https://orcid.org/0000-0002-9999-1021
Figures
Abstract
Human modification of water and nutrient flows has resulted in widespread degradation of aquatic ecosystems. The resulting global water crisis causes millions of deaths and trillions of USD in economic damages annually. Semiarid regions have been disproportionately affected because of high relative water demand and pollution. Many proven water management strategies are not fully implemented, partially because of a lack of public engagement with freshwater ecosystems. In this context, we organized a large citizen science initiative to quantify nutrient status and cultivate connection in the semiarid watershed of Utah Lake (USA). Working with community members, we collected samples from ~200 locations throughout the 7,640 km2 watershed on a single day in the spring, summer, and fall of 2018. We calculated ecohydrological metrics for nutrients, major ions, and carbon. For most solutes, concentration and leverage (influence on flux) were highest in lowland reaches draining directly to the lake, coincident with urban and agricultural sources. Solute sources were relatively persistent through time for most parameters despite substantial hydrological variation. Carbon, nitrogen, and phosphorus species showed critical source area behavior, with 10–17% of the sites accounting for most of the flux. Unlike temperate watersheds, where spatial variability often decreases with watershed size, longitudinal variability showed an hourglass shape: high variability among headwaters, low variability in mid-order reaches, and high variability in tailwaters. This unexpected pattern was attributable to the distribution of human activity and hydrological complexity associated with return flows, losing river reaches, and diversions in the tailwaters. We conclude that participatory science has great potential to reveal ecohydrological patterns and rehabilitate individual and community relationships with local ecosystems. In this way, such projects represent an opportunity to both understand and improve water quality in diverse socioecological contexts.
Citation: Jones EF, Frei RJ, Lee RM, Maxwell JD, Shoemaker R, Follett AP, et al. (2021) Citizen science reveals unexpected solute patterns in semiarid river networks. PLoS ONE 16(8): e0255411. https://doi.org/10.1371/journal.pone.0255411
Editor: Andreas C. Bryhn, Swedish University of Agricultural Science, SWEDEN
Received: March 1, 2021; Accepted: July 15, 2021; Published: August 19, 2021
This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Data Availability: All data and code used are available at doi.org/10.4211/hs.85f3584dccb54afba5698ac615ff949a.
Funding: This work was supported by the Utah Division of Wildlife Resources through the Watershed Restoration Initiative program and National Science Foundation (award numbers EAR-2011439 and EAR-2012123) to BWA. Additional support was provided to BWA, ZTA, and NCH by the Sant Family Foundation through the Roger and Victoria Sant Endowment and the Provo River Watershed Council. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Agriculture, wastewater, and fossil fuel use have exceeded global thresholds for nitrogen (N) and phosphorus (P) [1–4], resulting in eutrophication of two-thirds of freshwater ecosystems globally [5–8]. Excess nutrients and other water pollutants such as heavy metals and waste from humans and livestock degrade aquatic ecosystem integrity, leading to trillions of USD in economic damages and the premature death of approximately 2 million people annually [9–12]. Mitigating these global water crises will require improved monitoring and management, which themselves depend on public understanding and financial support [13–16]. Consequently, improving public understanding and interaction with aquatic ecosystems is a planetary priority [17–19].
Because of the high spatiotemporal variability typical of both surface and subsurface aquatic ecosystems [20–22], identifying pollutant sources requires repeated sampling of many locations in the watershed [23–25]. This type of synoptic sampling provides a high-resolution view of water chemistry throughout the network, potentially generating insights into hydrological and biogeochemical processes [26–28], as well as identifying the location and spatial extent of pollutant sources [29–31]. Pollutant sources range in size from discrete point sources, such as a wastewater discharge, to diffuse nonpoint sources, such as runoff from agricultural fields [32, 33]. They also vary in duration, from persistent sources that are always active to intermittent sources that only deliver pollutants to the ground or surface water during certain ecohydrological conditions [34–36].
While measuring or modeling water chemistry continually everywhere in the stream network remains impossible, a suite of ecohydrological metrics have been developed to inform management based on repeated synoptic sampling [1, 23, 27, 29, 37]. For example, the relationship between spatial variability and watershed size can reveal the patch size of processes driving water chemistry, revealing the relative importance of delivery from terrestrial environments and processing in aquatic environments [26, 29, 38, 39]. Likewise, the persistence of spatial patterns in water chemistry through time can reveal changes in pollutant sources or sinks, informing the necessary sampling frequency [23, 27, 40, 41]. In practice, high spatial persistence of a solute in a watershed allows for a single synoptic sampling event to representatively characterize its dynamics [23, 29, 40]. While synoptic sampling requires substantial resources and coordination, combining such metrics with traditional analysis of watershed land use and land cover could improve effectiveness of monitoring and restoration measures. So far, these analyses have primarily been done in temperate and high-latitude ecosystems [23, 27, 38, 40], leaving important unknowns about other ecosystem types.
In semiarid and arid regions, the degradation of aquatic ecosystems has been particularly extreme because of high water demand and low water availability for green, blue, and gray water use [17, 34, 42–44]. This combination often results in intense hydrological and chemical disruption of surface and groundwater in semiarid and arid watersheds, which are often endorheic [17, 42, 43, 45]. In addition to anthropogenic pressure, aquatic ecosystems in semiarid regions are naturally dynamic because of extreme hydrological and biogeochemical variability [46–48]. Changes in precipitation and evapotranspiration result in large expansions and contractions of the surface water network, and wetting and drying cycles create heterogeneous biogeochemical conditions [49–51]. This combination of human and natural variability could create nutrient source and sink patterns that are distinct from temperate regions, potentially complicating the identification and mitigation of nutrient sources in heavily impacted semiarid watersheds.
In this context, we organized a series of participatory synoptic sampling events in the Utah Lake watershed in the western US. We used a citizen science approach for two reasons. First, waterbodies in this region are experiencing eutrophication and water overallocation [43, 52–54], partly because of a lack of public connection with local aquatic ecosystems [55]. For example, this disconnect has led to a lack of public will to implement wastewater treatment measures that could decrease delivery of bioavailable nutrients to Utah Lake [56, 57] and even the consideration of a radical reengineering of the lake, including large artificial islands [58, 59]. Second, the Utah Lake watershed is nearly 8,000 km2, making traditional synoptic sampling impractical and expensive. By collaborating with local nonscientists, we were able to sample nearly 200 locations throughout the watershed within a few hours, reducing variability from temporal changes in water flow and chemistry [27, 60]. We focused on three major questions: 1. What are the primary sources of carbon and nutrients in the Utah Lake Watershed, 2. How much solute retention or release is there in the surface water network, and 3. What are the general patterns of solute chemistry in these semiarid river networks. Though we did not collect quantitative data on the attitudinal effects of participating in the samplings, we hypothesized that learning about and spending time in the diverse tributaries to Utah Lake would improve public awareness and proclivity to address local environmental issues [16, 18, 61, 62]. Regarding the ecohydrological issues surrounding Utah Lake, we hypothesized that the complex human footprint and variable hydrology in this semiarid watershed would create low spatial persistence of pollutant sources and high critical source area behavior with a few influential areas disproportionately influencing water quality [29, 31, 33]. To test these hypotheses, we analyzed volunteer-collected samples for a broad suite of physicochemical parameters, calculating the ecohydrological metrics we describe below.
