Purpose and Scope

Since 1975, the SRWTP has been discharging variable amounts of treated wastewater into Bright Angel Wash, which is oriented along the Bright Angel Fault (fig. 1). The potential influence of this wastewater through faults and fractures to springs along the South Rim of the Grand Canyon has never been directly measured. The treated wastewater could be introducing CECs to sensitive spring ecosystems, which could have unintended effects on spring water quality and aquatic organisms.

A 2-year USGS investigation in cooperation with the NPS was completed to test for the presence of CECs in springs discharging along the South Rim of the Grand Canyon. The April 2021 sampling was intended to be a synoptic characterization of CECs for the purpose of establishing a baseline for more comprehensive studies. Samples were collected once in April 2021 from seven springs with potential flow paths downgradient of Bright Angel Fault and analyzed for wastewater indicator CECs, focusing on pharmaceuticals and PFAS. Samples were also collected from Bright Angel Wash downstream of the SRWTP and a nearby shallow groundwater well dug into the Kaibab Limestone known as Rowe Well. Results from the April 2021 sampling were combined with data retrieved from the Water Quality Portal (WQP; National Water Quality Monitoring Council, 2023) from 1980 through 2022 to assess other tracer compounds that may indicate historical wastewater influence. The information from this study will help identify if there is a connection between the SRWTP discharge and downstream springs. It will also help to establish baseline characterization of the types and concentrations of CECs to aid Grand Canyon National Park in their discussion on infrastructure changes at the SRWTP.

Description of Study Area and Climate

The study area is in the Colorado Plateau physiographic province of northern Arizona. The Colorado River in the Grand Canyon divides the Colorado Plateau into the Kaibab Plateau (North Rim) and Coconino Plateau (South Rim; Metzger, 1961). Each plateau has a somewhat different climate and different hydrogeochemistry of groundwater sources (fig. 1; Metzger, 1961; Huntoon, 1974). The youngest rocks in the South Rim study area are Paleozoic sedimentary rocks underlain by Proterozoic igneous and metamorphic basement rocks (Metzger, 1961). Groundwater in this study was sampled from the Redwall-Muav aquifer (Mississippian to Cambrian in age), which is near the bottom of the Paleozoic stratigraphic sequence (Billingsley, 2000). Water is thought to recharge the Coconino aquifer about 50 miles to the south of the Grand Canyon (Bills and others, 2007; Crossey and others, 2009; Solder and others, 2020). Solder and others (2020) propose that some additional smaller areas of localized recharge may be closer to the South Rim (Knight and Huntoon, 2022). Along the South Rim at 7,000 feet (ft), the vegetation consists primarily of Pinus ponderosa Douglas ex P. Lawson & C. Lawson (ponderosa pine) forests along with additional species such as Quercus gambelii Nutt. (Gambel’s oak). The springs sampled for this study range in elevation between 3,700 and 4,270 ft. Within the spring watersheds are Populus fremontii S. Watson (Fremont’s cottonwoods) and other riparian species. Springs and perennial streams below the South Rim support riparian habitat for wildlife and aquatic biota, some of which are endangered. Much of this diversity can be attributed to the Park’s dramatic topographic spectrum. Its elevational variety provides microhabitats for natural processes supporting rare and endemic plant and wildlife species. For example, nearly 70 percent of the aquatic beetle species of the Grand Canyon region require springs or spring-fed streams and do not appear in other habitats (Grand Canyon Wildlands Council, Inc., 2004). According to Usher and others (1984), the perennial stream reach below the confluence of the Garden and Pipe Creeks contains several native fish species, such as Rhinichthys osculus (Girard, 1856; speckled dace), Catostomus discobolus yarrowi (Cope, 1874; Zuni bluehead), and Catostomus latipinnis (Baird and Girard, 1853; flannelmouth sucker). The spring ecosystems are incredibly diverse and unique to the region and are therefore a high priority for Grand Canyon National Park management to preserve and protect (National Park Service, 2010).

Groundwater movement in the Grand Canyon region is dominated by fault and fracture-controlled flow paths wherein some older geologic structures serve as barriers to flow and younger structures provide pathways for rapid movement of water through the subsurface, as evidenced by recent dye tracer studies along the North Rim of the Grand Canyon (Jones and others, 2018). Dye was placed into sinkholes there and subsequently moved vertically several hundred meters and horizontally several kilometers to springs in less than a year (Jones and others, 2018). Master joints likely increase the hydraulic conductivities of brittle lower carbonate rocks of the region and facilitate fracture flow; such joints are frequently spaced between major structures (Huntoon, 1974). Northwest trending faults are present in the Horn Creek and Monument Creek watersheds and intersect the Bright Angel and Hermit Faults (Maxson, 1961). These faults may provide preferential pathways for groundwater movement. Springs are present in each of the major drainages that cut into the South Rim of the Grand Canyon and may also represent locations of preferential, structurally controlled flow paths.

The average annual rainfall of the South Rim of the Grand Canyon measured by the NPS Tusayan weather station during water years (WY) 2004–21 was 14.8 inches, and the average snowfall was around 58 inches (fig. 2A; Walking Shadow Ecology, 2022). During the 2-year period spanning the April 2021 sampling, the average annual precipitation for WY 2020 and WY 2021 was about 4 inches less than the average in WY 2004–21 and was near the lowest tenth percentile for that period. August had the greatest median precipitation, and June had the lowest (fig. 2B). Mean maximum and minimum temperatures were 65 and 31 degrees Fahrenheit, respectively, at the NPS Tusayan weather station. The South Rim is hotter and drier than the North Rim and requires the delivery of Roaring Springs water from the North Rim to the South Rim via the Transcanyon Waterline (National Park Service, 2014). Roaring Springs, however, is highly dependent on snowpack (Ross, 2005), which has led Grand Canyon National Park to look for alternative or additional sources of water, like Bright Angel Creek, which flows from the North Rim to its confluence with the Colorado River at Phantom Ranch.

Water Supply and Wastewater Treatment History

In the 1890s, water supply was limited on the South Rim of the Grand Canyon. Water sources included Rowe Well, local springs, and rainwater catchments. The Atchison, Topeka & Santa Fe Railway Company (ATSF) started hauling water from Flagstaff in 1908 (Anderson, 1999). Prior to that, water was hauled over 125 miles from Del Rio Springs in Chino Valley, Arizona, to the South Rim (McConnell, 2006). By 1919, the ATSF was hauling as much as 100,000 gallons per day to support steam power and generators and to supply water to the affiliated Fred Harvey hotel facilities and other visitor accommodations (Anderson, 1999). In 1927, the ATSF gained control of Garden Spring and its substantial spring flow below the South Rim, and in 1932, a pumphouse and facilities were built for moving Garden Spring water 3,200 ft up to the South Rim. By 1934, the system piped as much as 150,000 gallons per day to Grand Canyon Village storage tanks (Steely, 2015). Water supply on the South Rim remained adequate for several decades (Anderson, 1999; John Milner Associates, Inc., 2005), but that changed with the rapid increase in visitation by the early 1950s.

There were 1 million park visitors in 1956, and the National Park anticipated millions more in the years to come, with the majority concentrated at the Grand Canyon Village (Steely, 2015). The projected visitation figures exceeded the Garden Spring and railroad-supplemented water supply. The North Rim’s Roaring Springs was selected as an alternate water source, and initial construction of the Transcanyon Waterline began in 1965 (Steely, 2015). The waterline became operational in 1970 and has been supplying water for the South Rim's visitor and management facilities ever since. The 12.5-mile waterline can transport as much as 1 million gallons per day of naturally flowing spring water from the North Rim’s Roaring Springs cave directly to the South Rim.

With the advent of a consistent water supply to the South Rim came the need to properly treat and sanitarily reuse that water. Around 1900, sewage treatment on the South Rim consisted of septic tanks and some wastewater filtration. Water scarcity was recognized early on by Grand Canyon Village, and wastewater reclamation was quickly adopted (Hommon, 1928; Anderson, 1999). The effluent, however, was unsanitary, contaminated with hydrogen sulfide, and could not be used for reuse activities like irrigating lawns near the hotel or for producing steam and heat with boilers in the long term (Hommon, 1928; McConnell, 2006). By the early 1920s, the septic systems were no longer meeting the needs of the Grand Canyon National Park owing to increased visitation.

A treatment facility was built in 1925. The facility initially used an activated sludge plant and the supplementary treatment of water (activated sludge-aeration, settling tanks, clarifier, and chlorination) in its treatment operations (Hommon, 1928; Garthe and Gilbert, 1968; Historic American Engineering Record, 1968). Treated wastewater was pumped 2 miles back to Grand Canyon Village and distributed for use in campground and El Tovar Hotel toilets, boiler feed water, cooling for electricity generation, and irrigation for shrubbery (National Park Service, 1978; McConnell, 2006). Unused wastewater was treated through the activated sludge portion of the plant and flowed to the evaporation lagoons several miles south that overflowed into a wash flowing southwest from the South Rim near the confluence of Bright Angel and Coconino Washes. These same lagoons continued to be used as a water supply for fire suppression (fig. 1). By the early 1950s, the NPS had assumed operations of the treatment plant. The antiquated sanitary facilities, however, could no longer handle the large crowds, and in 1972 a new activated sludge treatment plant was constructed to replace the old plant (National Park Service, 1978). Both treatment plant facilities were utilized between 1972 and 1985 while new holding tanks and infrastructure were added (National Park Service, 1978; McConnell, 2006). The new SRWTP is located approximately 0.3 mile east of the old facility.

