The goal of this study was to identify, classify, and synthesize predictable patterns in water quality responses along longitudinal stream flowpaths from many observations of combinations of chemical parameters in the context of distance downstream and watershed characteristics (Figure 2). This typology of longitudinal water quality patterns along stream flowpaths can provide a foundation for future studies, watershed assessments, evaluating management and restoration efforts, and making comparisons across watershed responses. In this paper, we define typology as a system for classifying longitudinal patterns in water quality along stream flowpaths in certain categories according to how they are similar. Based on literature studies and our previous work (Figure 2; Table 1), we propose a typology with the following longitudinal patterns in chemical concentrations: (1) increasing, (2) decreasing and dilution, (3) chemostatic, (4) transition zones, (5) pulses, (6) plateaus, (7) biogeochemical uptake, (8) transformation, and (9) complex piecewise changes along flowpaths (Figure 2; Table 1). There are a growing number of examples of longitudinal synoptic studies in the literature (Table 1), which present a broader context for proposing a typology and testing patterns of responses observed in our extensive regional data sets.

Throughout this paper, we focus on the synthesis of many original data sets from LSS monitoring to test our typology and its applications in water quality studies, instead of trying to report results in a “typical” hypothesis-field collection-measurement-results-discussion type of format. The integration of multiple chemicals as chemical cocktails along flowpaths is a novel component of our typology and approach and allows tracking of pollution sources and identification of water quality tradeoffs (e.g., one contaminant is retained, but another contaminant is released along flowpaths) (Kaushal et al., 2018; Kaushal et al., 2020; Kaushal et al., 2022). We use the narrative parts of this paper (and our original results and figures) to demonstrate and illustrate distinct longitudinal patterns of response observed in extensive data sets. Overall, the focus of our paper is on the synthesis of data to develop and test a typology of longitudinal water quality patterns and responses for informing water quality monitoring and management (instead of trying to report results from individual case studies).

Specifically, we compare and discuss our original data characterizing longitudinal patterns in water quality with many others from the literature to synthesize the typology of water quality trends and transitions along flowpaths. For example, we apply LSS monitoring of chemical cocktails to explore how water quality evolves as streams flow through: (1) progressively degraded areas due to urbanization, (2) restored reaches with hydrologically connected floodplains, and (3) forested conservation areas including national, regional, and local parks (natural recovery). Conservation of surrounding forested lands and wetlands, and restoration approaches involving stream-floodplain restoration can enhance attenuation and transformation of contaminants. There is a growing need for longitudinal monitoring approaches to assess whether the effects of conservation and restoration can be detected along flowpaths (Kaushal et al., 2023a).

Important mechanisms influencing attenuation and transformation include: (1) increased hydrologic residence times and contact between stream water and floodplain and hyporheic sediments, (2) ion exchange in lesser polluted soils, (3) complexation with organic matter from riparian plants and soils, (4) and biological uptake and transformation from plants and microbial communities (e.g., Grant et al., 2018; Kaushal et al., 2008b, Noe and Hupp, 2009, Roley et al., 2012, Kaushal et al., 2022, Mayer et al., 2022). We propose that an improved understanding of water quality gained through LSS monitoring and a chemical cocktail approach (tracking chemical combinations across space and time) is helpful for unlocking the potential for co-management of contaminants and ecosystem recovery based on conservation and restoration practices and best management practices (BMPs) (Kaushal et al., 2022; Kaushal et al., 2023a).

As mentioned previously, there are relatively fewer examples of longitudinal stream synoptic data compared with time-series of concentrations and fluxes from long-term monitoring sites within watersheds. Here, we investigated the sources, transport, attenuation, and transformations of chemical cocktails from data collected along 10 longitudinal stream flowpaths in the Baltimore-Washington D.C. metropolitan region (Table 2; Figure 3; Supplemental Table S1) in the Chesapeake Bay watershed within a combined 1,765 km² area (Table 2; Figure 2). Detailed information on study sites and synoptic sampling surveys are in Table 2 and (Supplemental Table S1). The longitudinal stream flowpaths include data from our long-term study sites. At these sites, we have characterized and compared the hydrology, biogeochemistry, and geochemistry over decades (e.g., Kaushal et al., 2005, Kaushal et al., 2008a, Kaushal et al., 2011, Kaushal and Belt, 2012; Kaushal et al., 2014a; Kaushal et al., 2014b; Kaushal et al., 2014c; Newcomer et al., 2012, Smith and Kaushal, 2015, Kaushal et al., 2019, Bhide et al., 2021, Grant et al., 2022, Kaushal et al., 2022, Kaushal et al., 2023b). These longitudinal flowpaths share similar climate and regional patterns but have: different land use characteristics (Supplemental Table S1), different management strategies, and different sources of pollution (e.g., pulses of sewage effluent, urban stormwater runoff, and road salt). Urban streams and rivers in this study flow either through conservation areas (regional and national parks) and stream-floodplain restoration areas or progressively degraded urban areas (site descriptions further below). Thus, we tested the ability of LSS monitoring to provide further insights regarding sources, transport, and transformations of different chemicals at our long-term monitoring sites and to synthesize a typology of typical water quality responses to help guide future longitudinal studies.