Methods
The Utah Lake watershed
Utah Lake is one of the largest freshwater lakes in western North America, with a surface area of 375 km2 and a natural watershed area of 7,640 km2 [54]. As a part of the Great Salt Lake watershed, Utah Lake is a remnant of Lake Bonneville, which covered up to 51,000 km2 of what is now Utah, Nevada, and Idaho until about 15,000 years ago [63–65]. The watershed ranges from 1,368 to 3,586 MASL and is characterized by relatively pristine high-elevation headwaters in the Wasatch Mountains—although mining, livestock grazing, rural subdivisions, and ski resorts are present. Low elevation valleys have a mix of urban and suburban development interspersed with irrigated agricultural land, with an overall human footprint of up to 93% in the valley bottom. Utah has one of the fastest growing populations in the US [66], and agricultural land is increasingly being converted for suburban development. Seven wastewater treatment plants, serving approximately 600,000 people in the valley region, discharge treated effluent into tributaries to the lake (S1 Fig).
In the watershed headwaters, the dominant geology is limestone and quartz [67]. The vegetation consists of mixed aspen, conifer, and maple forests at high elevations, transitioning to scrub oak and sagebrush at lower elevations. The high-elevation hydrology is complex, including gaining reaches (i.e., net flow of subsurface water into surface water flows) and losing reaches (i.e., net flow into subsurface water), particularly in areas of karst conduits and colluvial materials [68]. At the base of the mountains, where rivers flow into Lake Bonneville sediment deposits at the outer perimeter of the prehistoric lake, reaches become primarily losing, and water transport occurs largely through shallow groundwater flowpaths [69]. Near the lake, streams once again become gaining, mixing with new streams generated by natural and artificial recharge that feeds many springs flowing into the lake [70, 71].
Sampling design
We classified the subwatersheds into one of four categories based on land use and hydrologic condition (Table 1): Agricultural unregulated (Spanish Fork River), Mixed dammed (Provo River), Mountain urban (American Fork River and Hobble Creek), and Valley tributaries (Benjamin Slough, Goshen Valley, Mill Race, and others). The Agricultural unregulated subwatersheds have been slower to experience rapid population growth and still remain mostly under agricultural uses. One subwatershed (Diamond Fork) receives artificially diverted flow from the nearby Strawberry River, but for most of the Agricultural unregulated subwatersheds, the in-stream hydrologic modifications are minimal. All the categories have some degree of flow infrastructure (e.g., check dams, channelized reaches, dikes, etc.), but Mixed dammed subwatersheds include two large reservoirs (capacities of 0.18 and 0.39 km3, respectively) that drastically alter the downstream hydrology. Land use in Mixed dammed subwatersheds is a mixture of both agriculture and urban. Mountain urban subwatersheds include high contrast land use and degree of impact, with mostly pristine headwaters before entering highly developed land use and consequently modified stream channels at low elevations. Valley tributaries include the groundwater sourced tributaries that originate near the lake as described above. These subwatersheds also include return flow from agricultural water diversions, and effluent from wastewater treatment plants and other industrial facilities. The Utah Lake watershed also includes the West desert region (Cedar Valley, Tickville Gulch, and Lake Mountain), which has only small, remote ephemeral springs that are largely inaccessible. These subwatersheds were excluded from sampling because they are not connected to the lake via surface water.
We initially selected 500 sampling sites from the Ambient Water Quality Monitoring System (AWQMS), a database curated by the Utah Division of Water Quality, and the Utah State University Water Quality Extension citizen science program, Utah Water Watch. We used a clustering technique by including sites just above and below a confluence to maximize watershed coverage and minimize travel distances. We consolidated the initial 500 sites to 270 by merging redundant locations and removing inaccessible sampling locations (e.g., difficult terrain or private ownership) and sites with no surface flow even during snowmelt. Nearly all remaining sites were publicly accessible, and for the few locations on private property, we obtained verbal permission from landowners. We used the application USGS StreamStats to delineate watersheds, calculate watershed areas (km2), and estimate land use and land cover from the National Land Cover Database (NLCD) for 1992 and 2011. Land use was classified as forested, developed, impervious surface, or herbaceous upland for each watershed.
Participatory science
The practice of involving nonprofessionals in research (i.e., participatory or citizen science) has been extensively used in the natural and social sciences [16, 72, 73]. While there are tradeoffs to participatory approaches, including less control over data acquisition and costs for participants, there can be substantial community and scientific benefits [73–75]. On the community side, citizen science can create educational opportunities and foster trust between researchers, regulators, policymakers, and the public [74, 76–78]. On the scientific side, citizen science can provide opportunities and value for researchers, including enabling novel experimental designs (e.g., the collection of hundreds of samples synchronously) and informing research priorities by improving researchers’ awareness of local needs and policy priorities [10, 75, 76, 79].
We collaborated with an undergraduate course on watershed ecology to develop a multi-year participatory science program. This program included 2 public lectures, 7 community events (e.g., fairs, festivals, and educational evenings) where we made presentations or staffed interactive booths, and 6 synoptic samplings. From the beginning of the project, we partnered with existing water organizations, including the Provo River Watershed Council, Utah Water Watch, the Utah Lake Commission, the Utah Division of Water Quality, and the Utah Valley Earth Forum. These partnerships were pivotal in recruiting volunteers, designing the samplings, and disseminating the results. We additionally advertised through social media (Facebook, Twitter, and Instagram), email lists (including past participants), online videos, and paper fliers for approximately one month before each of the six sampling events. These advertisements targeted local university students, youth groups, environmental groups, and the broader community. At the community events, we presented a model watershed (EnviroScape, Herndon, VA) to demonstrate runoff and transport of pollutants represented by food coloring. In all our interactions with the public, we emphasized the historical context of human-water interactions in the watershed, the current ecological status of the lake system, and potential policy and personal actions to improve the health of the lake.
Sampling events
The synoptic sampling events were the central experiences for volunteers in the program. We carried out six sampling events in 2018 and 2019, but due to COVID-19 and other delays, the chemical analyses of the last three samplings have not yet been finalized. We report the results of the 2018 March (Spring), July (Summer), and October (Fall) samplings in this paper. Volunteers were invited to sign up in advance for the sites they wanted to sample on the project website (https://utahlakecollab.wixsite.com/utahlakecollab), choosing from an interactive online map of the 270 locations. This planning process encouraged volunteers to explore the entire watershed virtually before the event, looking at areas where they already had experiences and imagining locations they had not yet visited.
To reduce variability from sampling error, we used careful training, simplification of procedures, and replication of sampling (i.e., we asked multiple volunteers to collect samples from the same site so we could quantify sampling error). Other citizen science studies have found that when such precautions are taken, the benefit of using volunteers to collect large numbers of samples outweighs the tradeoffs of this kind of approach [80–82]. On the days of the samplings, we distributed informational flyers while volunteers waited to collect sampling kits and directions to sites. The flyers described the sampling procedure and provided context about water use and water pollution in the area (S2 Fig). Before distributing the sampling kits, we trained each volunteer individually, showing them how to collect water samples safely and reproducibly. We provided simple field sheets, where participants recorded site conditions and bottle numbers. After completing the sampling, participants returned their samples and datasheets to the distribution locations, where we performed quality control, noting samples with incomplete data or other irregularities (e.g., broken filters, partially filled bottles).
Because this research involved nonprofessional volunteers, we consulted the Institutional Review Board (IRB) at Brigham Young University (BYU), which oversees all research involving human subjects. We were informed that IRB approval was not needed because volunteers were not the subject of the research (i.e., we did not collect information about their identities or experiences). This limited our ability to quantitatively assess the community outcomes of the research but made participation less burdensome and improved inclusivity, particularly for members of the community unwilling to share personal information for philosophical or political reasons.
