for δ 13 C T A B L E 1 Mean and standard deviation of measured water quality parameters at each season and exchange zone Long-term standard deviation for the laboratory is 0.2 permil (‰) for 13 C and 0.3‰ for 15 N. All isotope values are expressed in standard delta notation (δ) relative to international standards Vienna PeeDee Belemnite (V-PDB) and air for carbon and nitrogen, respectively (Sharp, 2017). Due to high C:N ratios we used the equation presented in Post et al. (2007) to normalize δ 13 C values for lipid content, where δ 13 C normalized = δ 13 C untreated − 3.32 + 0.99 × C:N.
Due to the difficulty in isolating stable isotope signatures for phytoplankton in seston-rich systems (Marty & Planas, 2008), we used the POM sample with the highest chlorophyll α concentration collected in each season (ranging from 4.3 to 9.3 µg L −1 ) to represent the isolated phytoplankton signature. Season-specific estimates were selected as dissolved inorganic carbon fractionation can differ by season (Finlay & Kendall, 2007). This is undoubtedly not a pure phytoplankton signature as other constituents were included in each of those samples; however, by selecting samples with known elevated phytoplankton concentrations we were able to conservatively estimate the relative contribution of phytoplankton to consumers.

| Data analysis 2.4.1 | The relationship between hydrodynamic conditions and pelagic habitat
The tidal excursion (i.e., integrated velocity between slack water) was estimated using the mean cross-sectional velocity collected at the mouth of the channel (site number 11455335). Tidal excursion estimates were corrected with a scaling factor of 1.5 based on comparison with tidal excursion measured with neutrally buoyant global positioning system drifters. The extent of the tidal excursion from the mouth of the channel was estimated for every tide for a spring-neap cycle centered around each of the six study periods. The range of tidal excursions was used to define three exchange zones: high, low, and no exchange, from the mouth to the upstream end of the channel (based on P. R. Stumpner, Burau, et al., 2020). We used these results to validate sampling site location relative to hydrodynamic exchange zones (high, low, and no exchange).
To assess the influence of hydrodynamic process on pelagic habitat, we compared physical (specific conductance, temperature, turbidity) and biological (chlorophyll α and fDOM) water quality variables using twoway ANOVA with site and season as factors. All analyses were done in Program R (R Core Team, 2019).

| The distribution of pelagic zooplankton and nekton
Densities of zooplankton and nekton were compared using two-way ANOVA with site and season as factors. Zooplankton and fish communities were compared using permutational multivariate analysis of variance (PerMANOVA) with site and season as factors. Fish diets were compared using permutational multivariate analysis of variance (PerMANOVA) on prey item counts with site and season as factors where sample sizes were appropriate (n > 10 for at least two of the three sites). All analyses were done in Program R (R Core Team, 2019) with the package "vegan" (Oksanen et al., 2019).
Biomass and total carbon were calculated for zooplankton and nekton (i.e., shrimp and fish). Zooplankton carbon was calculated by converting biovolumes measured during identification to carbon with the relationship: log BV = −1.429 + 0.808 log C, where BV is zooplankton biovolume (mL m −3 ) and C is zooplankton carbon (mg C m −3 ; Kimmel et al., 2006;Wiebe et al., 1975). Shrimp wet biomass was calculated using the following equation derived from collected and B is wet biomass (g). Biomass was converted to carbon following relationships for other caridean shrimp, where carbon is approximately 20% of wet biomass (Torres et al., 1994). We calculated fish wet biomass using published relationships for the San Francisco Estuary (Kimmerer et al., 2005). We converted fish biomass into carbon assuming 10% of fish biomass is carbon (Nixon et al., 1986).