Treatment processes at the SRWTP consist of influent screening and aerated grit chambers, biological treatment using activated sludge, secondary clarification, and tertiary treatment consisting of flocculation, coagulation, sedimentation, filtration, and disinfection with chlorine gas (Roberts and others, 1999). Sludge is treated by aerobic digestion and drying beds, then transported by truck to offsite storage lagoons (Roberts and others, 1999). Treated effluent from the SRWTP is discharged to the reclaimed water system storage tanks within Grand Canyon Village (Roberts and others, 1999). When reclaimed water demand is less than the plant treated wastewater flow, the excess treated wastewater is discharged into Bright Angel Wash or to lagoons during emergency bypasses or for fire suppression. During normal operations, about one-third of the treated water is returned to the reclaimed water system, one-third is released to the lagoons, and the remaining third is discharged into Bright Angel Wash.

The Park was reissued an Arizona Pollutant Discharge Elimination System permit for the SRWTP in 2010—effluent discharge records prior to 2010 are sparse. From 2010 through 2022, the average effluent discharge was about 6 million gallons per month (Mgal/mo; fig. 3A; Arizona Department of Environmental Quality, unpub. data, 2022). August has the highest average discharge, and December and January have the lowest. The maximum volume discharged in 2010–22 was 9.9 Mgal/mo in 2015 (fig. 3B). Discharge has varied over time, dropping as low as about half of what it was at its peak. Zukosky (1995) reported effluent discharge averaging 10 Mgal/mo during 1992 and 1993. This all indicates the volume of water discharging into Bright Angel Wash has been highly variable since the completion of the SRWTP. The volume of treated wastewater discharge is tied to many factors associated with the population residing and visiting at the Park. Evolving reuse strategies also affect the volume of water reuse (toilets, irrigation, fire suppression), as well as increases in water conservation. Visitation is the most influential factor of wastewater generation and discharge volume going to Bright Angel Wash (fig. 3C). For example, the average monthly discharge in 2020 was about 3 Mgal/mo less than that in 2019 because of decreased staffing and visitation related to the coronavirus 2019 pandemic.

Description of Springs and Associated Streams

In April 2021, seven springs, one well cistern (Rowe Well), and Bright Angel Wash were sampled for wastewater-related CECs using discrete or grab and time-integrated sampler methods. The seven springs were Pipe, Pumphouse, Garden, upper Horn Bedrock, Horn East Alluvium, Salt Creek, and Monument (fig. 1; table 1). Data from 1980 through 2022 were retrieved from the WQP (National Water Quality Monitoring Council, 2023) and include the previously mentioned springs, adjacent Hermit and Burro Springs, and streams downstream of these spring headwaters. The SRWTP and Tusayan Well were also included in the data retrieval. The following descriptions for the April 2021 sampling are modified from Monroe and others (2005). All spring locations have perennial flow, but streamflow is variable in the creeks. Garden Creek is perennial, whereas Salt, Horn, and Monument Creeks are intermittent.

Pipe Spring (USGS site no. 360410112055700) issues from the alluvium in the main Pipe Creek watershed about 900 ft upstream of the Tonto Trail Crossing (fig. 4A). Water was collected at the point of issuance on the right bank below the root mass of a tree. Discharge was measured about 10 ft downstream using a volumetric approach, and the flow is perennial. Pipe Spring water merges with Burro Spring water, continuing as Pipe Creek down to Garden Creek and the Colorado River.

The Garden Creek watershed includes several springs and seeps. This watershed has historically been described as a “collection gallery” for groundwater related to the Bright Angel Fault (Metzger, 1961). Spring development and landscape alterations since the early 1900s have affected the hydrology in the Havasupai Gardens area. The largest springs have been identified as Garden Spring, formerly referred to as Indian Garden Spring, Pumphouse or Two Trees Spring, and Pumphouse Wash. Inconsistent location descriptions and coordinates have led to uncertainty about the accuracy of historical sampling locations (Fitzgerald, 1996; Kobor, 2004; Nuyttens, 2022). Garden Spring has been described as the largest spring location on the east slope of the canyon along the fault, at the base of the Muav Limestone (Metzger, 1961; Zukosky, 1995). This description does not fit the current description of Garden Springs and potentially resembles Pumphouse Wash.

Pumphouse Spring (USGS site no. 360441112073201), shown in figure 4B, is about 350 ft southeast of the NPS Garden Creek Pump Station. Perennial spring water first emerges from beneath a large boulder that is about 100 ft upslope from two prominent Fremont’s cottonwoods (fig. 4C). Previous studies refer to this location as Two Trees Spring (Fitzgerald, 1996; Nuyttens, 2022; Curry and others, 2023). The spring discharges from hillslope alluvium overlying the Bright Angel Shale. Water samples were collected near the boulder. The water enters the alluvium immediately below the spring and re-emerges about 10 ft above the USGS streamgage station Pumphouse Wash near Grand Canyon, Arizona (USGS site no. 09403013). The spring has been altered and modified with a box construction and conveyance structure. About 300 ft downstream of the streamgage, the spring water meets the confluence of Garden Creek, which then flows to the Colorado River.

Garden Spring (USGS site no. 360439112073901), formerly referred to as Indian Garden Spring, is within the Garden Creek watershed and located within the Havasupai Gardens campground. The current issuance point is constructed at the base of a Fremont’s cottonwood (fig. 4D), but the spring quickly becomes a large discharge area with multiple seeps and thick aquatic vegetation growth of 17 plant species that provide habitat for several rare aquatic macroinvertebrate species in the area called Garden Creek (Nuyttens, 2022). Garden Creek gains flow from several springs and seeps, including Pumphouse Spring. The confluence with Pipe Creek occurs before the waters reach the Colorado River. The water samples were collected at the point of issuance at the base of the tree, which appears to be a historical alteration to constrain or convey flow. Between 1920 and 1940, the area that is now the Havasupai Gardens campground was developed and altered from its natural condition (John Milner Associates, Inc., 2005). This includes the spring development work by the Santa Fe Railroad to increase flow from the springs at Garden Creek in 1927 (John Milner Associates, Inc., 2005). Later, building construction and stream restoration by the NPS also altered the natural spring-flow conditions. The historical alterations of the Havasupai Gardens landscape, dotted with springs and seeps, may partially explain the uncertain naming and geographical locations of the original springs in the area.

The Horn Creek (eastern) watershed has an upper location called upper Horn Bedrock Spring (USGS site no. 360439112084601; fig. 4E) and a lower location called Horn East Alluvium (USGS site no. 360443112083300; fig. 4F). Upper Horn Bedrock Spring is about 1,500 ft upstream of Horn East Alluvium on a western watershed spur of the east subbasin. The spring emerges on the downstream side of the northwest-striking fault and forms a small pool below a sheer bedrock cliff. Water samples were collected from the small pool. About 1,600 ft upstream from the Tonto Trail crossing of the eastern Horn Creek watershed, the lower alluvium sample site is in the east tributary of Horn Creek, where water discharges from the channel alluvium that overlies the Bright Angel Shale. Water samples were collected at a small pool that was formed by boulders in the stream channel. The confluence with the east watershed is about 1,200 ft downstream of the Tonto Trail crossing at the west watershed. Flow is intermittent to the confluence with the Colorado River.

Salt Creek Spring (USGS site no. 360439112094101) emerges from a bedrock cliff in the main Salt Creek watershed about 2,600 ft upstream from the Tonto Trail crossing of Salt Creek (fig. 4G). Water discharges from bedding planes in the Muav Limestone and drips down a 26-ft high cliff face of Muav Limestone onto a small talus slope on the west side of the canyon. All water enters the channel alluvium immediately downstream from the talus slope. Water samples were collected from dripping cliff faces that form a small pool in a small impression in the headwall. The Salt Creek flows ephemerally to the Colorado River.

Monument Spring (USGS site no. 360356112103201) is at the headwall of a small tributary on the east side of Monument Creek about 2 miles upstream from the Tonto Trail (fig. 4H). The head of the Monument Creek watershed is known as The Abyss, and the near-vertical cliffs rise about 0.6 miles to the South Rim of the Grand Canyon. The spring is in an alcove above a series of waterfalls that support several hanging gardens that include Adiantum capillus-veneris L. (common maidenhair), Aquilegia chrysantha A. Gray (golden columbine), and other aquatic plants. There are numerous smaller springs and seeps in the alcove. The main spring discharges from a vertical fracture near the contact between the Redwall Limestone and the Temple Butte Limestone (Monroe and others, 2005). The combined gaining discharge pours over a cliff as a waterfall and collects at the waterfall’s base about 300 ft downstream of the primary spring issuance. Monument Creek is intermittent between the spring and the Tonto Trail crossing, where flow resurfaces and is perennial in most years. From there, it flows intermittently for about 1.5 miles and then joins with the Colorado River.