Laboratory analyses
Samples were filtered in the field with pre-rinsed 0.45 μm cellulose acetate filters into 60 ml high-density polyethylene bottles and immediately frozen or refrigerated until analysis (typically within 2 weeks of sampling). Anions (NO3-, NO2-, SO42-, Cl-, and PO43-) and cations (NH4+) were quantified by ion chromatography (Dionex Thermofisher HPIC). Soluble reactive phosphorus was quantified colorimetrically using the ascorbic acid method [83], reported hereafter as PO43- for simplicity. Dissolved organic carbon (DOC) and total dissolved nitrogen (TDN) were quantified using a C/N auto-analyzer (Elementar, Langenselbold, Germany). Dissolved inorganic nitrogen (DIN) was calculated as the sum of NO3-, NO2-, and NH4+ from the ion chromatography.
Ecohydrological metrics and statistical analyses
We calculated three recently developed ecohydrological metrics that indicate solute dynamics: spatial variance thresholds, subwatershed leverage, and spatial persistence [27, 29]. Spatial variance thresholds indicate the predominant spatial scale where solute delivery or removal is occurring, analogous to the representative elemental area concept [84, 85]. For each of the main tributaries (Table 1), we plotted the scaled (subtracted the mean and divided by the standard deviation) concentration for each solute against subwatershed area. Changes in variance with spatial scale can be caused by mixing of tributaries, in-stream processing, and changes in terrestrial-aquatic linkages. We used the pruned exact linear time (PELT) method to test for changes in variance, implemented in the changepoint package of R [86]. We hypothesized that the variance would decrease with spatial scale, following observations from other river networks [29, 38].
We calculated subwatershed leverage by multiplying the difference in subwatershed and watershed outlet nutrient concentrations with the ratio of subwatershed area to watershed outlet area [27, 29]. Assuming similar specific discharge throughout the watershed, leverage indicates the amount of flux at the outlet that can be explained by the contribution of each subwatershed [29, 38]. Positive leverage indicates the watershed is a net sources of the given solute and negative leverage indicates the watershed is a net sink. Very large leverage values can occur—for example, well over 100%—because some material is retained or removed as the water flows over and through the landscape before reaching the outlet. Thus, sites with large leverage values (i.e., >100%) can be considered highly influential contributors to solute concentrations at the outlet. Recently, mean leverage values across subwatersheds have been used to infer network-scale production or retention of solutes [27]. When the watershed leverage mean is positive, this implies there has been nutrient removal within the surface water network (i.e., there is more solute in the tributaries than can be accounted for at the outlet), whereas a mean negative leverage value indicates production within the surface water network. Valley tributaries subwatersheds were excluded from leverage analyses because they did not converge within a network to a single outlet (i.e., many discharged directly into Utah Lake).
To assess the consistency of observed spatial patterns, we calculated spatial persistence with Spearman’s rank correlation [23, 29]. This analysis calculates a correlation coefficient (ρ) of solute concentrations ranked from highest to lowest for each category and solute, across each pair of Seasons (i.e., Spring-Summer, Summer-Fall, and Spring-Fall). We graphed means and ranges of ρ values and tested for the effects of Season and land use category on solute persistence using ANOVA.
We also used parametric statistics to compare links between land use and land cover with water chemistry for the main tributaries. For each solute, we tested for differences among watersheds with Analysis of Variance (ANOVA). We tested for links between land use and catchments characteristics with water chemistry using generalized least squares models. The models used concentration of each select solute (PO43-, DIN, TDN, DOC, Cl-, and SO42-), including season and land use (e.g., developed, impervious, forest, and herbaceous) as independent variables. A second order Akaike Information Criterion (AICc) was calculated for each possible subset model, then the subset model with the smallest AICc score was selected as the final model (reported below).
We performed all statistical analyses in R, using the dplyr and ggplot2 packages [87–89]. We used ArcGIS Pro (ESRI) to map solute concentrations, using open-source base layers. Sample concentrations, including geographic information, and code used in analysis are available at https://doi.org/10.4211/hs.85f3584dccb54afba5698ac615ff949a.
Results
In 2018 and 2019, our outreach efforts reached approximately 6,500 community members directly. Across the first three sampling events (the ones reported in this paper), we had over 150 unique participants, with at least a third of participants attending more than one event. Most participants sampled in small groups although some worked alone. Detailed demographic data was not collected for participants in this phase of the project, however most attendees were estimated to be between 18 to 30 years old, with ages ranging from less than 1 over 80 years old. When informally asked about their experience, responses were overwhelmingly positive, including statements like, “I had no idea there were so many beautiful streams in Utah Valley,” and, “this opened my eyes to how much we depend on this water. It actually comes from somewhere before my sink!”
Most sampling kits were used correctly (e.g., bottles filled with filtered sample), with <10% of the samples rejected because of user error. Incomplete or illegible labeling was the most common error, though there were a few instances of complete communication breakdown (e.g., one bottle was returned filled with soil).
Solute concentrations
The spatial distribution of solute concentrations across the watershed is shown in Fig 1. This map includes data for samples collected from Utah Lake, which were excluded from other analyses. Point color corresponds with Utah’s numeric water quality standards for N and P (N > 4 mg L-1 and P > 0.05 mg L-1) and the 25th and 75th percentile of measured concentrations for other parameters. DOC concentration varied relatively evenly across the watershed, while DIN, TDN, PO43-, SO42-, and Cl- concentrations were highest at sites near or on Utah Lake (Tukey, p-adj.<0.001).
Samples were collected from the watershed (outlined in black) and averaged across three citizen science synoptic sampling events. Point color represents numeric water quality standards for N and P values, or 25th and 75th percentile for others. Point size is scaled to concentration. Basemap source: USGS National Map and Earth Resources Observation and Science Center.
Solute concentrations were different across the three sampling seasons (Spring, Summer, and Fall) through the year (ANOVA, F-stat = 5.896, p-value<0.01) and across categories (ANOVA, F-stat = 96.055, p-value<0.0001) for all solutes (Fig 2 and S1 Table). Pairwise analysis determined that solute concentrations in Spring were higher than Summer or Fall (Tukey, p-adj. = 0.002), but also depended on land use category (ANOVA, F-stat = 4.34, p-value<0.001). A pairwise analysis across all categories determined that DOC concentration was higher in Summer than Fall or Spring (Tukey, p-adj<0.001), and TDN concentration was higher in Fall than during Summer or Spring (Tukey, p-adj<0.05). All solute concentrations were higher at Valley tributaries sites than the other three categories (p-adj.<0.001), and Mountain urban were higher than Mixed dammed (p-adj = 0.005).
Samples were collected during synoptic sampling events in three seasons (SPR = Spring, SUM = Summer, FAL = Fall) in watersheds with different land use categories (Agricultural unregulated, Mixed dammed, Mountain urban, and Valley tributaries). Boxplots represent the 25th, 50th, and 75th percentiles, points within 1.5 times the interquartile range, and points beyond. The notches represent the 95% confidence interval of the median (non-overlapping notches suggest statistically significant differences between populations).
Regressions of solute concentration by Season and land use found that % impervious surface was positively correlated with higher concentrations for all solutes except DOC (Table 2). In regression models of TDN, DOC, and SO42-, backward exclusion criteria found that only Season was significant. In regression models of SO42- and Cl-, backward exclusion criteria found that only % herbaceous upland was significant. Correlation coefficients (R2) for the models were between 0.11 and 0.25.