We explored recovery in water quality at our long-term monitoring sites by comparing transport and transformation of contaminants using LSS monitoring (300 sampling sites along combined stream synoptic distances of 337 km across the 10 watershed flowpaths, in which the mainstem of the flowpath was sampled at 3–45 locations with a range in distance between sites from 0.4 to 20.85 km) (Table 2). We compared retention and release of chemical cocktails along flowpaths using statistical analysis of multiple elements or chemical cocktails. Although we conducted LSS monitoring at different times at different sites, longitudinal patterns can still be determined and quantified. In some cases, we present data from repeated LSS monitoring at our long-term monitoring sites to investigate water quality applications across seasonality and hydrologic events. Finally, we compared our longitudinal stream synoptic patterns with others from the literature to test a typology of water quality trends and transitions along flowpaths.

There are important caveats and limitations in our stream synoptic approach and the case studies that we explore throughout this paper. For example, not all stream synoptic surveys measured all the same constituents along each flowpath, not all stream synoptic surveys were subject to the same statistical tests, and not all synoptic surveys were sampled during the same season, and there were differences in spatial intervals or general time periods of different synoptic surveys. In general, the number of stream flowpaths, number of stream locations, and number of time samples were collected were based upon the following criteria: (1) long-term monitoring and background knowledge of study sites, (2) site access and safety, and (3) feasibility. Regarding long-term study sites, we have characterized the background hydrology and biogeochemistry of all of the streams mentioned in this paper including Gwynns Falls (Kaushal et al., 2008a; Kaushal et al., 2011), Campus Creek and Paint Branch (Kaushal et al., 2022; Wood et al., 2022), Scotts Level Branch (Newcomer et al., 2012; Galella et al., 2021), Sligo Creek (Galella et al., 2021, Maas et al., In Review), Bull Run (Bhide et al., 2021; Grant et al., 2022), Anacostia watershed (Kaushal et al., 2020; Kaushal et al., 2022; Kaushal et al., 2023b). In general, we typically sampled synoptic sites along every 100 m for small streams, and we sampled every 500–1,000 m along larger streams. The number of stream locations for synoptic sampling were based on site accessibility and safety. Many streams were located along bike paths where we could access sampling points quickly and frequently. At some sites we were able to conduct stream synoptic surveys routinely once per month at baseflow to characterize seasonality. In some case studies, we also sampled immediately after notable weather events, primarily during road salt applications and storms (e.g., Kaushal et al., 2018; Kaushal et al., 2022). Although there were important caveats and limitations, this study aggregates and compares many results of stream synoptic surveys to observe downstream responses and synthesize a typology of water quality patterns along flowpaths based on many original case studies and the global literature (Figure 2; Table 1).

Multiple longitudinal stream synoptic locations were located throughout each of the above watersheds to investigate longitudinal changes in downstream concentrations and/or isotopic compositions of multiple chemical contaminants. In addition, we also sampled shallow groundwater at Scotts Level Branch along the flowpath from uplands to the stream channel using transects of piezometers. Methods of sampling for groundwater at Scotts Level Branch and details on groundwater well installation are in (Wood et al., 2022). In each watershed, LSS monitoring was completed within 24 h or on consecutive days with similar streamflow conditions. We did not always move downstream along a watershed during the day; in some cases, we moved upstream or sampled the entire watershed flowpath by multiple teams simultaneously throughout the day (e.g., sampling timing was sometimes different for flowpaths and data collection could sometimes rotate through upstream, downstream, and simultaneous collection along flowpaths). Thus, the chemical patterns along flowpaths were not due to diurnal patterns throughout the day. Our sampling resolution between sampling points varied based on the size of the watershed, but was typically hundreds of meters or a few kilometers; thus, we sampled in detail along the length of each watershed to evaluate where urban degradation occurred and/or restoration activities were implemented. In some cases, these flowpaths extended from headwaters to the stream outflow to receiving waters. Sampling locations for the synoptic sites along each mainstem were chosen based on accessibility, presence of tributary junctions, and positioning of conservation and restoration features.

Characteristics (i.e., stream, sampling dates, longitudinal distance sampled, mainstem and tributary sampling points, and USGS station location and discharge) of each synoptic event are included in Table 2. Water samples were collected at the tributary outflows and at least 100 m downstream from the tributary confluence along the mainstem to ensure well-mixed conditions (almost all data presented in this paper is from along the mainstem). Latitude and longitude for synoptic sites were recorded using GPS systems or applications. Along the Gwynns Falls, our synoptic sampling also included discharge measurements using a Marsh McBirney 2000 (Hach Co., Loveland, CO, United States) velocity meter. LSS monitoring was conducted mostly during baseflow conditions, and there was also targeted sampling after snow events along Rock Creek, Bull Run, and Anacostia River, Campus Creek.