In the early 1890s, Sanford Rowe established the closest viable well to Grand Canyon Village, which was about 2.5 miles to the southwest of the Village (Hommon, 1928). Rowe operated a small tourist camp near his well, known as Rowe Well (USGS site no. 360205112104601), and a livery service between Williams, Arizona, and his camp. During this period, the well served as a water source for Grand Canyon Village residents and NPS facilities. The well is dug into the Kaibab Formation and acts as a cistern for rainwater runoff or very localized infiltration from a nearby seep. A submersible pump was used to purge and sample the well (fig. 4I). Sample bottles were lowered into the well to sample for PFAS to avoid concentration bias from the polytetrafluoroethylene tubing.

The SRWTP discharges treated wastewater into Bright Angel Wash. The Bright Angel Wash sampling location (USGS site no. 360230112093401; fig. 4J) was 0.7 miles downstream of the outfall. Bright Angel Wash is an ephemeral channel when wastewater effluent is not being discharged (fig. 4K). The surface flow infiltrates into the ephemeral channel within 3 miles downstream of the outfall. The stream channel continues southwest and converges with Coconino Wash about 5 miles downstream. The stream channel resides in a forested area that includes ponderosa pines and Gambel’s oaks. The SRWTP outfall is alternatively named Clearwell Overflow in the dataset retrieved from the WQP. References to a “Clear Well” are made in Hommon (1928), which describes it as a basin tank with an overflow. Later studies refer to samples collected at Clearwell Overflow but geographically plot near the SRWTP outflow, which flows into Bright Angel Wash (Zukosky, 1995; Ingraham and others, 2001; Curry and others, 2023). For the purposes of clarity in this USGS report, Bright Angel Wash includes Clearwell Overflow WQP data.

Previous Investigations

The USGS and NPS have studied water quality in several of the springs along the South Rim (Grand Canyon Wildlands Council, Inc., 2004; Monroe and others, 2005; Dyer and others, 2016; Dyer and Stumpf, 2017). Much of the research has focused on understanding South Rim spring water sources, residence times, and flow paths for the purpose of water extraction, determining aquatic ecosystem water requirements, and the potential effects from legacy and future uranium mining on the water quality of springs in the region (Zukosky, 1995; Fitzgerald, 1996; Ingraham and others, 2001; Kobor, 2004; Monroe and others, 2005; Beisner and others, 2020; 2023a, b; Solder and others, 2020; Nuyttens, 2022; Curry and others, 2023).

Metzger (1961) completed one of the first groundwater investigations of the South Rim of Grand Canyon National Park. The investigation was made at the request of the NPS to determine whether a sufficient water supply could be developed to increase supply for Grand Canyon Village. He provided a detailed description of the geology and sources of springs along the South Rim and documented the numerous faults and the complexity of spring flow in the region. There are minor faults in the eight canyons descending to the main canyon. Metzger (1961) illustrated the complexity of the area’s geology by mentioning the more than 50 faults in the Hermit Basin concentrated in less than a mile along the contact of the Toroweap Formation and Coconino Sandstone. These faults had a throw ranging from a few inches to a maximum of 30 ft. Metzger (1961) also described the springs at Garden Springs as an effective “collection gallery” for groundwater in the vicinity, yielding about 300 gallons per minute (gal/min), and indicated they are closely associated with the Bright Angel Fault. The sources, pathways, and volume of spring discharge became a primary research topic for the NPS because there were distinct differences between the geochemical type, age, and discharge volume of water at the different springs.

The source and age of water have always been key questions for understanding the persistence of water available for human and wildlife use. Dye tracer studies on the North Rim have shown a rapid response to snowmelt and runoff (Jones and others, 2018). The South Rim receives about half as much precipitation as the North Rim, but Goings (1985) indicated a relatively rapid response in South Rim spring discharge from precipitation. Goings (1985) developed correlations between precipitation and spring flow at the Salt Creek, Monument, Horn, and Hermit Springs; the lag period was between 1 and 2 months. The rapid response suggested the water is young and structurally controlled, which allows interconnected faults to route water quickly and increase the speed of travel time within the basins. This was particularly apparent in larger basins, like Hermit and Monument, which are extensively fractured. Tritium and uranium isotopes were used later by Fitzgerald (1996) to show that, contradictory to Goings (1985), the South Rim water residence times were on the order of decades. Fitzgerald (1996) also classified the springs into different spring-water types on the basis of geochemical and inorganic analytes. One group-type included the Hermit, Pumphouse, and Pipe Springs; the springs were determined to have longer residence times based on tritium results (before the bomb peak of the mid-1960s¹ ). Another group-type water, which included the Monument, Salt Creek, and Horn Springs, was determined to have shorter residence times (after the bomb peak of the mid-1960s), but these were still on the order of a few decades.

Further complicating the issue of spring water source and age was the Transcanyon Waterline delivering North Rim water from Roaring Springs to meet an ever-increasing demand for water on the South Rim. Additional below-ground inputs come from leaks in the waterline and other water infrastructure in Grand Canyon Village. Another significant input of North Rim water comes from the treated effluent discharged into Bright Angel Wash, which infiltrates within the first few miles of the wash. Zukosky (1995) hypothesized that the 10 Mgal/mo was likely infiltrating into the South Rim hydrologic system because of the permeable nature of the Kaibab Limestone and the proximity of the Bright Angel Fault to the discharge area (fig.1). Zukosky (1995) distinguished South Rim water from North Rim water using rare-earth elements (REEs) and stable oxygen and hydrogen isotope ratios of spring water and groundwater. Furthermore, Zukosky (1995) suggested that Garden Spring water and Bright Angel Wash (named Clearwell Overflow in Zukosky [1995]) wastewater had similar characteristics to North Rim and South Rim waters, indicating a mixing of the treated wastewater, infiltrating via Bright Angel Fault and local groundwater. This implies that a portion of the discharge at Garden Spring may originate from the treated wastewater. However, the discrepancy in measured chloride concentrations between Bright Angel Wash wastewater and Garden Spring water presented contradictory evidence for this recharging-mixing relationship hypothesis. The elevated chloride Zukosky (1995) measured at Monument Creek was not highlighted in the study and could have been related to Bright Angel Wash wastewater, which had very high chloride concentrations. This mixing of water hypothesis and stable isotope approach was further pursued to better understand the source of water at Garden Spring, which serves as an important source in quantity and quality for the riparian and aquatic ecosystems that rely on it.

Ingraham and others (2001) used stable oxygen and hydrogen isotope ratios and conservative anions (chloride and nitrate) to support Zukosky’s (1995) finding that water discharging at Garden Spring had an isotopic signature like that of the SRWTP treated wastewater, and possibly as much as half of the annual discharge may originate from the treated wastewater. However, like Zukosky (1995), they found that Garden Spring contained no appreciable amounts of the anions associated with wastewater. Ingraham and others (2001) attributed this finding to a leak in the Transcanyon Waterline discovered above Garden Spring and suggested that a portion of that spring’s discharge may have its origin in water directly from the waterline, which ultimately dilutes the anion concentration. A follow-up review that focused on several springs along the South Rim indicated that there was no direct evidence of contaminants from the SRWTP affecting springs (Tobin and others, 2018). According to Tobin and others (2018), the lack of anions suggested that the shift in isotopic signature at Garden Spring was again more likely tied to leakage from the waterline rather than contamination from wastewater.

Monument Spring and Horn Spring, just west of Garden Spring, were highlighted by Monroe and others (2005) for having high nitrate and chloride concentrations. In Monument Spring, the concentrations indicated that groundwater discharge was influenced either by wastewater contamination on the South Rim of the Grand Canyon or by exposure to rocks and (or) geologic features in the subsurface (for example, breccia pipes or faults) not encountered by discharge from the other springs. The nitrate sources were unknown, but concentrations in Monument Spring were consistently elevated, and concentrations in Bright Angel Wash treated wastewater were consistently above the U.S. Environmental Protection Agency (EPA)’s maximum contaminant level (MCL) of 10 milligrams per liter (mg/L; U.S. Environmental Protection Agency, 2023). Horn Creek Spring’s nitrate concentration (1.2 mg/L) was not as high as that of Monument Spring (6.8 mg/L), but it was elevated compared to that of the nearby Salt Creek Spring (0.2 mg/L). Although nitrates in groundwater can come from natural sources, the most common source of elevated nitrate is human and animal waste sources or wastewater without denitrifying filtration. During Monroe and others’ (2005) study, concentrations at Monument Spring ranged from 5.7 to 6.8 mg/L, below the EPA’s MCL of 10 mg/L, but much higher than any of the 25 other springs discharging along the South Rim of the Grand Canyon.