Leverage and spatial persistence
In general, scaled concentrations and leverage by watershed area for the different solutes did not show a funnel shape that is typical of humid and temperate watersheds (Figs 3 and 4A) [38]. Instead, many of the solutes exhibited hourglass shapes, with higher scaled concentration and leverage at the largest subwatershed size. However, the funnel shape did occur for DIN concentrations in the valley tributaries, which had particularly high leverage in the smaller catchments (<100 km2). Variance thresholds (km2), calculated from the scaled concentration data using the PELT method, are listed in S3 Table. Only two solutes (DOC and TDN) were calculated to have a single changepoint: TDN, 0.78 km2, and DOC, 0.62 km2 (although DOC concentrations showed no variance collapse). PELT calculated two variance thresholds for both Cl- (3.69, 666.00) and SO42- (62.6, 666.0), each having one changepoint closer to the headwaters and one nearer to the lake. DIN had three variance thresholds, all of which were relatively small subwatershed sizes (0.78, 14.3, 18.0). Six changepoints were calculated for PO43- (0.23, 10.3, 10.9, 14.3, 18.0, 644), with five out of the six thresholds being found at small subwatershed sizes. Occasional outliers (single points at mid-range and large watershed size) may have had an oversized effect on the overall pattern.
Theoretical diagram of spatial variability collapse (left) and spatial variability pattern observed in Utah Lake watershed (right).
Scaled solute concentration (a, left) and leverage (b, right) by watershed area for solutes of interest (DOC, PO43-, TDN, DIN, Cl-, and SO42-) for sites within four land use and hydrologic categories of the Utah Lake watershed. Samples were collected on synoptic sampling events conducted in three seasons: Spring, Summer, and Fall of 2018. Horizontal lines represent the means of the raw concentration data for that particular solute; vertical lines represent change points detected by PELT analysis.
Most subwatersheds exhibited moderate to low leverage (i.e., < ±25% leverage) on watershed outflow concentrations of Cl-, and SO42- (85, and 84% of watersheds respectively; Figs 4B and 5 and S2 Table). DOC, N, and P dynamics were much more concentrated (i.e., showing a critical source area behavior), with around two-thirds (59–69%) of watersheds having a moderate to low effect on outflow concentrations and the remaining third controlling flux. There were many more highly influential subwatersheds (i.e., >100% leverage) for TDN and PO43- than for DOC, Cl-, and SO42-, in line with the expected pattern from discrete sources of N and P in the watershed. Unexpectedly, there were very few highly influential subwatersheds for DIN concentrations, potentially due to the very high relative flux of DIN at the watershed outlet (i.e., large loads in subwatersheds still remain <100%). The number of subwatersheds that were highly influential (i.e., >100% leverage) ranged from 11–16% for TDN and PO43-, 8% for DOC, and only 1–2% for DIN, Cl-, and SO42-.
Diamonds represent the mean leverage value for each subwatershed category. The horizontal black line at y = 0 represents a neutral mass balance. Diamonds that are above the black line indicate solute production and diamonds below the black line indicate solute removal within the surface water network.
Mean leverage values for each solute and subwatershed category further illustrated the dynamic behavior of N and P in the spring and summer in comparison with DOC, Cl-, and SO42- (Fig 5). For example, TDN exhibited a strong removal signal (mean leverage >0) for all categories in the spring and summer, with means closer to 0 indicating a conservative transport (0 net production or removal) in the fall. DIN exhibited a weak production signal in all categories and seasons. PO43- varied considerably by category and season. In contrast, Cl- and SO42- maintained consistent neutral mass balances across category and season. DOC switched from a net source to a sink from spring to summer in the Unregulated agricultural subwatersheds, but remained neutral for the Mountain urban subwatersheds, and maintained a mostly consistent sink capacity in the Mixed dammed subwatersheds.
Spatial persistence was specific to solute and land use (Fig 6 and S3 Table). DOC and PO43- had lower spatial persistence (0–0.5) than the other solutes (0.3–1). Cl- and SO42- had the highest spatial persistence (>0.7). DIN and TDN had intermediate levels of persistence (0.5–0.7). Persistence was highest overall at Mountain urban and lowest in Valley tributaries and Agricultural unregulated subwatersheds, although the order of persistence as dependent on solute (ANOVA, F-stat = 3.514, df = 30, p-value<0.001).
Spatial persistence is calculated as a pairwise Spearman rank correlation coefficient (ρ) among synoptic samplings from three seasons (Spring, Summer, and Fall). Points represent mean ρ for all pairwise comparisons within each category, and bars give the range of values. The horizontal grey line marks ρ = 0.7; indicating that the majority of the spatial pattern was retained between the two samplings.
Discussion
Novel spatial hydrochemical patterns
Our study emphasizes the unique hydrochemistry of semiarid and mixed natural-urban regions. In our results, PO43- was less spatially stable than other major ions, which was also the case in temperate agricultural watersheds [23, 29] and in Arctic tundra watersheds [27]. The low spatial persistence of PO43- compared with other solutes like Cl- suggests two non-exclusive factors: one, sources other than natural geologic weathering introduce variability and two, biological processes involving P uptake are variable. For example, changes in hydrological connectivity across seasons and storm events can alter the delivery and chemical availability of PO43- and other nutrients [90–93]. Likewise, P is often a limiting nutrient for some subwatersheds, introducing another source of variability as biological demand may exceed supply [36, 94, 95].
Valley tributaries stood out as having particularly low spatial persistence across all solutes (Fig 6). We hypothesize that this is at least partially a function of the natural difference in solute concentrations between dilute snowmelt stream water and solute-rich groundwater, combined with changes in source based on hydrologic conditions [69]. When discharge is high, stream water diluted with snowmelt flows farther into the valleys before the change from losing to gaining occurs. During low flow, the switching point of losing to gaining moves upstream, creating large variability in solute concentrations observed at valley sites between the mouth of the canyons and the lake.
Our analysis did not return a single spatial threshold for variance collapse in solute concentration, unlike in other studies [50, 96–98]. For example, Northern Boreal watersheds tend to have variance collapse in DOC at 15 km2 [40]. In Arctic tundra, the threshold for DOC and NO3− is similar (10–20 km2), and PO43- is slightly larger (25 km2) [27]. Mined Kentucky headwaters have variance collapse in major anion and cation concentrations between 15 and 75 km2 [99]. DOC, PO43-, TDN, and DIN all had threshold values at very small subwatershed areas (<1 km2). These low thresholds are likely influenced by point sources, such as wastewater effluent and field drains, which we reported as subwatershed areas of 0 km2. PO43-, Cl-, and SO42- had thresholds at small and large subwatershed sizes. The hourglass pattern in spatial variance may be also be impacted by the fact that the average subwatershed areas for the Valley tributaries and Mountain urban were much smaller (40 and 32 km2, respectively) than the Mixed dammed and Agriculture unregulated (200 and 281 km2, respectively). The lack of variance collapse in our study, like the low spatial persistence, may have been due to increased solute concentrations at groundwater-influenced sites near the lake [69]. Alternatively or additionally, we may have detected differences in spatial variance at finer scales because of the high-resolution sampling (i.e., such variability could exist in other areas but not have been detected because of coarser spatial sampling).