Major element (e.g., Na, Ca, K, Mg) and trace element (e.g., Mn, Zn, Sr, Cu) concentrations were measured by inductively coupled plasma optical emission spectrometry in an acidified (0.5% high-purity nitric acid) analytical matrix on a Shimadzu Elemental Spectrometer (ICPE-9800). For major element measurements, the acidified samples were nebulized in radial mode (across a plasma flame). For trace element measurements, the acidified samples were nebulized in axial mode (down plasma flame). The instrument was calibrated to the range of trace metals that are commonly observed in urban streams in accordance with analytical guidelines for surface water analysis (Galella et al., 2021). There were no non-detects in the measurements of trace elements we report for these polluted urban streams. Major and trace elements in Gwynns Falls acidified stream samples were analyzed by ICP-MS at the United States Environmental Protection Agency Lab in Ada, Oklahoma, United States. Dissolved organic carbon (DOC), dissolved inorganic carbon (DIC) and total dissolved nitrogen (TDN) were measured using a Shimadzu Total Organic Carbon Analyzer (TOC-V CPH/CPN) and total nitrogen module, TNM-1 (Haq et al., 2018). Isotope ratios of ¹⁵N/¹⁴N in nitrate in stream samples from the Anacostia River and its tributaries were measured at the UC Davis Stable Isotope Facility. Isotope ratios of ¹⁵N/14N were measured using a ThermoFinnigan + PreCon trace gas concentration system interfaced to a ThermoScientific Delta V Plus isotope-ratio mass spectrometer.

Summary statistics were calculated for all chemical constituents measured on each stream. We used linear regression analysis to test for increasing or decreasing trends between chemical concentrations and distance downstream. If concentration versus distance did not follow a linear trend, the polynomial fits in Excel or the curve fitting toolbox in MATLAB (cftool Mathworks, 2001) was used for quantitative solutions for longitudinal patterns in chemical concentrations.

Principal component analysis was used to assess the temporal and spatial chemical cocktails along the flowpath in Bull Run and Gwynns Falls. Before performing PCA, the data was evaluated using histograms and quantile-quantile plots to determine the normality. The data was determined to be positively skewed and log-transformed to avoid violating the normality constraints of PCA (Davis and Sampson, 1986). The data was standardized to have a mean of zero and a standard deviation of one. The PCA was performed in R studio version 1.4 and used the packages FactoMineR and factoextra (Lê et al., 2008). For both analyses, a resampling-based stopping rule (Peres-Neto et al., 2005; Rippy et al., 2017) was used to identify principal components that explained significantly more data variance than expected due to chance (significance determined at a 95% confidence level). Only these significant components were retained and interpreted.

LSS monitoring illustrated distinct typologies in chemical concentrations along the Gwynns Falls flowpath and one of its major tributaries, Scotts Level Branch (Figure 4). Along the Gwynns Falls flowpath, we observed statistically significant increasing trends for calcium, magnesium, sulfur, boron, and molybdenum (increasing trends of approximately 0.014, 0.005, 0.003, 0.005, 3 × 10⁻⁵ mg/L per km, respectively) (p< 0.05) (Figure 4A). We also observed significant decreasing trends in TDN, Ba, and Mn concentrations (decreasing trends of approximately -0.002, -0.001, -0.0007 mg/L per km, respectively) (Figure 4B). Redox sensitive elements like Fe and Mn showed a pulse typology with increasing distance downstream and varied in concentrations suggested by the presence of floodplain restoration sites and parks (Figure 4E). Some elements like K and Sr showed transition typologies characterized by stepwise changes with increasing distance downstream (Figure 4F). Some trace elements like Cu and Pb (at very low concentrations) showed no pattern with distance downstream (Figure 4C). Overall, these results suggest that there can be distinct typologies in longitudinal changes in chemical concentrations along stream flowpaths.

Streamflow increased longitudinally along the Gwynns Falls, and there were distinct relationships between elemental concentrations and increasing discharge (Figure 5). Not all elements were simply diluted with increasing streamflow along the flowpath (Figure 5). In addition, longitudinal patterns in chemical concentrations along the mainstem did not appear to show abrupt pulses or dilutions from discharge from tributaries (Supplemental Figure S1); a detailed mass balance analysis of loads along the Gwynns Falls mainstem and inputs from tributaries can be found in Kaushal et al., (2014a). Instead, longitudinal concentration-discharge patterns showed distinct patterns and processes along the Gwynns Falls UWC. For example, concentrations of TDN and Ba decreased linearly whereas S concentrations increased linearly and B concentrations increased curvilinearly (Figure 5). There was no evident relationship between streamflow and stable Na concentrations (chemostatic behavior) (Figure 5). Differences in the longitudinal concentration-discharge relationships suggested the possibility to group elemental transport into distinct chemical cocktails along the Gwynns Falls flowpath.

LSS monitoring revealed biogeochemical transformations and transition zones along the Gwynns Falls flowpath (Figure 6). Along the Gwynns Falls flowpath, we observed polynomial declining trends in concentrations of CO₂ and N₂O with increasing distance downstream (Figure 6). There was also a decline in the humification index of organic matter (HIX), which suggested a shift from terrestrial to aquatic sources of DOC along the flowpath (Smith and Kaushal, 2015) (Figure 6). In addition, there was an abrupt transition typology and increase in DOC concentrations approximately 20 km along the Gwynns Falls flowpath (Figure 6). Thus, CO₂, N₂O, and HIX appeared to show a downstream dilution typology (Figure 2) whereas DOC appeared to show a transition zone typology (Figure 2) where it increased abruptly downstream.