During this April 2021 USGS investigation, a study was published by Curry and others (2023) that continued with the working hypothesis of Ingraham and others (2001) that Roaring Springs water from the Transcanyon Waterline infiltrates along the Bright Angel Fault and intermingles with groundwater. Curry and others (2023) estimated the proportion of Roaring Springs water contributed to each spring using geochemical tracers (major ions and stable isotopes) to define end members and develop mixing models. They proposed that Garden Spring water was about 40 percent Roaring Springs water, Pipe Spring was between 60 and 80 percent Roaring Springs water based on stable isotopic data alone, and the Hermit, Monument, Salt Creek, and Horn Springs were all between 0 and 20 percent Roaring Springs water. Curry and others (2023) observed some wastewater compounds at the Garden and Pipe Springs and attributed them to wastewater from the SRWTP, although several of the compounds are not known to be persistent and were not detected by Curry and others (2023) in the treated wastewater from the SRWTP. There was high variability between their two sampling events, and no concentrations of wastewater compounds were measured from any South Rim spring in February 2021. They did not sample the Horn, Salt Creek, or Monument Springs.

Field Collection

Spring samples for inorganic analytes were collected as close to the point of issuance from the ground as possible using a peristaltic pump with flexible silicon tubing. Water samples collected for PFAS and pharmaceuticals were unfiltered grab samples collected straight from the water source to avoid contamination or sorption associated with sampling equipment. Field parameters, including pH, water temperature, specific conductance, dissolved oxygen, and barometric pressure, were measured at each location immediately after these water samples were collected.

A water sample from the Rowe Well was collected using a low volume Grundfos Redi-Flo2 (Royal Eijkelkamp, Giesbeek, Netherlands) submersible sampling pump. The underground cistern dimensions were difficult to determine because of the limited view inside the opening. The Rowe Well cistern volume was never determined. The water level was approximately 15 ft below land surface, and the total depth was undetermined. A purge was attempted with a pump rate of around 1.4 gal/min, which amounted to a purge of approximately 200 gallons. During the 135-minute purge, the water level and field parameters did not change much: pH was within 0.05 pH units, specific conductance was within 0.5 percent, dissolved oxygen was within 0.02 mg/L, and temperature was within 0.02 degrees Celsius (°C). These conditions met stability criteria for sample collection (U.S. Geological Survey, variously dated). The water appeared to be a local source and perhaps sourced from a small, perched aquifer in the Kaibab Formation. The water was likely a combination of precipitation and localized runoff recharge.

Water samples collected for major ions, trace elements, and nutrients were filtered through a 0.45-micrometer capsule filter. Pharmaceutical samples analyzed at the USGS National Water Quality Laboratory (NWQL) in Lakewood, Colorado, were collected using a syringe and filtered with 0.45-micrometer glass fiber into a 20-milliliter (mL) baked amber glass vial. Unfiltered samples were collected in two 250 mL polycarbonate bottles for PFAS testing. Wastewater samples were collected in a 1-liter baked amber glass bottle and kept chilled on ice before shipping to the USGS Integrated Water Chemistry Analytical Laboratory (IWCAL) in Boulder, Colorado, for analysis. Detailed field and analytical descriptions are published by Beisner and others (2017, 2023a). Water samples were also collected for the analysis of tritium, stable isotope ratios of oxygen and hydrogen (δ¹⁸O and δ²H), carbon isotopes, uranium isotopes, and sulfur isotopes. Most of these sample results were not part of the interpretive analysis conducted in this investigation. Additional analysis of the associated samples can be found in Beisner and others (2017; 2023a, b).

Laboratory Analysis

Water samples were analyzed for major, trace, and REEs by the USGS National Research Program Laboratory in Boulder, Colorado (USGSTMCO), using inductively coupled plasma atomic emission spectrometric methods described by Garbarino and Taylor (1979, 1996)48 and Taylor (2001). The precision for all methods analyzed by the USGSTMCO was 4 percent or better, depending on the element. Water samples analyzed for nutrients were sent to the NWQL. These were analyzed using colorimetric determination methods described by Patton and Truitt (2000) and by Patton and Kryskalla (2011).

PFAS were analyzed at SGS Orlando using EPA method 537.1, which uses solid phase extraction and liquid chromatography-tandem mass spectrometry (Shoemaker and Tettenhorst, 2018). Wastewater tracer compounds were analyzed at the IWCAL, and pharmaceuticals were analyzed at the NWQL. Wastewater tracer compounds were analyzed at the IWCAL using continuous liquid-liquid extraction and then measured by gas chromatography-tandem mass spectrometry in multiple-monitoring mode following methods described by Barber and others (2000). Surrogate standards were added prior to extraction and work-up procedures, and isotopically labeled internal standards were added to extract immediately prior to analysis. Pharmaceutical analyses were conducted by the NWQL using a high-performance liquid chromatograph coupled to a triple quadrupole tandem mass spectrometer (Furlong and others, 2014). Carbamazepine and diphenhydramine were analyzed in both the wastewater and pharmaceutical method. A mean concentration and standard error are used in the results and discussion. Stable isotope ratios of oxygen (δ¹⁸O) and hydrogen (δ²H) were measured at the USGS Reston Stable Isotope Laboratory following mass spectrometer methods described by Révész and Coplen (2008a, b).

Data Analysis

Sample data collected for this investigation are stored on the USGS National Water Information System (NWIS; U.S. Geological Survey, 2024). Quality control information is stored on the internal USGS database and is not visible on the public webpage, but it is available upon request. Data were retrieved from the WQP (National Water Quality Monitoring Council, 2023), which also interfaces with the NWIS. The SRWTP outfall discharge and nitrate concentration data were retrieved from the Arizona Department of Environmental Quality in August 2022 (Arizona Department of Environmental Quality, unpub. data, 2022).² Data retrieved from the WQP include data from past USGS sample collection, NPS sample collection, and sample collection associated with academic research. The data retrieved from the WQP included data from studies by Goings (1985), Zukosky (1995), Fitzgerald (1996), Ingraham and others (2001), Dyer and others (2016), and Dyer and Stumpf (2017). The WQP data retrieved for the period of record is 1980 through 2022, but the amount of available data during that period varies from analyte to analyte. Sample data were first grouped by location and then by watershed for the analysis (fig. 5; table 2).

Location groups include sample locations within a few hundred feet around the spring or section of stream. For example, Pipe Spring includes samples at or near the spring, which can vary in location depending on yearly precipitation. Pipe Creek would include surface water samples downstream of the spring, but the upper part of the creek was distinguished from the lower part of the creek, which would be a mile or more downstream. Watershed groups are composed of spring and creek samples collected within each respective watershed. The Hermit Creek watershed was included in the analysis because of its large size and nearby proximity to the SRWTP. The WQP data were examined for erroneous data, but no quality control measures were taken to verify collection locations. Data and location information in the WQP have inconsistent collection methods and site geolocation methods because of the different entities (universities, State and Federal agencies) collecting the data. Geolocation methods have changed from paper maps to global position systems. In addition, the point of spring issuance can change because of anthropogenic landscape changes, channel alteration from flooding, and seasonal hydrologic conditions. For example, the Havasupai Gardens has historically been described as an area comprised of several points of issuance in the quickly gaining stream reach of Garden Creek, and at least one of the points of issuance becomes ephemeral during dry years (Metzger, 1961; Zukosky, 1995). Metzger (1961) described the largest spring at the Havasupai Gardens issuing from the east slope of the canyon along the fault, at the base of the Muav Limestone of Tonto Group and the other from the creek bed at the uppermost end of the Havasupai Gardens. Several other seeps flow within the creek. The area was developed throughout the 1900s and affected the hydrology of the Havasupai Gardens area. This included the addition of permanent structures with a leaching field, the settling ponds at the Transcanyon Waterline-Garden Creek Pump Station, and the streamgaging at Pumphouse (John Milner Associates, Inc., 2005). After inspection of the data retrieved from the WQP and review of the associated publications, the names and locations of Garden Spring, Pumphouse Spring, and Two Trees Spring appear to be inconsistent (Metzger, 1961; Zukosky, 1995; Fitzgerald, 1996; Ingraham and others, 2001; Kobor, 2004; Nuyttens, 2022). The locations are separated in the analysis if the sample location description or water chemistry appear to be inconsistent with conditions observed during this investigation. The primary example is Garden Springs data collected by Zukosky (1995) and the possible erroneous coordinates of the location in the WQP.

The WQP data were aggregated by watershed relative to the Bright Angel Fault (fig 5; table 2) to describe regional spatial patterns. Descriptions of the sites and location accuracy can be retrieved from the WQP (National Water Quality Monitoring Council, 2023). Because of the different researcher objectives and laboratory methods, the analysis and reporting limits are not always compatible for direct statistical comparison, although aggregated watershed data were tested for statistical difference using a nonparametric Wilcoxon signed-rank test (refer to the “Discussion” section for more information). Primarily, these data were used for generalized distribution comparisons (boxplots). Included among the inorganic analytes considered in the WQP data retrieval comparison are nitrate, chloride, boron, trace elements, and stable hydrogen (δ²H) and oxygen (δ¹⁸O) isotope ratios.