Urban sources affect all solutes except DOC
This study identified hot spots of solutes of concern (Fig 1), and determined that position within watershed is important in determining concentration dynamics [100]. PO43-, TDN, DIN, Cl-, and SO42- concentrations were highest in Valley tributaries and other low-elevation reaches, and solute removal or dilution occurred at mid-elevation reaches, indicated by the map of concentrations (Fig 1) and narrowing of solute concentrations in mid-range subwatershed sizes (Fig 4A). Impervious surfaces contributed to overall higher solute concentrations (Table 2), but solute concentrations were more variable in reaches with agricultural activity (Fig 6). This could be due to both the direct effect of impermeability on nutrient export (e.g., stormwater drainage) as well as a correlation between percent impermeability and the presence of wastewater effluent [34, 101]. We note that our linear models had low predictive capability (up to 25%), suggesting the need for additional explanatory metrics (e.g., geology and updated land cover data). Decreases in solute concentrations at sites in the valley could be due to losses to groundwater [69] or sorption of P to Lake Bonneville sediments [54].
Valley tributaries were significant sources of N, a majority of which was DIN (Figs 2 and 4A). Multiple linear regression analyses showed that models of TDN and DIN had higher correlation with % impervious surface than any of the other solutes (Table 2), and both categories of N had lower persistence across Valley tributaries than the other subwatershed categories. Valley tributaries, which had the highest percentage of developed land use (Table 1), had the highest variability in N concentrations. Biological activity could also be responsible for high variability and seasonal differences in N concentrations, similar to what was observed in other mountainous western US watersheds [102].
This study confirms that urban point sources may disrupt spatial variability collapse. For example, in highly urbanized watersheds in New York, NO3− decreased in variability with increased watershed size, but not SRP or NH4 + [94]. We found DIN increased below wastewater treatment plants in Mixed dammed and Mountain urban watersheds, although in Mixed dammed subwatersheds, DIN is subsequently diluted and/or removed at downstream sites (Fig 1). PO43- concentration exceeded State of Utah numeric criteria for nutrient pollution throughout all subwatershed categories and sizes, suggesting that point and non-point sources (e.g., stormwater, agricultural water, and natural geological deposits) contribute to elevated P in local streams.
Citizen science
In this study, we provided the opportunity for thousands of local citizens to learn more about point and non-point sources of water pollution in a deep and meaningful way [18]. This engagement has the possibility of creating public support for efforts to address water quality in the Utah Lake watershed [103]. Future directions of this work include using educational research tools to quantify the impact of participation on knowledge, attitude, and behavior. This study demonstrates that citizen scientists can help professional researchers accomplish study methodologies that are otherwise prohibitive. This has the dual benefits of extending capacity for scientific observation, and fundamentally changing public awareness and mentality. Both benefits subsequently influence how water resources are managed [61]. In this sense, participatory water quality monitoring is not only a means of increasing understanding of how water and nutrients propagate through watersheds; it is a mechanism to improve water quality itself and encourage sustainable stewardship [79, 80].
Conclusion
Our results demonstrate the high spatial and temporal variability of PO43- within this watershed. However, at intermediate subwatershed sizes, PO43- removal or dilution occurred (Fig 4A). Point sources and groundwater around the lake contributed N in the form of DIN, and, like PO43-, decreased in concentration in mid-range subwatershed size. In addition to high solute concentrations, Valley tributaries had low spatial persistence, indicating temporally dynamic changes in sources and sinks of solutes. Even though there are inputs from natural (e.g, geology) and anthropogenic (e.g., mining and grazing) sources that contribute to variability in solute concentrations in the headwaters, the variability decreases at intermediate watershed sizes as these diverse headwaters mix. It is at the highly impacted, urban reaches (Valley tributaries and subwatershed sites >500 km2 that normal solute behavior is affected to the degree that variability unexpectedly increases.
The Utah Lake watershed is fundamentally different in network-scale hydrochemistry than previously described temperate, urban, and Arctic watersheds, due to the unique hydrology and human impacts, especially in the lower reaches of the watershed. We encourage including more semiarid regions, specifically endorheic basins, in hydrologic studies because understanding their distinctive hydrologic characteristics is critical to preserving these unique ecosystems, many of which are threatened by climate change and human development [43].
Supporting information
S1 Fig. Map of Utah Lake watershed sites synoptically sampled.
Points colored by land use and hydrologic modification category (red = Agricultural unregulated, green = Valley tributaries, blue = Mixed dammed, purple = Mountain urban). Yellow triangles represent wastewater treatment plants. Basemap source: USGS National Map and OpenStreetMap.
https://doi.org/10.1371/journal.pone.0255411.s001
(TIF)
S2 Fig. Documents provided to citizen science synoptic sampling participants.
Includes sampling instructions and datasheet for recording sample information. Instruction sheet shows photo of co-author GML demonstrating sampling technique.
https://doi.org/10.1371/journal.pone.0255411.s002
(DOCX)
S1 Table. ANOVA test comparing solute concentrations in streams from different land use categories over three synoptic sampling events in the Utah Lake watershed.
https://doi.org/10.1371/journal.pone.0255411.s003
(DOCX)
S2 Table. Distribution of leverage values in each subwatershed for each solute.
https://doi.org/10.1371/journal.pone.0255411.s004
(DOCX)
S3 Table. ANOVA test comparing spatial persistence values for solutes measured in different land use categories over three synoptic sampling events in the Utah Lake watershed.
Asterisks denote significant p-values: ’***’ < 0.001; ’**’ <0.01; ’*’ < 0.05; ’.’ < 0.1.
https://doi.org/10.1371/journal.pone.0255411.s005
(DOCX)
Acknowledgments
We would like to thank Greg Carling, Paul Frandsen, Courtney Fulton, Blair Hansen, Jordynn Scheuller, Hannah Scow, Lindsey Steinhorst, Jesse Morris, and Julianne Capito for contributing to the initial project design; Brian Brown and Isaac St Clair for assisting with data compilation; and the students of PWS 306 in 2019 who created the watershed delineations in USGS StreamStats. We recognize and thank all the participants and volunteers throughout 2018 and 2019, without whom this project would not have been possible.
References
- 1. Frei RJ, Abbott BW, Dupas R, Gu S, Gruau G, Thomas Z, et al. Predicting Nutrient Incontinence in the Anthropocene at Watershed Scales. Front Environ Sci. 2020;7.
- 2. Lade SJ, Steffen W, de Vries W, Carpenter SR, Donges JF, Gerten D, et al. Human impacts on planetary boundaries amplified by Earth system interactions. Nat Sustain. 2019 [cited 19 Dec 2019].
- 3. Le Moal M, Gascuel-Odoux C, Ménesguen A, Souchon Y, Étrillard C, Levain A, et al. Eutrophication: A new wine in an old bottle? Science of The Total Environment. 2019;651: 1–11. pmid:30223216
- 4. Yan Z, Han W, Peñuelas J, Sardans J, Elser JJ, Du E, et al. Phosphorus accumulates faster than nitrogen globally in freshwater ecosystems under anthropogenic impacts. Liu L, editor. Ecology Letters. 2016;19: 1237–1246. pmid:27501082
- 5. Foley JA, Ramankutty N, Brauman KA, Cassidy ES, Gerber JS, Johnston M, et al. Solutions for a cultivated planet. Nature. 2011;478: 337–342. pmid:21993620
- 6. Galloway JN, Dentener FJ, Capone DG, Boyer EW, Howarth RW, Seitzinger SP, et al. Nitrogen Cycles: Past, Present, and Future. Biogeochemistry. 2004;70: 153–226.