LSS monitoring revealed biogeochemical transformations along the Anacostia River flowpath (Figure 7). The importance of biogeochemical transformations along the Anacostia River flowpath can be explored using LSS monitoring across time. Na and K concentrations typically showed significant decreases longitudinally downstream (Figure 7), as the river flowed through conservation areas of Anacostia National Park with tidal wetlands and forests. Concentrations of K, a biologically limiting nutrient, were substantially lower than Na along the flowpath (Figure 7). In contrast, concentrations of dissolved Fe showed a pulse typology as the river flowed through adjacent forest and wetland areas likely due to changes in redox conditions and biogeochemical reactions (similar to pulses in Fe and Mn in restoration features at Scotts Level Branch) (Figure 7).

LSS monitoring suggested potential transformations of nutrients and organic matter along the Anacostia River flowpath (similar to changes in dissolved Fe as discussed above), as water flowed from degraded suburban and urban headwaters into conserved tidal wetlands and forests (Figure 8). Farther downstream (particularly along the Northwest Branch through Burnt Mills Park and along the Anacostia mainstem through Anacostia National Park), there appeared to be uptake of nutrients and increases in organic matter concentrations (Figure 8). Nitrate (NO₃ ⁻) and soluble reactive phosphorus (SRP) declined rapidly along the Anacostia mainstem, while particulate organic carbon and the C:N ratio of particulate organic matter increased (which suggests the importance of biological uptake of nutrients and transformation into organic matter) (Figure 8). Paint Branch and Sligo Creek showed significantly elevated δ¹⁵N-NO₃ ^- above +10 within the range of sewage or caused by fractionation (Kaushal et al., 2011) (Figure 8). The δ¹⁵N-NO₃ ^- (Figure 8D) decreased along flowpaths draining extensive riparian conservation areas such as the Northwest Branch (through Burnt Mills Park) and Anacostia mainstem through Anacostia National Park. Conversely, the δ¹⁵N-NO₃ increased along urban flowpaths with minimal riparian buffers such as Sligo Creek, the Northeast Branch, and Paint Branch (Figure 8D).

LSS monitoring was also able to delineate the spatial and temporal extent of biogeochemical transformations and transition zones associated with stream and floodplain restoration activities along the Campus Creek flowpath (Figure 9). For example, there were substantial declines in dissolved oxygen along the Campus Creek flowpath, as it flowed through reaches with regenerative stormwater conveyance (pools and wetlands with longer hydrologic residence times) (Figure 9). During summer months, dissolved oxygen concentrations were less than 2 mg/L indicating hypoxia (Figure 9). As dissolved oxygen declined along the regenerative stormwater conveyance pools and hydrologically connected floodplains, there were increases in concentrations of dissolved Fe and Mn (similar to restored reaches in Scotts Level Branch and forest and wetland conservation areas in Anacostia National Park) (Figure 9). This was likely due to low oxygen conditions and reduction of Fe and Mn to soluble forms. There was also a relationship between % forest and shrub cover in the watershed and concentrations of Fe and Mn in Campus Creek (Figure 9). In contrast to redox sensitive elements, Na showed a dilution typology as the stream flowed through the regenerative stormwater conveyance pools, which may be expected from biologically conservative elements.

Results from LSS monitoring also revealed attenuation of pollution from wastewater, road salt, and urban stormwater along flowpaths through restoration and conservation areas (Figure 10). There were also significant linear decreasing relationships between % watershed forest and shrub cover and chemical concentrations in these same watersheds (Figure 11). We observed two types of rapid declining patterns in concentrations of many chemical constituents (Na, K, Ca, Mg, Mn, TDN, DOC, conductivity) along the Bull Run flowpath downstream of the UOSA wastewater treatment plant. First, there was a very rapid initial decline in concentrations between the UOSA effluent and our first sampling point downstream likely due to dilution as a physical driver, as evidenced by the consistent and similar decreasing slopes for chemicals between UOSA at this first sampling station (Figure 10A). Different chemicals showed varying rates of decline and transport lengths likely due to their reactivity, changes in their phases (dissolved vs. particulate), and other biogeochemical processes, as Bull Run flowed through extensive forest and wetland areas in a regional park. Please see Figure 11 for examples of significant relationships between chemical concentrations and % forest and shrub cover in Bull Run.

A plateau typology in chemical concentrations was observed in Rock Creek along a stream flowpath through conservation areas and a national park. During a winter road salt event, there was increasing specific conductance and concentrations of major and trace elements such as Na, Ca and Mn with distance downstream (Figure 10B). However, these chemical concentrations appeared to plateau and remain more constant once Rock Creek flowed through the heavily forested National Park with extensive forest riparian buffers (Figure 10B). Please see Figure 11 for examples of significant relationships between chemical concentrations and % forest and shrub cover in Rock Creek.