With the exception of Kobor (2004) and Nuyttens (2022), a detailed streamflow data compilation and water-quality analysis have not been conducted for spring discharge along the South Rim. This is partially related to the scarce and inconsistent data collection of spring discharge. Streamflow data were compiled from USGS streamgages (continuous stage-discharge relation) and the discrete discharge data from the WQP (includes USGS discrete measurements) to look for trends or patterns that might indicate a relation or shift with wastewater discharge influence over time. The period of record for each spring or creek varies temporally, and the discrete measurement frequency also varies over time. Inconsistent overlap makes comparisons or conducting trend analysis statistically challenging. In addition, the Arizona Department of Environmental Quality only has Arizona Pollutant Discharge Elimination System Permit discharge records going back to 2010 (figs. 3B, 3C; Arizona Department of Environmental Quality, unpub. data, 2022), which incidentally represents a period when minimal concurrent streamflow data were collected.

The method of speciation and the descriptions of dissolved and total nitrate data can be ambiguous within the analyte descriptions in the WQP (U.S. Environmental Protection Agency, 2017). The same amount of nitrate in a single sample can be reported two ways, nitrate as nitrogen (N) or nitrate as nitrate (NO₃^-). For example, 1 mg/L of nitrate as N and 4.5 mg/L of nitrate as NO₃^- would represent the same amount of nitrate in a single sample (U.S. Environmental Protection Agency, 2017). These data were checked from the WQP data retrieval and verified that all data were reported as nitrate as N.

The nitrogen data retrieved from the WQP consists of nitrate as N for filtered nitrate (dissolved) and unfiltered nitrate (total nitrate). Filtered and unfiltered water sample data were combined for general statistics but were labeled as such in the boxplots. Theoretically, the unfiltered data should tend to bias higher than the filtered in a 1-to-1 ratio; however, the mean ratio is closer to 1 (Sprague and others, 2017). This 1-to-1 ratio was verified with a data retrieval from the WQP for 809 pairs of filtered versus unfiltered nitrate groundwater and surface water samples for locations distributed throughout Arizona (U.S. Geological Survey, 2024). The bivariate fit produced a coefficient of determination (R²) of 0.98 (p<0.0001), and the bivariate equation was Filtered Nitrate as N = 0.010 + 0.989*Unfiltered Nitrate as N. The R² of 0.989 for Arizona samples has a mean ratio close to 1. A ratio of 1 is a perfect agreement. This supported combining the filtered and unfiltered samples in the analysis.

Gadolinium (Gd) is a naturally occurring REE commonly used as a contrast agent in magnetic resonance imaging (MRI). The medical use of Gd-based contrast agents has led to an input of anthropogenic Gd (Gd-anth) into natural environments (Brünjes and Hofmann, 2020). In many investigations, total Gd (Gd-total) is measured because the anthropogenic component (Gd-anth) cannot be measured directly (Brünjes and others, 2016). This approach requires the geogenic Gd (Gd-geo) background to be estimated to determine presence of an anthropogenic anomaly. This approach has advantages over trying to measure individual species of Gd-based contrast agents because Gd must first be complexed with strong organic chelators before administering to humans. However, there is no standard methodology on how to quantify Gd-geo in an aqueous sample, which has generated a variety of different approaches. Most calculations to determine Gd-anth are based on either interpolations or extrapolations of Gd-geo from the shale-normalized lanthanides by means of linear, logarithmic, geometric equations, or third-order polynomial fit (Brünjes and others, 2016).

REE concentration data are normalized to a geologic reference, such as chondrites, regional shales, or mantle material, and are plotted in order of decreasing atomic number. This reduces the Oddo-Harkins effect, which causes a sawtooth pattern in the REE plot. This is caused by the chemical elements with even atomic numbers being more abundant than the adjacent odd atomic number elements (Verplanck and others, 2010). Because of the Oddo-Harkins effect, Gd is naturally enriched compared to its nearest neighbors on the periodic table, europium (Eu) and terbium (Tb). However, despite its proximity to Gd, Eu was excluded from this analysis because of interferences caused by barium oxide and barium hydroxide in the inductively coupled plasma mass spectrometry measurements (Duvert and others, 2015). Terbium does not interfere with the analysis and was retained in the plots.

Different reference materials can be used to normalize the REEs and provide the smoothest REE patterns. Normalization allows the relative concentrations of the REE in different samples to be compared visually and also determines if Gd anomalies from Gd-anth could be observed. Differences between normalizing materials are often minimal, and two materials were chosen to compare for this investigation. The North American Shale Composite from Haskin and Haskin (1968) and the chondrite values from McDonough and Sun (1995) were used for the normalization of REEs. To quantify the magnitude of the Gd enrichment or observe an anomaly, the normalized dissolved Gd value was divided by the interpolated Gd between samarium and Tb using an equation described by Verplanck and others (2010, eq. 1).

Quality Assurance

For the one-time synoptic sampling in April 2021, two field blanks and one replicate were collected. A field blank was collected with certified inorganic-free blank water at Garden Spring. A replicate and a field blank were collected with certified organic-free and PFAS-free blank water at Bright Angel Wash. Multiple standard reference samples were analyzed by the USGSTMCO during every analytical run. A detailed analysis of the standard reference sample performance for the respective laboratories is presented by Beisner and others (2017, 2020, 2023a).

Parameters

Spring water temperature ranged from 12 °C to 18.6 °C, pH ranged from 7.3 to 8.5, specific conductance ranged from 417 to 1,080 microsiemens per centimeter (μS/cm), and dissolved oxygen ranged from 6.2 to 9.2 mg/L. Specific conductance in Bright Angel Wash wastewater effluent (1,270 μS/cm) was highest compared to any of the springs. Historically, the WQP data (1992–2002) indicate the Wash’s specific conductance was lower; its mean specific conductance was 874 μS/cm (table 3). Among the spring samples collected in April 2021, the highest specific conductance was measured at the Monument and upper Horn Bedrock Springs (774 μS/cm and 1,080 μS/cm, respectively), and lowest was measured at the Pumphouse and Garden Springs (417 μS/cm and 437 μS/cm, respectively). Historically (1992–2005), Monument Spring had a mean specific conductance of 838 μS/cm and at times was above 1,000 μS/cm. The concentrations exceeded 2,000 μS/cm at locations in lower sections of Monument Creek (1983–2008), Salt Creek (1983–2017), and Hermit Creek (1983–2008). Horn Creek Spring historically (1983–2019) had the highest mean specific conductance (840 μS/cm); however, the Monument and Salt Creek Springs were periodically more than 1,500 μS/cm, which is greater than the maximum specific conductance at Horn Creek Spring (1,180 μS/cm).

During the April 2021 sampling, the pH measured in samples taken at the Monument and Salt Creek Springs was above 8.00. At Garden Spring, pH was the lowest at 7.27. Historically (1992–93), Garden Spring also had the lowest mean pH at 7.23, and, similar to the April 2021 samples, the historical mean pH at Salt Creek Spring (2000–16) was above 8.00. Sample locations in watersheds west of Bright Angel Fault have historically had significantly greater median specific conductance concentration and median pH compared to the watersheds east of the fault (Wilcoxon rank-sum test, p<0.0001; table 4).

Contaminants of Emerging Concern

The April 2021 water samples were analyzed for 28 PFAS compounds, but only perfluoroalkyl acids (PFAAs) were found in the samples. No polyfluoroalkyl substances or fluoropolymers were detected in the samples, which is likely related to those compounds’ production, use, and chemical structure. The following nine PFAAs were detected: perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluorobutanesulfonic acid (PFBS), and perfluorooctanesulfonic acid (PFOS; fig. 6). All but one of the PFAS compounds detected in spring water (the short-chained PFBS) were also detected in the wastewater discharged into Bright Angel Wash. Monument Spring had five detections of PFAS compounds (in decreasing order of concentration: PFBS, PFPeA, PFOA, PFHxA, and PFOS; table 5), and upper Horn Bedrock Spring had only two detections (the short-chained PFBA and PFBS). No other sites had detections of PFAS compounds. PFBS was only found in samples collected from these two springs and not in the wastewater samples from Bright Angel Wash; the highest PFBS concentration was found at Monument Spring (3.8 nanograms per liter [ng/L]; fig. 7A).

The sample composition of PFAS compounds found at Bright Angel Wash and Monument Spring was similar, but proportionally, Monument Spring was higher in PFPeA (24.6–17.7 percent), PFHxA (13.8–12.6 percent), and PFOS (11.5–1.23 percent; fig. 7B; table 5). Proportions of PFAAs were more evenly distributed at Monument Spring, whereas Bright Angel Wash was dominated by PFOA (58.0 percent), PFPeA (17.7 percent), and PFHxA (12.6 percent). Bright Angel Wash had the highest concentration of PFOA (84.6 ng/L) and exceeded the Environmental Protection Agency maximum contaminant level of 4 ng/L (U.S. Environmental Protection Agency, 2023). The upper Horn Bedrock Spring had the highest concentration of PFBA (6.00 ng/L), only slightly higher than the PFBA concentration found in the Bright Angel Wash wastewater (5.60 ng/L).