| The importance of trophic pathways to consumers across hydrodynamically driven pelagic habitats
We used the package MixSIAR 3.1 (Stock & Semmens, 2016) in Program R (R Core Team, 2019) to determine the relative proportion of different primary productivity pathways to sampled pelagic consumers. This modeling technique incorporates uncertainty in source contributions, including trophic enrichment (Phillips et al., 2014). We ran each model independently for zooplankton, shrimp, and each fish species, with taxonomic group and site as random effects and used both δ 13 C and δ 15 N.     (Table 2).
PerMANOVA results showed significant community differences associated with both site (F 2,6 = 2.91, p = .014) and season (F 3,6 = 2.23, p = .045). Site differences were largely driven by greater density of cladocerans at the upstream site relative to downstream. Seasonal differences were driven largely by typical species turnover within the San Francisco Estuary (Kimmerer et al., 2018), with calanoid (Pseudodiaptomus forbesi) and cyclopoid (Limnoithona tetraspina) copepods dominating summer communities and other taxa dominating other seasons (Table 2).

| Nekton distribution
In all, 13,965 individual organisms were sampled with the midwater trawl representing pelagic nekton. The nektonic community was dominated by the Siberian Prawn Palaemon modestus, which comprised 89% of sampled organisms (84% of nekton biomass;   (Table 3). POM was typically more depleted than vascular vegetation, and phytoplankton estimates were always more depleted than vascular vegetation (Table 3). Mean δ 15 N for vascular vegetation differed spatially, with more depleted values farther upstream (Table 3). Emergent vegetation exhibited seasonal variability in δ 15 N (1.46‰-9.1‰; Figure 5), likely due to senescence , with little evidence for seasonal variability in other macrophyte groups. POM was most F I G U R E 3 Sampled water quality parameters associated with each exchange zone. Spring and summer from different years are pooled due to low variability while winters are kept separate. Box represents the median and interquartile range (IQR) and the whiskers extend to 1.5 × IQR. Points represent individual samples. Colors as in Figure 1. SC, specific conductance F I G U R E 4 Biomass of pelagic community constituents, with colored bars representing the mean and whiskers representing the standard error. Nekton are separated into shrimp and pelagic fish. Zooplankton densities have been converted to match nekton units; for zooplankton biovolume see Table 2. Colors as in Figure 1 T A B L E 2 Mean biomass of sampled pelagic taxa by each season and exchange zone

| Mixing model results
Both zooplankton groups were reliant on similar trophic pathways, with phytoplankton a dominant contributor in all season/region combinations (mean-76%; Figure 6). The contribution of vegetation to zooplankton was typically highest at the low exchange habitat in Winter. Both zooplankton groups (calanoid copepods and cladocerans) exhibited similar trends.
Abbreviations: EMV, emergent vegetation; FAV, floating aquatic vegetation; POM, particulate organic matter; SAV, submersed aquatic vegetation; TER, terrestrial vegetation. often due to low exchange and higher water age in this region (Downing et al., 2016;Gross et al., 2019). The gradient in specific conductance and elevated turbidity near the maximum extent of the tidal excursion support our characterizations of the three sites as high, low, and no exchange. Elevated chlorophyll α in the no exchange habitat is consistent with high residence time waters regionally (E. B. Stumpner, Bergamaschi, et al., 2020) but may also be related to water clarity, as elevated turbidity in the low exchange site may have limited phytoplankton growth (Cloern, 1987). Variability in phytoplankton community may also have affected in situ chlorophyll α measurement, as phytoplankton communities are noted to differ across regional habitats (E. B. Stumpner, Bergamaschi, et al., 2020).
Notably, these mechanisms alone do not explain low chlorophyll α that the influence of hydrodynamics on pelagic habitats can be sampled or measured on relatively small spatial scales, specifically in tidally influenced terminal channels.