Finally, we observed a decreasing typology of urban stormwater pollution in Scotts Level Branch for a portion of the flowpath, where there is a stream restoration project involving hydrologic reconnection of the floodplain with the stream. After a rainstorm, we observed declines in specific conductance and concentrations of carbon, nutrients, and major elements in Scotts Level Branch, as it flowed from an urban storm drain through restored floodplains and a local park; however, there were too few sampling points along this particular stream reach to analyze rates of attenuation (Figure 10C). Please see Figure 11 for examples of significant relationships between chemical concentrations and % forest and shrub cover in Scotts Level Branch. There was a coinciding increase in Fe and Mn concentrations downstream of the restored stream-floodplain complexes as mentioned previously (Figure 4), likely due to a change in redox conditions favoring Fe and Mn dissolution and denitrification. Along lateral groundwater flowpaths, there was also a decline in Na with increasing distance away from the stream along the riparian zone in Scotts Level Branch (Figure 10D). Attenuation of chemical cocktails along the Scotts Level Branch flowpath are likely due to increased biogeochemical transformations such as denitrification and cation exchange in floodplain soils.

Applications of LSS monitoring showed potential to track fate and transport of multiple chemical cocktails along flowpaths using typologies (similar longitudinal patterns among chemicals). Principal component analysis also revealed that there were distinct chemical cocktails of organic matter and elemental mixtures formed along the Bull Run flowpath (Figures 12, 13). In Figure 12, the dashed line indicates movement along the flowpath from upstream to downstream. The ellipses represent the land use changes along the flowpath. The different regions they encapsulate were identified statistically using hierarchical clustering of principal components (R, function HCPC). They were subsequently characterized (interpreted) both visually using maps along the flowpath and analytically by measuring the changes in impervious surface cover, forest cover, and riparian buffer width in the region. Dissolved organic matter downstream of the UOSA wastewater treatment plant was enriched in the biological freshness index (BIX) and protein-rich peaks B and T, which can indicate more labile DOM from microbial sources (Huguet et al., 2009; Coble et al., 2014). In contrast, sites with DOM showing higher humification index (HIX), which can indicate more refractory DOM from humic substances (Zsolnay et al., 1999; Ohno, 2002), were located either upstream of UOSA or considerably further downstream near the reservoir in Bull Run Regional Park (Figure 12). An alternative interpretation is that B and T peaks were correlated with PC1, but A, C, and M peaks have very similar loading strengths. This may suggest that all peaks were enriched, which can be explained as higher concentrations, which result in higher intensities for all peaks. Indices that drive PC2, or ratios of peaks can give concentration-independent metrics. At the start of the flowpath, dissolved organic matter had a stormwater signature that changed to a wastewater effluent signature, and then finished with a forested groundwater signature in Bull Run Regional Park. Thus, there were distinct changes in chemical cocktails of organic matter that could be traced along the Bull Run flowpath using LSS monitoring (see black trace in Figure 12).

LSS monitoring was also able to track transitions in chemical cocktails upstream of the wastewater treatment plant, immediately downstream of the UOSA wastewater treatment plant, and within forest and wetland conservation areas in the regional park (Figure 13). Elemental mixtures immediately downstream of UOSA were enriched in DOC, Na, TDN, Cu compared with elemental mixtures at sites within the forests and wetlands of Bull Run Regional Park (Figure 13). Given that N is a limiting nutrient in ecosystems, N from wastewater treatment plant discharges can also serve as a reactive tracer in downstream ecosystems (Figure 13). There were significant positive linear relationships between TDN concentrations and a variety of chemical parameters (DOC, Na, humic-rich peaks A and C in organic matter, and Cu) along Bull Run (Figure 13). There were other typologies of asymptotic trends and negative polynomial trends for chemical cocktails downstream (Figure 13). Results from longitudinal patterns illustrated transformation of different chemical cocktails with transitions spanning from urban wastewater effluent to downstream forest/wetland conditions.

We observed longitudinal patterns in chemical concentrations along the urban watershed continuum (UWC), which can be classified into increasing, decreasing, transitions, or no trends (Figure 2; Table 1). The typologies of longitudinal patterns can be influenced by a variety of factors depending on land use, geology, reactivity of chemical constituents, hydrology, and time (Table 1). Increasing longitudinal trends can occur downstream when biogeochemical supply exceeds the attenuation capacity of stream ecosystems (biological uptake, dilution, ion exchange) (e.g., Gove et al., 2001; Bhatt and McDowell, 2007; Kaushal et al., 2017) (Table 1). Decreasing longitudinal trends can be due to dilution, biological uptake, or ion exchange along flowpaths (e.g., Marti et al., 2004; Welty et al., 2023) (Table 1). There can be abrupt longitudinal transitions in water quality caused by shifts in surrounding land use, management, hydrology, underlying geology, etc. (Table 1). There can also be idiosyncratic, erratic, and pulsed inputs (e.g. Sivirichi et al., 2011; Pennino et al., 2016a; Pennino et al., 2016b) (Table 1). The absence of longitudinal trends can also be due to stable steady state conditions (importance of groundwater contributions). Chemostatic behavior has been used to refer to steady state chemical behavior over time; cases where ion concentrations, for example, do not change with time, even after a major intervention (presumably due to very diverse flow paths and transit times). However, it is important to note that chemostatic behavior can also be used to describe spatial patterns along flowpaths (Hale and Godsey, 2019). We discuss some examples of longitudinal patterns and processes below:

(1) Increasing trends in concentrations along flowpaths–We observed increasing downstream trends in some elemental concentrations along the Gwynns Falls (Ca, S, Mg, B, Mo) (Figure 4). These elements could accumulate in groundwater and lead to increasing downstream trends (Ledford and Lautz, 2015; Gabor et al., 2017) or there could be increasing contributions to surface waters from downstream sewage leaks, weathering of impervious surfaces, etc. Base cation and chloride concentrations can increase longitudinally along streams due to increased inputs from groundwater (Gabor et al., 2017), increased road salt accumulation in floodplains (Ledford and Lautz, 2015), increased ion exchange (Kaushal et al., 2022), and weathering from impervious surfaces (Kaushal et al., 2017). (Figures 1, 2). Organic carbon concentrations may also increase along some stream flowpaths due to increased terrestrial inputs (e.g., Sabater et al., 1993) or increased production of algal and bacteria (Finlay, 2011; Kaushal et al., 2014a).

For a more complete list of typologies and classifications of synoptic-scale patterns based on observations from the literature, please see Figure 2; Table 1.

Time, seasonality, and specific pollution events are other important factors influencing LSS patterns. LSS monitoring showed relatively consistent longitudinal decreasing typologies in Na and K concentrations along the Anacostia watershed during spring and summer baseflow, as the river flowed through Anacostia National Park (Figure 7). In contrast, concentrations of Fe showed a pulse typology as the Anacostia River flowed through Anacostia National Park, which was likely due to Fe reduction due to extensive fringing wetlands (Figure 7). In a similar manner to the Anacostia River, there were also similar and consistent longitudinal patterns (a pulse typology) in concentrations of dissolved oxygen, Fe, and Mn along the Campus Creek flowpath, as water flowed through hydrologically connected floodplains and regenerative stormwater conveyance pools (Figure 9). In addition to the effects of winter road salt applications, longitudinal patterns may change with hydrologic conditions over time. Our analysis of relationships between longitudinal patterns in streamflow and elemental concentrations showed that downstream concentration-discharge relationships along flowpaths could be specific to certain chemical cocktails (Figure 5); these downstream concentration-discharge relationships are influenced by similar sources, flowpaths, and reactivity. For example, our previous work along the Gwynns Falls flowpath has shown longitudinal declines in concentrations and fluxes of chemical cocktails of metals and ions due to changes in downstream dilution and groundwater recharge patterns (Kaushal et al., 2018). Analyzing changes in longitudinal patterns in water quality within the same watersheds across varying streamflow conditions has the potential to reveal insights into the importance of hydrologic processes influencing longitudinal patterns (e.g., Figure 5 shows diverse longitudinal concentration-discharge relationships for chemicals influenced by attenuation, dilution, flushing, chemostasis from chronic groundwater inputs, and other processes).

LSS monitoring has considerable potential to reveal biogeochemical transformations and transition zones along flowpaths. We discuss some examples below.

Transformations of carbon and nutrients along flowpaths - Our results suggested coupled nutrient uptake and carbon accumulation along some urban watershed flowpaths. We observed downstream declining concentrations of nitrogen along flowpaths of the Gwynns Falls, Scotts Level Branch, Anacostia River mainstem, and Bull Run (e.g., Figures 4, 8, 13). There were also longitudinal changes in the quality and composition of organic matter along the Bull Run flowpath (Figure 12). Significant quantities of nitrogen can be retained and transformed via algal and microbial uptake and denitrification along watershed flowpaths (e.g., Marti et al., 2004; Haggard et al., 2005; Newcomer Johnson et al., 2014; Kaushal et al., 2014a; Ledford and Lautz, 2015; Jin et al., 2018). Nitrogen uptake along flowpaths is influenced by stream size, temperature, organic matter availability, stream velocity and turbulence, nitrogen concentrations, and other factors (e.g., Dodds et al., 2002; Hall and Tank, 2003; Wollheim et al., 2008; Grant et al., 2018; Loken et al., 2018). Along the Anacostia River watersheds, we observed coinciding increases in dissolved organic carbon and particulate organic carbon and/or increasing C:N ratios along the flowpath as nutrient concentrations declined longitudinally (Figure 8). Similarly, there was a decline in nitrogen concentrations along the Bull Run flowpath (Figure 10) and coinciding changes in organic matter quantity and quality, as revealed by fluorescence spectroscopy (Figure 12). Previous work has observed increases in DOC concentrations with increasing urbanization along the Gwynns Falls (Kaushal and Belt, 2012; Kaushal et al., 2014a) and other work has observed longitudinal increases in organic matter due to the importance of floodplain wetlands and bacterial density (Sabater et al., 1993). Previous work has also shown strong downstream linkages between C and N uptake and spiraling along forested and human-impacted streams (e.g., Bernhardt and Likens, 2002; Brookshire et al., 2005; Kaushal et al., 2014a; Plont et al., 2020).