Of the 113 wastewater and pharmaceutical compounds analyzed in the April 2021 samples, the 39 compounds detected in the Bright Angel Wash samples were typical of wastewater and included several pharmaceuticals (including pain relievers and nonprescription drugs), plasticizers, and detergent metabolites (fig. 8). The highest concentrations (greater than 500 ng/L) were of pharmaceuticals and these included diphenhydramine (antihistamine), sitagliptin and metformin (diabetes drugs), and guanylurea (primary transformation product of metformin). Methocarbamol (muscle relaxer), fluconazole (antifungal), and carbamazepine (anti-epileptic) were also above 500 ng/L. Nonionic detergent metabolites made up a large number of the detections in the wastewater. This group contains biodegradation products of the alkylphenol polyethoxylates, one of the major ionic surfactants produced and used in the United States (Kveštak and Ahel, 1995). These surfactants are widely used in commercial and household detergents and include 4-Nonylphenolmonoethoxylate (NP1EO), 4-Nonylphenoldiethoxylate (NP2EO), 4-tert-Octylphenolmonoethoxylate (OP1EO), and 4-tert-Octylphenoldiethoxylate (OP2EO). Other commonly known wastewater compounds, like DEET (insect repellent), caffeine (nonprescription drug), cotinine (nicotine byproduct), and triclosan (antimicrobial compound), were also detected in the wastewater but not in any of the springs. Bisphenol A (plasticizer polycarbonate plastics), which is ubiquitous in the environment (Foreman and others, 2012), was detected at multiple sites.

Wastewater and pharmaceutical compounds were only detected in the Monument Spring and Bright Angel Wash samples. Of the wastewater and pharmaceutical compounds analyzed, eight were found at Monument Spring (fig. 9A). Diphenhydramine (106 ng/L)³ and carbamazepine (91.3 ng/L)⁴ were the compounds with the highest concentrations at Monument Spring. In addition, metformin, fluconazole, bisphenol A, venlafaxine (antidepressant), sulfamethoxazole (antibiotic), and tramadol (opioid pain reliever) were detected at Bright Angel Wash and Monument Spring. Other than bisphenol A, no other sites had detections of wastewater or pharmaceutical compounds.

Of the seven pharmaceutical compounds detected in Monument Spring and Bright Angel Wash samples, the relative magnitudes (spring to wastewater concentration ratio) ranged from 0.01 to 0.74 (fig. 9B; table 5). Compounds closer to a ratio of 1 suggest less transformation from infiltration to issuance, and smaller numbers may indicate that compounds are degrading or adsorbing along the flow path. The sulfamethoxazole concentrations measured at Bright Angel Wash and Monument Spring were the most similar and had the highest relative concentration ratio of 0.74 (fig. 9B). In Monument Spring water, carbamazepine concentrations were high (91.3 ng/L) and in the middle range for the relative concentration ratio (0.17; fig. 9B). The diphenhydramine was the highest concentration (106 ng/L) found in Monument Spring water, but the second-lowest relative concentration ratio (0.05). Tramadol had the second-highest relative concentration ratio at Monument Spring (0.33), but fluconazole (32.6 ng/L) and metformin (7.58 ng/L) had higher overall concentrations in Monument Spring than tramadol (4.30 ng/L).

Stable Isotope Ratios

Stable oxygen (δ¹⁸O) and hydrogen (δ²H) isotope ratios for groundwater samples near the South Rim of the Grand Canyon fall within the range of winter precipitation values, meaning most ratios plot left of −11.0 δ¹⁸O and below −85.0 δ²H (Monroe and others, 2005; Solder and others, 2020) in figure 10. The stable δ¹⁸O and δ²H isotope ratios for waters sampled in April 2021 and those from the historical datasets plotted to the right of the local and global meteoric water lines, suggesting the water may have undergone evaporation either during precipitation or as it moved through the unsaturated zone toward the water table (Beisner and others, 2023a). There are two general groups that roughly coincide with the Bright Angel Fault, which is more apparent when sites are plotted as average ratios (fig. 10B). Although the Bright Angel Fault is mostly aligned with Garden Creek, sites in the Garden Creek watershed and in the Pipe Creek and Burro Creek watersheds tend to cluster more than sites west of the fault in the Horn Creek, Salt Creek, Monument Creek, and Hermit Creek watersheds, which indicates less variability among the former. The Rowe Well sample had the most enriched stable isotopic ratios, but that is also likely an evaporated sample because the water resides in a hand-dug well that is likely disconnected from the larger aquifer or related to a locally perched aquifer. Roaring Springs and water from the Garden Creek Pump Station, which is Roaring Springs water that is stored until transferred for delivery to the South Rim, had a similar depleted isotopic ratio and plotted to the left of Bright Angel Wash in figure 10B. Bright Angel Wash represents a mixture of Roaring Springs water and a more enriched ratio from the SRWTP’s treated wastewater, which is more evaporated. The δ¹⁸O ratio for Bright Angel Wash wastewater samples has been more variable over time, which is expected owing to seasonal fluctuations of the treated wastewater volume. Other sites with variable δ¹⁸O ratios were the Monument, Horn, and Salt Creek Springs (table 6). The δ¹⁸O ratios of the Pumphouse and Pipe Springs were the least variable or had the lowest standard deviation.

Temporal trends of δ¹⁸O and δ²H ratios were analyzed at each site using a nonparametric Mann-Kendall Test (fig. 10A; table 7). Samples collected at Monument Spring before 2000 have plotted between the two clusters of sites, but samples collected after 2000 have been trending negatively over time (fig. 10B). The trends of the Monument, Salt Creek, and Horn Springs were more negative in samples collected in April 2021 than in the historical data. Three sites had a significant trend in one or both of the δ²H and δ¹⁸O ratios over time. The Pumphouse (Two Trees Spring) δ¹⁸O ratio trends more negative over time, while Garden Spring δ²H ratio is trending more in the positive direction over time (table 7). Monument Spring trends to a more depleted signature over time (both δ¹⁸O and δ²H ratios are negative; table 7), which is in the direction of the Bright Angel Wash ratios (table 7).

Nitrate, Chloride, and Boron

Nitrate concentrations ranged from 0.355 to 6.10 mg/L for spring samples collected in April 2021. All spring sample concentrations of nitrate were below the EPA’s drinking water MCL of 10 mg/L (U.S. Environmental Protection Agency, 2023); Monument Spring had the highest concentration of the spring samples at 6.10 mg/L, and Bright Angel Wash wastewater had the highest nitrate concentration of any sample at 59.9 mg/L, exceeding the EPA drinking water MCL.

Historically (1992–2021), the WQP nitrate concentrations measured at Bright Angel Wash have been consistently high, averaging 28.1 mg/L. Although only half the concentration measured in the April 2021 sample, this is still well above the EPA drinking water MCL. Nitrate concentrations at Monument Spring and Creek (1992–2021) have consistently been an order of magnitude higher than the mean concentration (0.63 mg/L) of all other springs (fig. 11; table 8). The nitrate concentration has been as high as 7.02 mg/L at Monument Spring. The concentrations at both Horn Springs (Upper Horn Bedrock and Horn East Alluvium) have also been historically higher than most other South Rim springs, ranging between 0.72 and 1.5 mg/L (2012–21). The WQP mean nitrate concentration at the Garden Spring location⁵ was similar to both Horn Spring sites and was double the concentration of the sample collected in the April 2021 study. The sample locations in watersheds west of Bright Angel Fault had a significantly greater median nitrate concentration compared to the watersheds east of the fault (Wilcoxon rank-sum test, p<0.0001; table 8).

Chloride data from the WQP (gathered between 1980 and 2022) for the area showed consistently high concentrations at Monument Spring (table 9). The mean concentration of 350 mg/L at the downstream Monument Creek location near the mouth of the Colorado River confluence was more than three times the mean concentrations at the upper watershed locations. The mean concentration measured at the downstream Monument Creek location was several times greater than concentrations measured at Bright Angel Wash (121 mg/L; fig. 12). Monument Creek near Colorado River had the only concentrations exceeding the EPA recommended maximum secondary drinking water value of 250 mg/L (U.S. Environmental Protection Agency, 2023). This is likely a result of the water’s flow path through geologic materials like the Tapeats Sandstone, which is a marginal marine deposit high in salts. The Horn, Salt, and Hermit Creeks also showed elevated concentrations of chloride (greater than the overall spring and creek median of 21 mg/L). Overall, chloride concentrations increased as water flowed from a spring’s source to its respective downstream creek location. This pattern is a result of water coming in contact with geologic materials and the subsequent dissolution of minerals, which are mobilized by the flowing water. The WQP sample locations in watersheds west of Bright Angel Fault had a significantly greater median chloride concentration compared to watersheds east of the fault (Wilcoxon rank-sum test, p<0.001; table 9).