- 7. Haygarth PM, Condron LM, Heathwaite AL, Turner BL, Harris GP. The phosphorus transfer continuum: Linking source to impact with an interdisciplinary and multi-scaled approach. Science of The Total Environment. 2005;344: 5–14. pmid:15907506
- 8. Vautier C, Kolbe T, Babey T, Marçais J, Abbott BW, Laverman AM, et al. What do we need to predict groundwater nitrate recovery trajectories? Science of The Total Environment. 2021; 147661. pmid:34034194
- 9. Díaz S, Settele J, Brondízio ES, Ngo HT, Agard J, Arneth A, et al. Pervasive human-driven decline of life on Earth points to the need for transformative change. Science. 2019;366. pmid:31831642
- 10. Errigo IM, Abbott BW, Mendoza DL, Mitchell L, Sayedi SS, Glenn J, et al. Human Health and Economic Costs of Air Pollution in Utah: An Expert Assessment. Atmosphere. 2020;11: 1238.
- 11. Landrigan PJ, Fuller R, Acosta NJR, Adeyi O, Arnold R, Basu N (Nil), et al. The Lancet Commission on pollution and health. The Lancet. 2017;0. pmid:29056410
- 12. Ward MH, Jones RR, Brender JD, de Kok TM, Weyer PJ, Nolan BT, et al. Drinking Water Nitrate and Human Health: An Updated Review. Int J Environ Res Public Health. 2018;15. pmid:30041450
- 13. Abbott BW, Bishop K, Zarnetske JP, Hannah DM, Frei RJ, Minaudo C, et al. A water cycle for the Anthropocene. Hydrological Processes. 2019;33: 3046–3052.
- 14. Adler RW. Coevolution of Law and Science. 1. 2019;44: 1–66.
- 15. Cheng FY, Van Meter KJ, Byrnes DK, Basu NB. Maximizing US nitrate removal through wetland protection and restoration. Nature. 2020; 1–6. pmid:33328640
- 16. Nardi F, Cudennec C, Abrate T, Allouch C, Annis A, Herman Assumpção T, et al. Citizens AND HYdrology (CANDHY): conceptualizing a transdisciplinary framework for citizen science addressing hydrological challenges. Hydrological Sciences Journal. 2020; 02626667.2020.1849707.
- 17. Abbott BW, Bishop K, Zarnetske JP, Minaudo C, Chapin FS, Krause S, et al. Human domination of the global water cycle absent from depictions and perceptions. Nature Geoscience. 2019;12: 533–540.
- 18. Bonney R, Cooper CB, Dickinson J, Kelling S, Phillips T, Rosenberg KV, et al. Citizen science: a developing tool for expanding science knowledge and scientific literacy. BioScience. 2009;59: 977–984.
- 19. Fandel CA, Breshears DD, McMahon EE. Implicit assumptions of conceptual diagrams in environmental science and best practices for their illustration. Ecosphere. 2018;9: 1–15.
- 20. Bochet O, Bethencourt L, Dufresne A, Farasin J, Pédrot M, Labasque T, et al. Iron-oxidizer hotspots formed by intermittent oxic–anoxic fluid mixing in fractured rocks. Nat Geosci. 2020; 1–7.
- 21. Kolbe T, Dreuzy J-R de, Abbott BW, Aquilina L, Babey T, Green CT, et al. Stratification of reactivity determines nitrate removal in groundwater. PNAS. 2019; 201816892. pmid:30692250
- 22. Wollheim WM, Mulukutla GK, Cook C, Carey RO. Aquatic Nitrate Retention at River Network Scales Across Flow Conditions Determined Using Nested In Situ Sensors. Water Resources Research. 2017;53: 9740–9756.
- 23. Dupas R, Minaudo C, Abbott BW. Stability of spatial patterns in water chemistry across temperate ecoregions. Environmental Research Letters. 2019;14: 074015.
- 24. Kaufmann PR, Herlihy AT, Mitch ME, Messer JJ, Overton WS. Stream chemistry in the eastern United States: 1. Synoptic survey design, acid-base status, and regional patterns. Water Resources Research. 1991;27: 611–627.
- 25. Minaudo C, Meybeck M, Moatar F, Gassama N, Curie F. Eutrophication mitigation in rivers: 30 years of trends in spatial and seasonal patterns of biogeochemistry of the Loire River (1980–2012). Biogeosciences. 2015;12: 2549–2563.
- 26. McGuire KJ, Torgersen CE, Likens GE, Buso DC, Lowe WH, Bailey SW. Network analysis reveals multiscale controls on streamwater chemistry. PNAS. 2014;111: 7030–7035. pmid:24753575
- 27. Shogren AJ, Zarnetske JP, Abbott BW, Iannucci F, Frei RJ, Griffin NA, et al. Revealing biogeochemical signatures of Arctic landscapes with river chemistry. Sci Rep. 2019;9: 1–11. pmid:30626917
- 28. Radke AG, Godsey SE, Lohse KA, McCorkle EP, Perdrial J, Seyfried MS, et al. Spatiotemporal Heterogeneity of Water Flowpaths Controls Dissolved Organic Carbon Sourcing in a Snow-Dominated, Headwater Catchment. Front Ecol Evol. 2019;7.
- 29. Abbott BW, Gruau G, Zarnetske JP, Moatar F, Barbe L, Thomas Z, et al. Unexpected spatial stability of water chemistry in headwater stream networks. Ecology Letters. 2018;21: 296–308. pmid:29282860
- 30. Giri S, Qiu Z, Prato T, Luo B. An Integrated Approach for Targeting Critical Source Areas to Control Nonpoint Source Pollution in Watersheds. Water Resour Manage. 2016;30: 5087–5100.
- 31. Liu R, Xu F, Zhang P, Yu W, Men C. Identifying non-point source critical source areas based on multi-factors at a basin scale with SWAT. Journal of Hydrology. 2016;533: 379–388.
- 32. Bernhardt ES, Blaszczak JR, Ficken CD, Fork ML, Kaiser KE, Seybold EC. Control Points in Ecosystems: Moving Beyond the Hot Spot Hot Moment Concept. Ecosystems. 2017;20: 665–682.
- 33. Heathwaite AL, Quinn PF, Hewett CJM. Modelling and managing critical source areas of diffuse pollution from agricultural land using flow connectivity simulation. Journal of Hydrology. 2005;304: 446–461.
- 34. Blaszczak JR, Delesantro JM, Urban DL, Doyle MW, Bernhardt ES. Scoured or suffocated: Urban stream ecosystems oscillate between hydrologic and dissolved oxygen extremes. Limnol Oceanogr. 2019;64: 877–894.
- 35. Covino T. Hydrologic connectivity as a framework for understanding biogeochemical flux through watersheds and along fluvial networks. Geomorphology. 2017;277: 133–144.
- 36. Moatar F, Abbott BW, Minaudo C, Curie F, Pinay G. Elemental properties, hydrology, and biology interact to shape concentration-discharge curves for carbon, nutrients, sediment, and major ions. Water Resour Res. 2017;53: 1270–1287.
- 37. Lutz SR, Trauth N, Musolff A, Breukelen BMV, Knöller K, Fleckenstein JH. How Important is Denitrification in Riparian Zones? Combining End-Member Mixing and Isotope Modeling to Quantify Nitrate Removal from Riparian Groundwater. Water Resources Research. 2020;56: e2019WR025528.
- 38. Asano Y, Uchida T, Mimasu Y, Ohte N. Spatial patterns of stream solute concentrations in a steep mountainous catchment with a homogeneous landscape. Water Resources Research. 2009;45.