We also observed increasing downstream trends in organic matter along the Gwynns Falls and Anacostia flowpaths (Figures 6, 8). As urban flowpaths are overloaded with organic matter and nutrients (Mallin et al., 2006; Blaszczak et al., 2019a; Blaszczak et al., 2019b; Carter et al., 2021), the capacity for stream metabolism of carbon and nitrogen can become saturated, thereby contributing to transport of excess organic matter and nutrients further downstream (Marti et al., 2004; Wollheim et al., 2008; Wollheim et al., 2018). Increased nutrient concentrations along human-dominated river continuums cause eutrophication and hypoxia (Diaz and Rosenberg, 2008; Rabalais et al., 2010). However, less work has considered production and accumulation of reactive organic matter along stream and river flowpaths as an additional cause of downstream hypoxia in coastal waters (Mallin et al., 2006; Blaszczak et al., 2019a; Blaszczak et al., 2019b; Carter et al., 2021). Oxidation of bioavailable organic matter can further contribute to biological oxygen demand, increase organic nutrient delivery and primary production, and contribute to harmful algal blooms (Rabalais et al., 2010). Impacts of urbanization on organic matter overloads along flowpaths warrant further research and consideration in water quality (e.g., Mallin et al., 2006, Kaushal and Belt, 2012, Kaushal et al., 2014b; Blaszczak et al., 2019a; Blaszczak et al., 2019b; Carter et al., 2021).

Transformations of greenhouse gasses along flowpaths - We observed longitudinal changes in concentrations in greenhouse gasses along the Gwynns Falls (decreasing and transition typologies) (Figure 6). Streams and rivers can be important sources of GHGs to the atmosphere due microbial processing of terrestrial inputs of organic matter and inorganic nitrogen into globally substantial quantities of CO₂ and N₂O to the atmosphere (Beaulieu et al., 2011; Raymond et al., 2013; Yao et al., 2020). Concentrations of CO₂ and N₂O declined with polynomial functions with increasing distance from headwaters along the Gwynns Falls (Figure 6). Biogeochemical processing of excess organic matter and nutrients along urban watershed flowpaths can significantly enhance production of GHGs (Smith et al., 2017; Jin et al., 2018). However, excess organic matter may also accumulate along urban watershed continuum flowpaths and become more labile and reactive (Kaushal and Belt, 2012; Kaushal et al., 2014a) when organic matter supply exceeds biological demand. Sewage inputs can also contribute to increased GHG concentrations along flowpaths (Jin et al., 2018), and sewage leaks are prevalent along the Gwynns Falls watershed (Kaushal et al., 2011). Concentrations of GHGs along the urban watershed continuum are influenced by other complex factors such as relationships of algal blooms, gas concentrations with discharge, gas exchange with the atmosphere, turbulence and seasonal hydrology (Harrison et al., 2005; Raymond et al., 2012; Grant et al., 2018). Overall, LSS monitoring of carbon, nutrients, and greenhouse gasses revealed coupled biogeochemical cycles along flowpaths and the potential for organic carbon overloads to downstream ecosystems (sensu Mallin et al., 2006; Blaszczak et al., 2019a; Blaszczak et al., 2019b; Carter et al., 2021).

Transformations of metals and salts along flowpaths - The potential for water quality transformations along flowpaths is not just limited to carbon and nutrients, but also includes transformations in forms and types of metal and salts through biogeochemical processes. Excess anthropogenic salt ion inputs shift the natural chemistry of freshwaters and pose emerging risks for drinking water supplies, aquatic ecosystems, and infrastructure both regionally and globally (e.g., Kaushal et al., 2005, Cañedo-Argüelles et al., 2013, Cañedo-Argüelles et al., 2016, Kaushal et al., 2020; Kaushal et al., 2022; Kaushal et al., 2023b; Bhide et al., 2021, Hintz et al., 2022b, Grant et al., 2022). We observed longitudinal variations in transport of both metals and salts along flowpaths. As expected, there were increasing trends in salts and metals with distance downstream along flowpaths draining progressively urban areas like the Gwynns Falls (Figure 4). These increasing concentrations were likely due to impervious surfaces and storm drains transporting metals and salts from pollution sources such as vehicles, tires, atmospheric deposition, road salts, sewage, etc. (Paul and Meyer, 2001; Kaushal et al., 2020; Morel et al., 2020) as watershed urbanization increased with distance downstream.

Salts and metals can increase or decrease along longitudinal flowpaths due to geochemical transformations, even over relatively small spatial scales along watersheds (Driscoll et al., 1988; Kaushal et al., 2022). Longitudinal patterns of conductivity (an indicator of salinity) and metals showed similar downstream decreasing patterns along Bull Run, Rock Creek, and Scotts Level Branch (Figure 10). Complexation of metals with salt ions, organic compounds, and inorganic materials can affect metal transport and transformation along flowpaths (Kaushal et al., 2020). Chloro-complexation between chloride ions and some metals could explain some similar patterns between conductivity and metals (Kaushal et al., 2019; Galella et al., 2021). For example, Cu can become soluble in response to increased salinity due to chloro-complexation (Kaushal et al., 2019). Ion exchange can also be important in explaining synoptic-scale patterns of salts and metals. Ion exchange retains ∼30–40% of incoming Na on exchange sites in sediments of Campus Creek (Kaushal et al., 2022). Sodium ions can mobilize other base cations and metal ions by directly competing for binding sites on sediment and ion exchange reactions (Kaushal et al., 2019; Kaushal et al., 2022). Thus, co-mobilization of salt ions and metals can occur along urban flowpaths (Kaushal et al., 2019; Kaushal et al., 2020; Galella et al., 2021; Kaushal et al., 2022; Kaushal et al., 2023b; Kaushal et al., 2023).

