Purpose and Scope

The purpose of this report is to evaluate the concentrations and sources, in both water and bed material, of fecal bacteria in the impaired reach of the Rio Grande. Spatial and temporal variabilities were analyzed, and instantaneous loads were calculated. Learning more about the extent of fecal-origin bacteria and its sources during the dry season will enable decision makers to focus best management practices to reduce E. coli concentrations from both point sources (discrete and discernible) and nonpoint sources (diffuse) in the Rio Grande watershed.

Study Area

The study area is in the unincorporated South Valley of Albuquerque, New Mexico (figs. 1 and 2), which is more agricultural than the metropolitan area of Albuquerque and has been for over 400 years (Visit Albuquerque, 2021). Historically, the southern part of the South Valley is more rural, and residences rely on groundwater wells and septic systems instead of city water and sewer systems. The current septic use based on Bernalillo County wastewater atlas data is estimated to be approximately 2,100 permitted systems in the South Valley and approximately 980 unpermitted systems (D. McGregor, Bernalillo County, written commun., 2022). Five water-quality sites located within two impaired AUs along the Rio Grande were sampled in this study (table 1, fig. 2).

Watershed Characterization

Hydrologic unit codes (HUC) are units by which the watersheds of the United States are classified. The HUCs are defined at different levels with smaller HUCs nested within larger HUCs, from the smallest geographic area (cataloging unit) to the largest geographic area (region) (USGS, 2022a). Albuquerque and the South Valley are within the HUC 8 watershed 13020203, the Rio Grande-Albuquerque watershed, with an area of 3,216 square miles (mi²) (fig. 1). The HUC 10 watershed where the study sites are located is HUC 1302020303 (City of Albuquerque-Rio Grande, 269 mi²). Directly upstream from the sites is HUC 1302020301 (Arroyo de Las Calabacillas-Rio Grande, 330 mi²) (fig. 3). HUC 1302020302 (Tijeras Arroyo, 132 mi²) feeds into the Rio Grande between the Above Tijeras and Below Tijeras Arroyo sites (figs. 2 and 3).

The three HUC 10 watersheds contributing to the study area vary in land cover (fig. 3; Dewitz, 2021). The Arroyo de Las Calabacillas-Rio Grande watershed, located directly upstream from the sites, contains 87 percent shrubland and is mostly undeveloped. The Tijeras Arroyo watershed contains approximately 94 percent shrubland and forested land. The City of Albuquerque-Rio Grande watershed is 34 percent developed and 47 percent shrubland with a small component (4 percent) of pasture, hay, and crops.

There are several potential point and nonpoint sources of E. coli in the South Valley (fig. 4). According to the EPA, sources of fecal indicator bacteria can include leaking septic systems, wastewater treatment plants, stormwater runoff, and domestic animal and wildlife waste (EPA, 2021b). On the basis of locations of existing water infrastructure, the density of water infrastructure is lower in the southeastern portion of the study area (M. Haley, Albuquerque Bernalillo County Water Utility Authority, written commun., 2021). If there is less water infrastructure in the area, it is more likely that residences will be on septic systems. LaMotte (2018) used 1990 and 2010 census data to estimate the number of people on septic systems in the coterminous United States on the basis of census blocks from the 2010 census. In the South Valley, the estimated number of people on septic systems was 13,320, as compared to the total population of 46,807 (fig. 4). However, Bernalillo County estimates that number to be much lower (approximately 3,000 systems) after initiatives to transition residences to municipal wastewater lines in the late 2000s (D. McGregor, Bernalillo County, written commun., 2022). Additional nonpoint sources similar in signature to leaking septic systems could include leaking sewer lines, sewer overflows into stormwater drains, illegal septage discharge from haulers into drains or the river, or public defecation into drainages in the study area.

There are several point sources within the study area. A wastewater treatment plant (WWTP) is located along the Rio Grande upstream from the confluence of Tijeras Arroyo (figs. 2 and 4), which allows for a 30-day average of 47 cfu/100 mL and a daily maximum of 88 cfu/100 mL for E. coli (NMEDSWQB, 2019c). The WWTP is the largest wastewater treatment plant in New Mexico, currently treating 50 to 60 million gallons of wastewater per day (Albuquerque Bernalillo County Water Utility Authority, 2023). A dog park is located just east of the Rio Grande and upstream from Bridge Blvd. Three pump stations convey stormwater from channels and storm drainage systems throughout the city and county into the Rio Grande. The pump stations primarily contribute to the Rio Grande during storm events. One pump station is directly upstream from Bridge Blvd and two other pump stations are downstream from Bridge Blvd, one near the Westside Community Center precipitation gage and the other upstream from and to the west of the Above Tijeras site. Both downstream pump stations are operated by the City of Albuquerque (COA). The Rio Grande and its surrounding bosque (a riparian cottonwood forest) are largely designated by the COA as open space and contain a large network of trails for hiking, including dog walking (City of Albuquerque Parks and Recreation Department Open Space Division, 2014). Bernalillo County Public Works has implemented several initiatives to reduce stormwater pollution, including an initiative to reduce dog waste in the watershed (Bernalillo County, 2022) that includes dog waste depositories across the bosque. Waterfowl fecal matter can be a source of E. coli, and the Rio Grande and bosque are also a major flyway for migratory birds. Many birds inhabit the area during the winter months (typically November through March) at places like Valle de Oro National Wildlife Refuge, which is located near the southern border of the study area (fig. 2; Sullivan and others, 2009; Herzenberg, 2018).