- 39. Creed IF, McKnight DM, Pellerin BA, Green MB, Bergamaschi BA, Aiken GR, et al. The river as a chemostat: fresh perspectives on dissolved organic matter flowing down the river continuum. Canadian Journal of Fisheries and Aquatic Sciences. 2015;72: 1272–1285.
- 40. Temnerud J, Bishop K. Spatial Variation of Streamwater Chemistry in Two Swedish Boreal Catchments: Implications for Environmental Assessment. Environ Sci Technol. 2005;39: 1463–1469. pmid:15819198
- 41. Sabo RD, Clark CM, Compton JE. Considerations when using nutrient inventories to prioritize water quality improvement efforts across the US. Environ Res Commun. 2021;3: 045005.
- 42. Wang J, Song C, Reager JT, Yao F, Famiglietti JS, Sheng Y, et al. Recent global decline in endorheic basin water storages. Nature Geoscience. 2018;11: 926. pmid:30510596
- 43. Wurtsbaugh WA, Miller C, Null SE, DeRose RJ, Wilcock P, Hahnenberger M, et al. Decline of the world’s saline lakes. Nature Geoscience. 2017;10: 816.
- 44. Hale RL, Scoggins M, Smucker NJ, Suchy A. Effects of climate on the expression of the urban stream syndrome. Freshwater Science. 2016;35: 421–428.
- 45. Wine ML, Laronne JB. In Water-Limited Landscapes, an Anthropocene Exchange: Trading Lakes for Irrigated Agriculture. Earth’s Future. 2020;8: e2019EF001274.
- 46. Austin AT, Yahdjian L, Stark JM, Belnap J, Porporato A, Norton U, et al. Water pulses and biogeochemical cycles in arid and semiarid ecosystems. Oecologia. 2004;141: 221–235. pmid:14986096
- 47. Harms TK, Grimm NB. Hot spots and hot moments of carbon and nitrogen dynamics in a semiarid riparian zone. J Geophys Res. 2008;113: G01020.
- 48. Fisher SG, Grimm NB, Martí E, Holmes RM, Jones JB Jr. Material spiraling in stream corridors: a telescoping ecosystem model. Ecosystems. 1998;1: 19–34.
- 49. Campo R del, Gómez R, Singer G. Dry phase conditions prime wet-phase dissolved organic matter dynamics in intermittent rivers. Limnology and Oceanography. 2019;64: 1966–1979. https://doi.org/10.1002/lno.11163.
- 50. Hale RL, Godsey SE. Dynamic stream network intermittence explains emergent dissolved organic carbon chemostasis in headwaters. Hydrological Processes. 2019;0.
- 51. Harjung A, Sabater F, Butturini A. Hydrological connectivity drives dissolved organic matter processing in an intermittent stream. Limnologica. 2018;68: 71–81.
- 52. Hansen CH, Williams GP. Evaluating Remote Sensing Model Specification Methods for Estimating Water Quality in Optically Diverse Lakes throughout the Growing Season. Hydrology. 2018;5: 62.
- 53. Hansen CH, Burian SJ, Dennison PE, Williams GP. Evaluating historical trends and influences of meteorological and seasonal climate conditions on lake chlorophyll a using remote sensing. Lake and Reservoir Management. 2020;36: 45–63.
- 54. Randall MC, Carling GT, Dastrup DB, Miller T, Nelson ST, Rey KA, et al. Sediment potentially controls in-lake phosphorus cycling and harmful cyanobacteria in shallow, eutrophic Utah Lake. PLOS ONE. 2019;14: e0212238. pmid:30763352
- 55.
Farmer J. On Zion’s Mount: Mormons, Indians, and the American Landscape. Harvard University Press; 2010.
- 56.
Baker W. My view: Algae blooms in Utah Lake. Deseret News. 22 Jul 2016. Available: https://www.deseret.com/2016/7/22/20592441/my-view-algae-blooms-in-utah-lake. Accessed 11 Feb 2021.
- 57.
PSOMAS. Utah Lake TMDL: Pollutant Loading Assessment and Designated Beneficial Use Impairment Assessment. Salt Lake City, Utah: State of Utah Division of Water Quality; 2007 Aug p. 88. Available: https://deq.utah.gov/ProgramsServices/programs/water/watersheds/docs/2009/02Feb/Final_Draft_Task2_Task3_Memo%20_08-01-07.pdf.
- 58. Follett A. Bartering the Public Trust: Assessing the Constitutionality of the Utah Lake Restoration Act (2018). Hinckley Journal of Politics. 2019;20: 25.
- 59.
Abbott BW, Carling GT, Follett A. Op-ed: The present, future and past of Utah Lake. Deseret News. 8 Mar 2018. Available: https://www.deseret.com/2018/3/8/20641364/op-ed-the-present-future-and-past-of-utah-lake. Accessed 5 Mar 2020.
- 60. Hill BH, Bolgrien DW, Herlihy AT, Jicha TM, Angradi TR. A Synoptic Survey of Nitrogen and Phosphorus in Tributary Streams and Great Rivers of the Upper Mississippi, Missouri, and Ohio River Basins. Water Air Soil Pollut. 2011;216: 605–619.
- 61. Church SP, Payne LB, Peel S, Prokopy LS. Beyond water data: benefits to volunteers and to local water from a citizen science program. Journal of Environmental Planning and Management. 2018; 1–21.
- 62. Crall AW, Jordan R, Holfelder K, Newman GJ, Graham J, Waller DM. The impacts of an invasive species citizen science training program on participant attitudes, behavior, and science literacy. Public Understanding of Science. 2012;22: 745–764. pmid:23825234
- 63.
Gilbert GK. Lake Bonneville. U.S. Government Printing Office, Washington D.C.: U.S. Geological Survey; 1890 p. 438. Report No.: 1. Available: http://pubs.er.usgs.gov/publication/m1.
- 64.
Hunt CB, Varnes HD, Thomas HE. Lake Bonneville: Geology of northern Utah Valley, Utah. Washington, D.C.: U.S. Government Printing Office; 1953 p. 109. Report No.: 257-A. Available: http://pubs.er.usgs.gov/publication/pp257A.
- 65.
Thompson RS, Oviatt CG, Honke JS, McGeehin JP. Chapter 11—Late Quaternary Changes in Lakes, Vegetation, and Climate in the Bonneville Basin Reconstructed from Sediment Cores from Great Salt Lake. In: Oviatt CG, Shroder JF, editors. Developments in Earth Surface Processes. Elsevier; 2016. pp. 221–291. https://doi.org/10.1016/B978-0-444-63590-7.00011–1
- 66.
US Census Bureau. Utah is Nation’s Fastest-Growing State, Census Bureau Reports. In: The United States Census Bureau [Internet]. Dec 2016 [cited 23 Feb 2019]. Available: https://www.census.gov/newsroom/press-releases/2016/cb16-214.html.
- 67. Hintze L, Willis G, Laes D, Sprinkel D, Brown K. DIGITAL GEOLOGIC MAP OF UTAH. 2000.
- 68. Neilson BT, Tennant H, Stout TL, Miller MP, Gabor RS, Jameel Y, et al. Stream Centric Methods for Determining Groundwater Contributions in Karst Mountain Watersheds. Water Resources Research. 2018;54: 6708–6724. https://doi.org/10.1029/2018WR022664.
- 69.
Cederberg JR, Gardner PM, Thiros SA. Hydrology of Northern Utah Valley, Utah County, Utah, 1975–2005. Reston, VA: U.S. Geological Survey; 2009. Report No.: 2008–5197. Available: http://pubs.er.usgs.gov/publication/sir20085197.