Principal components analysis of multiple elements along longitudinal flowpaths showed the potential to track fate and transport of pollution sources. For example, principal components analysis revealed a shift from labile protein-rich organic matter to more recalcitrant humic-like fractions of DOM as Bull Run flowed from a wastewater treatment plant through a forested regional park (Figure 12). Along the flowpath of Bull Run, principal components analysis also showed how the water quality signature of sewage effluent (enriched in K, Ca, S) from the UOSA wastewater treatment plant shifted to a natural groundwater signature enriched in Mg from geologic sources along downstream reaches in the forested regional park (Figure 13). These examples suggest LSS monitoring can be a powerful approach for tracking pulses and persistence of organic pollution inputs along streams (and also characterizing lag times in transformation of chemical mixtures). Future LSS approaches may also consider characterizing endmembers of different pollution sources, and then using endmembers to quantify contributions from different pollution sources along flowpaths using mixing models. Thus, LSS monitoring may provide a semi-quantitative or quantitative approach for identifying and tracking the fate and transport of different pollution sources along streams and rivers.

Overall, LSS monitoring revealed the potential for attenuation and dilution of multiple chemical cocktails from wastewater, urban stormwater, and road salt. For example, there were rapid rates of chemical attenuation directly downstream of the UOSA wastewater treatment plant along the Bull Run flowpath through forested regional parks (Figure 10). This attenuation affected nutrients, organic matter, salts, and metals. There was rapid attenuation likely due to dilution and transport distances that were chemical/element specific likely due to their propensity for biological uptake, chromatographic ion exchange in sediments and soils, sorption, sedimentation, storage, etc. along flowpaths (e.g., Kaushal et al., 2018; Parker et al., 2021). There were also rapid transitions in downstream concentrations of chemical cocktails, as streamwater flowed through restored stream-floodplain complexes along Scotts Level Branch (Figure 10). There was attenuation and dilution of salt ions during spring and summer months along the Anacostia River flowpath as the river flowed through Anacostia National Park (Figure 7). There were also inverse relationships between some chemical concentrations and increasing % forest and shrub cover at the watershed scale along Bull Run, Rock Creek, and Scotts Level Branch flowpaths (Figure 11).

These attenuation processes and transitions could have sometimes been due to dilution in stream reaches draining into conservation areas in parks, which may have minimal inputs of pollution compared to upstream. In addition, there could have been elemental transformations of nutrients and metals from dissolved to particulate forms due to biological uptake and transformation to biomass, changes in pH and solubility, and deposition of particulates out of the water column (Driscoll et al., 1988; Marti et al., 2004; Haggard et al., 2005). There could have also been transient storage or reactive uptake of nutrients, salts, metals and organic matter in hyporheic zones, floodplain soils, and hydrologically connected wetlands in conservation and restoration areas (Butturini and Sabater, 1999; Fuller and Harvey, 2000; Kaushal et al., 2022). In addition, increased availability of natural organic matter could also have contributed to enhanced attenuation of nutrients and metals. For example, we observed distinct changes in the quality of organic matter along the Bull Run flowpath using fluorescence spectroscopy (Figure 12). Organic matter can influence denitrification and complexation between organic materials and metals (Groffman et al., 2005; Mayer et al., 2010; Kaushal et al., 2018). Non-etheless, the presence of forest cover, conservation areas, and parks may have the potential to influence localized water quality along stream reaches (Figure 11).

Surprisingly, we also observed some capacity for attenuating and diluting salt ions (both processes can be important) along flowpaths in conservation and restoration zones (Maas et al., 2021) (Figures 4,7,9,10). Freshwater Salinization Syndrome (FSS) from anthropogenic sources is altering the natural chemistry of freshwaters on local, regional, and global scales. FSS poses emerging risks for drinking water supplies, aquatic ecosystems, and infrastructure globally (Kaushal et al., 2005; Cañedo-Argüelles et al., 2013; Cañedo-Argüelles et al., 2016; Dugan et al., 2017; Kaushal et al., 2018; Hintz and Relyea, 2019; Oswald et al., 2019; Kaushal et al., 2020; Hintz et al., 2022b; Grant et al., 2022; Kaushal et al., 2022). Increased concentrations of salts at many of our sites were in ranges that could impair ecosystems and mobilize metals and other contaminants (Kaushal et al., 2019; Galella et al., 2021). There were decreasing concentrations of salt ions along Bull Run with increasing distance downstream from urban areas and wastewater treatment plant discharges (Figure 10). In addition, there was a plateau in concentrations of salt ions along Rock Creek, as the stream flowed from progressively urban areas into forested reaches of Rock Creek National Park (Figure 10). The presence of forest cover, conservation areas, and parks at the watershed scale may have the potential to attenuate salt pollution (Figure 11), but more work is necessary to assess limitations (Maas, 2022).