| Habitats and distribution of pelagic zooplankton and nekton
The distribution of zooplankton and nekton within the channel was strongly correlated with the hydrodynamically driven habitat zones described above. The upstream zone characterized by no exchange with downstream habitats supported higher zooplankton densities, and higher densities of Cladocera, a group more abundant in low-velocity, freshwater habitats regionally (Frantzich et al., 2018;Kimmerer et al., 2018). The downstream, high exchange zone supported the lowest densities of zooplankton, which could reflect low phytoplankton abundance observed in the study or simply the relatively low densities of F I G U R E 5 Biplot of primary producer and consumer isotope values. Light gray boxes represent phytoplankton estimates and dark gray boxes represent pooled vascular vegetation (see Table 3) | 79 zooplankton present in many nearby SFE pelagic habitats (Kimmerer et al., 2018). Nekton were generally more abundant in the intermediate, low exchange zone. Shrimp were almost exclusively observed in the low exchange zone and at concentrations rarely observed in the San Francisco Estuary, particularly in pelagic habitats (T. Brown & Hieb, 2014;Young et al., 2017). This could reflect a strong affinity for turbidity, unmeasured food availability related to detrital accumulation, larval retention and local recruitment, or some other unmeasured habitat association.
Pelagic fishes were most abundant in the low exchange zone and were generally positively associated with turbidity. Although this relationship has been observed with Delta Smelt, in many systems Threadfin Shad have negative relationships with turbidity (Bull et al., 1995). Given the high zooplankton densities upstream, low densities of pelagic fishes upstream run counter to expectations.
This discrepancy suggests that turbidity might be more important than food availability in dictating the distribution of pelagic fishes, potentially reflecting the abundance of predatory species not sampled effectively by our gear (e.g., Striped Bass Morone saxatilis).

| Hydrodynamics and benthic-pelagic coupling
Mixing model results showed that the contribution of different primary producer groups to consumers varied across the observed pelagic habitats, and when normalized by abundance of pelagic  (Jago et al., 2006;Suzuki et al., 2012) and variable phytoplankton densities across hydrodynamic habitats (E. B. Stumpner, Bergamaschi, et al., 2020).
Our measurements of how these different organic matter sources are incorporated into pelagic consumers are consistent with those of Suzuki et al. (2012), which documented elevated detrital organic material in an ETM, and elevated phytoplankton densities up-and downstream of the ETM.
The predominant contribution of phytoplankton to both zooplankton groups is consistent with other local studies, which highlight the primacy of phytoplankton to zooplankton diets (Grimaldo et al., 2009;Müller-Solger et al., 2002). However, our conservative estimates of phytoplankton stable isotope signatures used in the mixing model are not pure phytoplankton but rather reflect a combination of phytoplankton and other POM of undefined origin. This uncertainty makes it difficult to precisely identify the relative contributions of phytoplankton and other primary productivity pathways; however, the occasional contribution of macrophyte-derived carbon to zooplankton should not be discounted, as it may provide additional resilience in a dynamic system (sensu Polis et al., 1997). Similar studies of the calanoid copepod Acartia tonsa have demonstrated plasticity in trophic support across an ETM gradient, whereby detritus is consumed within the ETM and phytoplankton on the edges (Derisio et al., 2014). The degree of detrital consumption may be variable across systems, as bioavailability and quality of detrital POM varies (Martineau et al., 2004;Sobczak et al., 2002).
Nekton exhibited high variability in the utilization of trophic resources other than phytoplankton in certain seasons and sites. Fish diet data also reflected this variability, as the importance of prey items other than zooplankton to pelagic fish was highly mutable in space and time. Notably, zooplankton were generally less prevalent in diets during winter, and at downstream locations, which is reflected in the importance of other trophic resources in those seasons and sites. Trophic variability, coupled with the variability in measured nekton biomass, drove high variability in the contribution of littoral/benthic productivity to the pelagic food web. The contribution of littoral/benthic productivity, in the form of vascular macrophytes and associated detritus, to pelagic consumers is widely recognized as an important element of pelagic food webs (Perissinotto et al., 2003;Vadeboncoeur et al., 2002). Littoral/ benthic productivity can move across habitat boundaries (i.e., benthic-pelagic coupling) directly as either dissolved or POM or be transferred trophically as pelagic secondary consumers ingest littoral/benthic herbivores (Kneib, 2002 Figure 2), and the influence of this perturbation on local food webs may have been brief and not captured by our sampling strategy, particularly given the punctuated nature of runoff impacts and delays in tissue turnover for many organisms (Vander Zanden et al., 2015). Hydrology is a common driver of food web structure in many habitats, including estuaries (Possamai et al., 2020), and so the potential for hydrologic disruption of hydrodynamic-driven food web structure should not be discounted.