- 70. Zanazzi A, Wang W, Peterson H, Emerman SH. Using Stable Isotopes to Determine the Water Balance of Utah Lake (Utah, USA). Hydrology. 2020;7: 88.
- 71. Fuhriman DK, Merritt LB, Miller AW, Stock HS. HYDROLOGY AND WATER QUALITY OF UTAH LAKE. Great Basin Naturalist Memoirs. 1981; 43–67.
- 72. Silvertown J. A new dawn for citizen science. Trends in Ecology & Evolution. 2009;24: 467–471. pmid:19586682
- 73. Walker DW, Smigaj M, Tani M. The benefits and negative impacts of citizen science applications to water as experienced by participants and communities. WIREs Water. 2021;8: e1488. https://doi.org/10.1002/wat2.1488.
- 74. Dickinson JL, Zuckerberg B, Bonter DN. Citizen Science as an Ecological Research Tool: Challenges and Benefits. Annual Review of Ecology, Evolution, and Systematics. 2010;41: 149–172.
- 75. Buytaert W, Zulkafli Z, Grainger S, Acosta L, Alemie TC, Bastiaensen J, et al. Citizen science in hydrology and water resources: opportunities for knowledge generation, ecosystem service management, and sustainable development. Front Earth Sci. 2014;2.
- 76. McKinley DC, Miller-Rushing AJ, Ballard HL, Bonney R, Brown H, Cook-Patton SC, et al. Citizen science can improve conservation science, natural resource management, and environmental protection. Biological Conservation. 2017;208: 15–28.
- 77. Pandya RE. A framework for engaging diverse communities in citizen science in the US. Frontiers in Ecology and the Environment. 2012;10: 314–317. https://doi.org/10.1890/120007.
- 78. Den Broeder L, Lemmens L, Uysal S, Kauw K, Weekenborg J, Schönenberger M, et al. Public Health Citizen Science; Perceived Impacts on Citizen Scientists: A Case Study in a Low-Income Neighbourhood in the Netherlands. CSTP. 2017;2: 7.
- 79. Abbott BW, Moatar F, Gauthier O, Fovet O, Antoine V, Ragueneau O. Trends and seasonality of river nutrients in agricultural catchments: 18years of weekly citizen science in France. Science of The Total Environment. 2018;624: 845–858. https://doi.org/10.1016/j.scitotenv.2017.12.176 pmid:29274609
- 80. Poisson AC, McCullough IM, Cheruvelil KS, Elliott KC, Latimore JA, Soranno PA. Quantifying the contribution of citizen science to broad-scale ecological databases. Frontiers in Ecology and the Environment. 2020;18: 19–26. https://doi.org/10.1002/fee.2128.
- 81. Elliott KC, Rosenberg J. Philosophical Foundations for Citizen Science. Citizen Science: Theory and Practice. 2019;4: 9.
- 82. Breuer L, Hiery N, Kraft P, Bach M, Aubert AH, Frede H-G. HydroCrowd: a citizen science snapshot to assess the spatial control of nitrogen solutes in surface waters. Sci Rep. 2015;5: 16503. pmid:26561200
- 83.
4500-P PHOSPHORUS. Standard Methods For the Examination of Water and Wastewater. American Public Health Association; 2018. https://doi.org/10.2105/SMWW.2882.093
- 84. Blöschl G, Grayson RB, Sivapalan M. On the representative elementary area (REA) concept and its utility for distributed rainfall-runoff modelling. Hydrological Processes. 1995;9: 313–330.
- 85. Wood EF, Sivapalan M, Beven K, Band L. Effects of spatial variability and scale with implications to hydrologic modeling. Journal of Hydrology. 1988;102: 29–47.
- 86. Killick R, Eckley IA. changepoint: An R Package for Changepoint Analysis. Journal of Statistical Software. 2014;58: 1–19.
- 87.
Wickham H. ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York; 2016. Available: http://ggplot2.org.
- 88.
Wickham H, François R, Henry L, Müller K. dplyr: A Grammar of Data Manipulation. 2019. Available: https://CRAN.R-project.org/package=dplyr.
- 89.
R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2018. Available: https://www.R-project.org/.
- 90. Fanelli RM, Blomquist JD, Hirsch RM. Point sources and agricultural practices control spatial-temporal patterns of orthophosphate in tributaries to Chesapeake Bay. Science of The Total Environment. 2019;652: 422–433. pmid:30368173
- 91. Frazar S, Gold AJ, Addy K, Moatar F, Birgand F, Schroth AW, et al. Contrasting behavior of nitrate and phosphate flux from high flow events on small agricultural and urban watersheds. Biogeochemistry. 2019;145: 141–160.
- 92. Gu S, Gruau G, Dupas R, Rumpel C, Crème A, Fovet O, et al. Release of dissolved phosphorus from riparian wetlands: Evidence for complex interactions among hydroclimate variability, topography and soil properties. Science of The Total Environment. 2017;598: 421–431. pmid:28448934
- 93. Zarnetske JP, Bouda M, Abbott BW, Saiers J, Raymond PA. Generality of Hydrologic Transport Limitation of Watershed Organic Carbon Flux Across Ecoregions of the United States. Geophysical Research Letters. 2018;45: 11,702–11,711.
- 94. Hoellein TJ, Arango CP, Zak Y. Spatial variability in nutrient concentration and biofilm nutrient limitation in an urban watershed. Biogeochemistry. 2011;106: 265–280.
- 95. Dent CL, Grimm NB, Fisher SG. Multiscale effects of surface–subsurface exchange on stream water nutrient concentrations. Journal of the North American Benthological Society. 2001;20: 162–181.
- 96. Coble AA, Koenig LE, Potter JD, Parham LM, McDowell WH. Homogenization of dissolved organic matter within a river network occurs in the smallest headwaters. Biogeochemistry. 2019;143: 85–104.
- 97. Likens GE, Buso DC. Variation in Streamwater Chemistry Throughout the Hubbard Brook Valley. Biogeochemistry. 2006;78: 1–30.
- 98. Tiwari T, Buffam I, Sponseller RA, Laudon H. Inferring scale-dependent processes influencing stream water biogeochemistry from headwater to sea. Limnology and Oceanography. 2017;62: S58–S70.
- 99. Johnson B, Smith E, Ackerman JW, Dye S, Polinsky R, Somerville E, et al. Spatial Convergence in Major Dissolved Ion Concentrations and Implications of Headwater Mining for Downstream Water Quality. JAWRA Journal of the American Water Resources Association. 2019;55: 247–258. pmid:33354106
- 100. Dupas R, Abbott B, Minaudo C, Fovet O. Distribution of Landscape Units Within Catchments Influences Nutrient Export Dynamics. Frontiers of Environmental Science & Engineering. 2019;7: 43.
- 101. Hale RL, Turnbull L, Earl SR, Childers DL, Grimm NB. Stormwater Infrastructure Controls Runoff and Dissolved Material Export from Arid Urban Watersheds. Ecosystems. 2015;18: 62–75.
- 102. Gardner KK, McGlynn BL. Seasonality in spatial variability and influence of land use/land cover and watershed characteristics on stream water nitrate concentrations in a developing watershed in the Rocky Mountain West. Water Resources Research. 2009;45. https://doi.org/10.1029/2008WR007029.
- 103. Dickinson JL, Shirk J, Bonter D, Bonney R, Crain RL, Martin J, et al. The current state of citizen science as a tool for ecological research and public engagement. Frontiers in Ecology and the Environment. 2012;10: 291–297.