Concentrations of boron followed a similar pattern to chloride concentrations: lower concentrations at the spring source and higher concentrations downstream at the spring’s respective creek (table 10). In the one Bright Angel Wash sample, the boron concentration was 120 micrograms per liter (µg/L), which was above the overall median of 92 µg/L but similar to several of the springs and multiple orders of magnitude lower than Monument Creek’s median concentration of 550 µg/L. Hermit Creek and Monument Springs median concentrations were around 30 µg/L, but concentrations measured at those springs were 2 to 5 times higher near the mouths of their respective creeks downstream. Boron concentrations measured at Monument Spring increased from a median concentration of 39 µg/L to 180 µg/L at the trail crossing (approximately 1.5 miles downstream) and to 550 µg/L near the mouth at the Colorado River confluence. A similar pattern was observed at Hermit Creek. The WQP sample locations in watersheds west of Bright Angel Fault had a significantly greater median boron concentration compared to the watersheds to the east (Wilcoxon rank-sum test, p<0.01; table 10).

Gadolinium

The growing use of Gd-based contrast agents in MRI (used in healthcare applications) has led to an increasing input of Gd-anth into natural environments (Brünjes and others, 2016), which makes it a useful tracer. Gd-based contrast agents for MRI were first introduced to the market in 1988, and their use for MRIs has been increasing ever since, leading to the ubiquitous presence of Gd in freshwater environments. Because of the high stability of these contrast agents, they are readily passed through human bodies and then through conventional wastewater treatment plants, resulting in positive Gd anomalies in aquatic systems that receive treated effluents (Verplanck and others, 2010; Brünjes and others, 2016; Hatje and others, 2016; Boester and Rüde, 2020).

The WQP data retrieval and results from the April 2021 sampling show relatively similar Gd concentrations between station groups. The WQP results from Garden Spring location⁶ and Monument Spring have historically had the highest concentrations (greater than 0.01 µg/L). The wastewater in Bright Angel Wash had the highest concentration in this study (0.05 µg/L; fig. 13A). After normalizing the concentrations to a geologic reference, the anomaly patterns were visible at Bright Angel Wash. The Horn, Salt, and Monument Springs had slight upticks in Gd for the REE plots, but overall small enough to be inconclusive (not graphed). Although not as pronounced as Bright Angel Wash, one of the Rowe Well samples contained an anomaly for Gd. This is likely a false signature caused by evaporation in the hand-dug well (see “Description of Sites” section). Differences in the patterns were minimal in the two geological references used for the normalization. Brünjes and others (2016) also noticed minor effects when testing different normalizing approaches for several references.

Plotting the ratio of normalized Gd to an interpolated Gd-geo (or background level; Verplanck and others, 2010, eq. 1) showed higher ratios for the April 2021 sampling at Bright Angel Wash and Monument Spring samples, but less so for the historical samples (fig. 13B). Gd anomalies greater than 1.0 can indicate that Gd-anth is present in the sample (Brünjes and others, 2016). However, this threshold can be misleading, and some investigators have used Gd anomaly thresholds up to 1.5 to avoid overestimation of Gd-anth (Rabiet and others, 2006; Verplanck and others, 2010). The samples from the WQP dataset at Horn Creek, Salt Creek Spring, and the Monument Spring and Creek show elevated distributions of ratios above 1.5.

Streamflow

Streamflow data were compiled from USGS streamgages (continuous stage-discharge relation [1994–2022]; figs. 14A–D; table 11), and the historical discrete discharge data were compiled from the WQP (includes USGS discrete measurements [1983–2009]; figs. 15A–C; table 12). The data were then analyzed for trends or patterns that might indicate a relation or shift in wastewater influence over time. No obvious patterns or statistical trends in discharge were observed for most of the measured creek locations over the period of record. The creeks draining their respective springs show different patterns in the timing, volume, and variability of flow.

The continuous streamgages operated during a similar 2-year period for Garden and Pipe Creeks (figs. 14A,14B). Pipe Creek appeared to be more seasonally responsive to snowmelt in the spring months of March through April (fig. 14B). Garden Creek stations looked to be most affected by fall precipitation (late season convective storm activity; fig. 14A). The highest median monthly flow in Garden Creek occurred in September. Also unique to Garden Creek was the timing of its lowest median monthly flow. It was much earlier in the season compared to the other creeks, occurring in May.

The mean discharge at stations associated with Garden Creek and Hermit Creek was greater than 300 gal/min (figs. 14A, 14D; table 11). This is an order of magnitude greater than the continuous streamgages at Pipe Creek and Pumphouse Wash. Discrete measurements from the Horn, Salt, and Monument station groups indicate the lowest discharge rates were from the Horn Creek and Salt Creek stations (median discharge less than 1 gal/min; figs. 15A, 15B). Seasonal characteristics at these station groups are mostly consistent: higher mean flow occurred in March through May, and lower mean flow occurred in August and September. Flood events associated with monsoonal convective storms were recorded at Garden Creek, Hermit Creek, and Pumphouse Wash in late summer and fall. Overall, the majority of discharge at springs along the South Rim appears to be dominated by winter snowmelt and spring runoff. The Monument Creek watershed showed seasonal patterns consistent with those of the Horn Creek and Salt Creek watersheds, although data collection was infrequent (fig. 15C). Of the continuous streamgages, Garden Creek had variable discharge or the highest standard deviation of monthly flow (140 gal/min), followed by Hermit Creek (48 gal/min; table 11). Monument Spring had the highest standard deviation of the discretely measured sites (51.2 gal/min; fig. 15C). The observed patterns have high uncertainty associated with them because of inconsistent measurements and limited overlapping period of record.

On average, the SRWTP discharges a total volume of 230 acre-feet per year into Bright Angel Wash. Discharge at the SRWTP is generally at its peak in July and August, when the greatest number of visitors are in the park. Discharge from the SRWTP into Bright Angel Wash was highest in 2015 and lowest in 2021 during the pandemic (total of 96.6 million gallons per year and 50.2 million gallons per year, respectively; Arizona Department of Environmental Quality, unpub. data, 2022). Between 2009 and 2022, median monthly discharge was between 5 and 8 Mgal/mo (Arizona Department of Environmental Quality, unpub. data, 2022), and Zukosky (1995) indicated that discharge in the early 1990s was at least 9 Mgal/mo but could be over 13 Mgal/mo.

Quality Assurance

Pharmaceutical and PFAS compounds were not detected in the field blanks analyzed by the NWQL. Cholesterol, 2,6-Ditertbutyl 1-4 benzoquinone, and 4-t-OP3EO were present above the laboratory method detection limit in the field and lab blanks analyzed by IWCAL and were at similar concentrations to environmental samples. These three pharmaceuticals were not included in the discussion of anthropogenic compounds from this study.

The replicate sample collected at Bright Angel Wash in April 2021 showed good agreement between the original and replicate for most PFAS compounds: PFPeA (2 percent difference), PFHxA (4 percent), PFHpA (5 percent), PFOA (4 percent), PFNA (7 percent), and PFDA (0 percent). PFBA (58 percent difference) and PFOS (40 percent) had higher variability between the original and replicate samples, although both PFOS values were estimated concentrations below the laboratory minimum reporting limit but above the method detection limit (1.8 and 1.2 ng/L). Pharmaceutical replicate samples from NWQL showed good agreement with the original sample; there was less than a 10 percent difference between concentrations of most analyzed compounds. The following compounds had higher variability between the original and replicate samples: caffeine (16 percent difference), sulfamethoxazole (30 percent), pseudoephedrine plus ephedrine (11 percent), meprobamate (14 percent), fexofenadine (12 percent), and nordiazepam (31 percent). Of these, caffeine, sulfamethoxazole, and nordiazepam had values below the laboratory minimum reporting limit but above the method detection limit and may represent values with greater quantitative uncertainty. The Bright Angel Wash wastewater replicate samples from IWCAL also showed good agreement. There was less than a 10 percent difference between most compounds. The following compounds had higher variability between the original and replicate samples: DEET (38 percent difference), diphenhydramine (20 percent), galaxolide (17 percent), 4-NP2EO (14 percent), 4-t-OP1EO (11 percent). 4-t-OP2EO and 5-methyl-1-H-Benzotriazole had a value detected in the environmental sample but not in the replicate sample.

Ecological Considerations

Grand Canyon National Park has five life zones and a variety of diverse ecosystems within those zones because of an extreme elevation gradient (National Park Service, 2014). The South Rim ecosystems rely on much less precipitation than the North Rim ecosystems, and spring habitat within South Rim ecosystems is more sensitive and vulnerable to changes in water quantity and quality. Uranium and other trace elements are a concern (Beisner and others, 2017), but for the most part, the Grand Canyon National Park is absent of major contaminant issues. The wastewater influence from the South Rim is a unique situation in a hydrologically complex setting where some appreciable amount of wastewater is influencing at least one spring along the South Rim of the Grand Canyon. A consistent influence suggests that species relying on these spring habitats are potentially exposed to chronic low levels of CECs.