Hydrology

The Rio Grande-Albuquerque watershed is part of the Rio Grande Rift physiographic province, with mountains forming the eastern and western boundaries. To the west, the mountains are volcanic and to the east, the mountains are uplifted igneous and sedimentary geologic units. The valley floor consists of compacted sands and gravels of the Santa Fe Group, which is Paleogene in age and locally includes Quaternary aged alluvium (Natural Resources Conservation Service, 2002).

The Santa Fe Group aquifer is an unconfined aquifer that underlies the Rio Grande, and the depth to groundwater varies widely throughout the Middle Rio Grande Basin (fig. 1), a geologic basin that contains the study area (Bartolino and Cole, 2002). The Santa Fe Group thickness ranges from 1,400 to 14,000 feet (ft), and the depth to groundwater ranges from less than 2 ft near the Rio Grande, where groundwater and surface water are closely connected (Bartolino and Cole, 2002), to about 1,200 ft in areas west of Albuquerque.

In the Rio Grande-Albuquerque watershed, the Rio Grande is a highly regulated system with the flow controlled by dams and reservoirs upstream and downstream from the study area. Flow in the Rio Grande varies widely throughout the year, with the highest flows typically occurring during spring snowmelt (Bartolino and Cole, 2002). Discharges in 2020 and 2021 were historically low during the dry season and were often statistically below the 25th percentile (548 cubic feet per second [ft³/s]) of historical flows at USGS 08330000 (Albuquerque) (fig. 5; Jian and others, 2008). Discharges for 53 percent of the days during the dry season were below the 25th percentile from November 1, 2019, through June 30, 2020, and those for 63 percent of days during the dry season were below the 25th percentile from November 1, 2020, through June 30, 2021. When the flow is very low at USGS 08330830 (Valle de Oro), diurnal fluctuations based on the effluent discharge can be observed.

An intricate system of riverside drains, irrigation canals, and community irrigation ditches known as acequias line the Rio Grande. The irrigation canals and ditches are primarily used during irrigation season in places where water from the river is diverted by the Middle Rio Grande Conservancy District (Bartolino and Cole, 2002). The riverside drains parallel both sides of the river and capture lateral groundwater flow derived as seepage loss from the river and irrigation return flow (as shallow groundwater flow), which help keep the river at a stable level.

Within the study area, the main tributaries of the Rio Grande are the WWTP effluent outfall, which was considered a tributary inflow because the water is originally drawn from the Santa Fe aquifer system (fig. 4; Bartolino and Cole, 2002), and Tijeras Arroyo. With its 132 mi² drainage area, Tijeras Arroyo is one of the largest arroyos in the Albuquerque area and is intermittent in flow, typically during storms. At streamgage USGS 08330600 Tijeras Arroyo near Albuquerque, N. Mex., discharge was measured during only 40 days in 2021 (USGS, 2022b). Tijeras Arroyo serves as the primary channel for snowmelt and stormwater from areas east of the Sandia Mountains (fig. 1; City of Albuquerque Parks and Recreation Department Open Space Division, 2014). Although not widely reported on, Tijeras Arroyo does contribute a notable amount of sediment into the Rio Grande during storm events. During flash floods between 1937 and 1947 (Copeland, 1995), 26 suspended sediment samples were collected from Tijeras Arroyo. The average suspended sediment concentration for Tijeras Arroyo was 58,000 milligrams per liter (mg/L) at an average discharge of 300 ft³/s, and the average size class breakdown was 19 percent clay, 69 percent silt, and 12 percent sand.

Field Methods

Discrete surface-water samples, bed material samples, and field parameters were collected from five sites along the Rio Grande between April 2020 and August 2021 (table 1). Within the river cross section at each site, samples were collected from within 5 ft of the east bank, river center, and within 5 ft of the west bank to determine if there was heterogeneity within the sampling site. These east, center, and west sampling locations will collectively be referred to as “three-point sampling” at each site. This sampling method is atypical of the suggested bacteriological sampling method specified by the New Mexico Environment Department Surface Water Quality Bureau Standard Operating Procedure 9.1, where compliance sampling typically occurs at only one well-mixed location within the stream (NMEDSWQB, 2022b). Twelve sampling events occurred at each of the sites, six events in 2020 and six events in 2021. Four sampling events each year occurred during the dry season and two sampling events each year either occurred during the wet season or during the dry season at a discharge of greater than 1,000 ft³/s.