| Conservation and management implications
Identifying hydrodynamic controls on pelagic habitats, communities, and food webs will aid in understanding distributions of pelagic communities at multiple spatial scales, and is important information needed for effective conservation and management. Characterizing hydrodynamic controls on pelagic habitats is a mechanistic approach that can be broadly applied. Without using hydrodynamic information, different pelagic habitats can easily be identified but predicting where or how these may form or change is difficult without knowing the underlying physical controls. For example, we can measure turbidity, but it is more difficult to predict where turbidity will be high YOUNG ET AL.
| 81 as a result of restoration or other management actions. Delta Smelt, a pelagic fish endemic to the SFE, is strongly associated with turbid water throughout its extant distribution (Feyrer et al., 2007;Nobriga et al., 2008), and any factor influencing local turbidity may influence the distribution of these pelagic fish. Similarly, the distribution of phytoplankton (and thus chlorophyll α) is a known driver of zooplankton abundance (Ambler et al., 1985), important prey for Delta Smelt and other declining pelagic fishes. Chlorophyll α concentration is easy to measure, but harder to predict, and enhanced understanding of physical controls on these processes can improve conservation management.
Current management priorities in the SFE focus on habitat restoration, specifically recreation of dendritic tidal wetland channels and associated floodplains (Herbold et al., 2014), and anthropogenic modifications to hydrology, primarily through manipulation of water releases and withdrawals (Luoma et al., 2015). Hydrodynamics can strongly influence pelagic habitat and associated biological responses at multiple spatiotemporal scales, which could provide crucial information for habitat restoration managers. This information could be used to inform restoration site placement and prioritization along existing hydrodynamically driven pelagic habitat gradients and assist with forecasting the development of hydrodynamically driven gradients within restored sites. By understanding the extent of exchange between channels or within a channel, we are better able to predict and understand the distribution of pelagic habitats within that channel and their subsequent effects on pelagic communities, resulting in greater confidence in forecasts of restoration outcomes.
Terminal channel systems with more complex inputs (e.g., terminal tidal marsh channels with adjacent tidal floodplain) may differ in many respects, but the influence of tidal hydrodynamics on terminal channel pelagic communities will likely still be profound. It is important to note that habitat restoration, levee failure, or other large-scale perturbations to regional hydrodynamics (including changing precipitation patterns) can alter tidal forcing and thus the location of pelagic habitats within tidal channels (Herbold et al., 2014). Although hydrologic influences on these gradients were limited to a single season in our study, further study of flow impacts on hydrodynamic features is needed to evaluate seasonal persistence and potential impacts of altered hydrology. Down-scaled global climate models predict increases in water temperature, salinity and sea level, and decreased precipitation and freshwater outflow (L. R. Brown et al., 2013;Cloern et al., 2011), which may also impact the persistence or location of these hydrodynamically driven habitats.
The results of our study highlight the relationship between hydrodynamic processes and biological responses and suggest that understanding the physical process can improve the understanding of pelagic communities.

ACKNOWLEDGMENTS
This project was supported by an interagency agreement from the U. Burau and E. Stumpner for invaluable discussion and feedback and L.
Loken for thoughtful review and critique. Two anonymous reviewers greatly improved manuscript clarity. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are openly available in the USGS National Water Information System (NWIS; https://doi.