According to the following studies, this could potentially lead to greater exposure, biomagnification with increasing trophic levels, and possible generational effects: Giesy and Kannan (2001), Eggen and others (2011), Xie and others (2016), Blazer and others (2021), Fenton and others (2021), Brase and others (2022), Mahoney and others (2022), Pulster and others (2024). Most of the CECs measured at the springs during this study do not have drinking water standards or human health benchmarks, although these studies show some of the CECs detected can have endocrine-disrupting and physiological effects. Studies in laboratory animals have also demonstrated developmental toxicity, including neonatal mortality (Martin and others, 2004; Fenton and others, 2021; Blazer and others, 2021; Mahoney and others, 2022). Concentrations measured at the springs are mostly below the levels discussed in these studies; however, the conditions and species studied do not necessarily represent the unique habitat of South Rim springs. Furthermore, studies investigating CECs' effects are generally conducted in a controlled setting using lethal concentrations, and this does not represent species or environmental conditions found in the springs of the South Rim. The ecological considerations presented in this investigation are a way to provide a general characterization of harmful CECs and a discussion of levels shown to have effects.

Since the early 2000s, the bioaccumulation of PFAS in aquatic organisms has raised concerns about effects in the food web and potential effects on humans and wildlife (Giesy and Kannan, 2001; Pulster and others, 2024). PFAS compounds have been shown to bioaccumulate in aquatic invertebrates and vertebrates (Blazer and others, 2021; Brase and others, 2022; Mahoney and others, 2022), and PFAS are found in human and animal blood where they bind to serum proteins and organs such as liver and kidneys, as well as in muscle tissue. This is in contrast to other persistent legacy contaminants that accumulate in fatty tissue, like dichlorodiphenyltrichloroethane, polychlorinated biphenyls, polycyclic aromatic hydrocarbons, and polybrominated diphenyl ethers. Most research on the effects of PFAS in the environment has focused on the two perfluoroalkyl acid groups of PFAS: perfluoroalkane sulfonic acids and perfluoroalkyl carboxylic acids. This includes the most extensively studied—PFOA, PFOS, PFHxS, and PFNA—although there are thousands of PFAS compounds manufactured and many already in the environment that have not been studied (Houde and others, 2011).

Jantzen and others (2016) exposed embryonic Danio rerio (Hamilton, 1822; zebra danio) to acute exposures (2 micromoles, about 1,000 ng/L) of PFOS, PFNA, and PFOA. The study demonstrated that acute, embryonic exposure (5 days) to individual PFAS results in significant biochemical and behavioral changes in young adult zebrafish 6 months after exposure, leading to the conclusion that these three PFAS have long-term and persistent effects following short-term embryonic exposure that persist into adulthood. Similarly, Menger and others (2020) exposed zebra danio to several PFAS and observed altered swimming behavior in zebra danio embryos exposed to PFOS, PFOA, and PFNA, but found PFPeA, PFHxA, and PFBS did not cause any changes in behavior. A review by Banyoi and others (2022) looked at roughly 61 experimental publications and concluded that PFAS in concentrations of 13,500 ng/L have adverse effects on body size variables for secondary consumers like fish; however, there were no significant effects on fish fecundity or liver or gonad somatic indices. The review did reveal that there are large research gaps for PFAS effects on different organisms in aquatic ecosystems at environmentally relevant concentrations. The concentration of PFOA found in the Bright Angel Wash wastewater was 84.6 ng/L, and Monument Spring water had a concentration of 20.8 ng/L. This is two to three orders of magnitude lower than the effect levels reported in these studies.

Scientific studies on the ecological effects of PFAS and PFAS in aquatic environments are relatively limited in number, and most studies are toxicological in nature, testing the upper bounds of levels, which do not always address ecologically relevant concentrations or the broader implications for ecosystem health. There are thousands of PFAS chemicals about which there is little or no scientific knowledge about toxicity and environmental risks (Pulster and others, 2024). Meanwhile, new chemicals are continually being developed and released into the environment (Buck and others, 2011). Although several PFAS have been banned, there is no clear indication that concentrations of PFOS and PFOA are declining in the environment or in the aquatic organisms exposed to them (Pulster and others, 2024). New emerging short-chained variants like GenX, PFBS, and PFBA are replacing the longer-chain chemicals, but there is a lack of research and regulation for these, and as a result, no consistent monitoring (Xu and others, 2021). The effects of PFAS from exposures possibly experienced by aquatic organisms in Monument Spring are unclear because there are no studies looking at low levels in semiarid aquatic ecosystems like the South Rim of the Grand Canyon.

Pharmaceuticals at low concentrations can cause both acute and long-term chronic effects on aquatic macroinvertebrates, fish, algae, and microbial communities (Pulster and others, 2024). These effects range from bacterial resistance to interference with the endocrine system of vertebrates. Pharmaceuticals are designed to bind to animal receptors and, as a result, interact with similar receptors in organisms exposed in the environment. Despite the low, below water-quality criteria concentrations measured in the April 2021 samples, some of the detected pharmaceuticals could still cause negative effects because of the complex interactions of environmental conditions and pharmaceutical mixtures in the environment.

Carbamazepine is an anti-epileptic and mood stabilizing drug that is used primarily in the treatment of epilepsy and bipolar disorder and is not considered to be a known endocrine-disrupting compound but has shown some anti-estrogenic activity (Biau and others, 2007). Studies have shown that exposure to concentrations in aquatic systems similar to those measured at Bright Angel Wash (399–529 ng/L) can have detrimental effects. Jarvis and others (2014) studied the effects of exposure to two environmentally relevant concentrations (200 and 2,000 ng/L) of carbamazepine on macroinvertebrate communities and determined the drug may alter freshwater community structure and ecosystem dynamics. Triebskorn and others (2007) found the lowest observed effect concentration of 1,000 ng/L in the range of environmentally relevant concentrations to negatively affect liver, kidney, and gill development of Oncorhynchus mykiss (Walbaum, 1792; rainbow trout) and Cyprinus carpio (Linnaeus, 1758; common carp).

Haack and others (2012) demonstrated that sulfamethoxazole concentrations between 240 and 520 μg/L—four orders of magnitude greater than Bright Angel Wash concentrations—may influence ecological function through changes in community composition and could promote antibiotic resistance through selection of naturally resistant bacteria.

Diphenhydramine, an antihistamine, can bioaccumulate in fish. Xie and others’ (2016) study demonstrated evidence of oxidative stress and suppressed swimming activity and feeding rates of Carassius auratus (Linnaeus, 1758; crucian carp) at concentrations of 27,100 ng/L of diphenhydramine. The lowest concentration tested, 840 ng/L, had no discernable effect from the control. Bright Angel Wash had a concentration of 2,349 ng/L. The results confirm that diphenhydramine can accumulate, be metabolized in fish, and exert a negative effect at different levels of biological organization (biochemical and behavioral responses).

When fish larvae were exposed to venlafaxine, an antidepressant, behavioral changes like a decrease in predator avoidance and a reduction in total escape performance were observed at the lowest experimental dose concentrations of 500 ng/L (Painter and others, 2009). At Bright Angel Wash, the concentration was 23 ng/L. Parrott and Metcalfe (2017) observed changes in nest-defense behaviors of Pimephales promelas (Rafinesque, 1820; fathead minnows) at concentrations of 88,000 ng/L and no effects at 880 ng/L, an environmentally relevant venlafaxine concentration (level common in wastewater).

Niemuth and Klaper (2015) demonstrated the potential endocrine-disrupting effects of metformin, a diabetes medication. Fathead minnow were exposed to 40,000 ng/L of metformin for a full life cycle. Niemuth and Klaper (2015) observed intersex conditions, reduced fecundity, and reduction in overall size of male fish. The concentration at Bright Angel Wash wastewater was 992 ng/L.

Tramadol, an opioid pain reliever, was studied in common carp and zebra danio by Sehonova and others (2016). They found the lowest observed effect concentration to be 10,000 ng/L of tramadol hydrochloride; this had significant effects on hatching, early ontogeny, morphology, histopathology, and morphometric and condition characteristics. When Buřič and others (2018) exposed Procambarus virginalis (Lyko, 2017; marbled crayfish) to concentrations as low as 100 ng/L for seven days, there were significant differences in behavior patterns, such as an overall decrease of velocity and distance moved during the exposure and an increase in the time spent in shelters. The concentration at Bright Angel Wash was 13.2 ng/L.

Azole fungicides (for example, fluconazole) are potent endocrine-disrupting chemicals for fish. The chemicals have anti-androgenic effects and disrupt steroidogenesis. Concentration effects have been observed multiple orders of magnitude larger than levels found in Bright Angel Wash (570 ng/L; Pacholak and others, 2020). Their study found fluconazole increases bacterial oxidative stress and promotes nanoscale modification of bacterial cell walls, indicating the presence of azoles (fluconazole) may be harmful to environmental bacterial strains.