Dry season sampling criteria for this study have been set by evaluating (1) past sampling conditions in relation to precipitation and discharge and (2) the National Pollutant Discharge Permit for the WWTP (NMEDSWQB, 2019c). The dry season, as defined by the permit, is November 1–June 30; wet season sampling, which occurs after a rain event during the time period of July 1–October 31, is defined as occurring when rainfall is greater than 0.25 in., and an antecedent dry period of at least 48 hours after a rain event greater than 0.1 in. in magnitude (NMEDSWQB, 2019c). Given the criteria of the permit for the dry season and the peaks in flow in the Rio Grande after snowmelt over the past few years, dry season sampling occurred when the following criteria were met within the greater Albuquerque area:

USGS precipitation and streamgages within the study area were used to monitor the streamflow and rainfall sampling criteria.

Three-point sampling surface-water samples for E. coli, MST markers, and suspended sediment concentration were collected using the discrete or grab method. The USGS National Field Manual states that the isokinetic depth integrated sampling is the standard procedure and typically is the preferred method if the cross section is uneven in velocity and (or) depth (USGS, variously dated). Additionally, isokinetic samples are composited into a churn that is not able to be autoclaved, so bacteria samples are traditionally collected as a grab sample. However, Paretti and others (2018) directly compared samples collected using the isokinetic depth integrated sampling technique to grab samples and found similar E. coli and suspended sediment concentrations using both methods. Therefore, this study used the grab sample technique.

Water and bed material samples for E. coli and MST markers were collected by following the USGS protocols described in Myers and others (2014). Before going into the field for sample collection, bottle kits containing the appropriate sterilized 1-liter (L) wide-mouth polyethylene bottles and sediment jars were assembled. At a minimum, all sampling equipment was soaked in phosphate free detergent for 30 minutes and rinsed with tap water followed by deionized water (USGS, 2004). After the initial cleaning step, the E. coli sampling bottles were autoclaved, and the MST marker sampling bottles and jars were sterilized with a sodium hypochlorite solution, neutralized with a sodium thiosulfate solution, and rinsed with sterile deionized water (Myers and others, 2014).

To prevent contamination, the least amount of equipment was used to collect a sample, and sample collection followed the USGS National Field Manual for bacteria sample collection (USGS, variously dated). If the sites were safely wadable, the sampling bottle was submerged to allow the bottle to fill with the opening pointed slightly upward toward the current. If the sites were not wadable, the samples were collected either from a bridge (Bridge Blvd and Rio Bravo) using a DH–95 sampler and a bridge board, or, at the sites with no accessible bridge (Above Tijeras, Below Tijeras, and Valle de Oro), the grab samples were collected from a kayak. All water samples were collected with about 2.5–5 centimeters of headspace in each bottle to allow for proper mixing. The E. coli and MST marker samples were bagged immediately to prevent contamination and stored on ice in a cooler until analysis.

Bed sediment samples were analyzed for E. coli and MST markers at the Ohio Water Microbiology Laboratory (OWML) and for fine sand break at the New Mexico Water Science Center (NMWSC) Sediment Laboratory. For the E. coli and MST bed sediment collection, three sterile, plastic, wide-mouth 125-milliliter (mL) jars were used at each sampling location. For the fine sand break analysis, a 16-ounce plastic jar was used. The lidded sampling jar was lowered to the river bottom where the lid was removed to scoop bed sediment into the jar. The lid was then secured tightly on the jar before bringing the jar to surface. The jars were bagged immediately to prevent contamination and stored on ice in a cooler until analysis. When river stage was too high to easily sample by hand, a tool with a hose clamp attached to a long handle was used to hold the sample jar and scoop bed sediment from the river bottom.

Four known-source fecal samples were collected from fresh material and of known origin. Solid samples were collected for geese (n=2) and canine (n=1) sources. One human source, a wastewater influent sample, was collected from the WWTP. All known-source fecal samples were stored on ice immediately after collection and shipped to the OWML for processing within 24 hours. Known-source samples were only collected once throughout the study.

Physical parameters, including water temperature, pH, specific conductance, dissolved oxygen, turbidity, and barometric pressure, were recorded at each three-point sampling location within each site using either a Yellow Springs Incorporated 6920 series or 600 series multiparameter water-quality sonde (YSI Incorporated, Yellow Springs, Ohio) or an In-Situ Aqua TROLL 600 (In-Situ Inc., Fort Collins, Colorado). Following the manufacturer instructions and USGS methods, a calibration check was performed on the sonde each morning before collecting samples (USGS, variously dated). Field parameters, including turbidity, dissolved oxygen, specific conductance, pH, and water temperature, were measured by USGS staff. Two USGS streamgages, Albuquerque and Valle de Oro, provided continuous discharge throughout the study (table 1). Daily mean discharge from Albuquerque was used for the Bridge Blvd and Rio Bravo sampling sites. Daily mean discharge from Valle de Oro was used for the Above Tijeras and Below Tijeras sampling sites, as it reflects the daily effluent outfall fluctuations from the WWTP, which is upstream from Above Tijeras. At Valle de Oro, discharge was measured during every sampling date to check the continuous streamgage at the site. Table 2 shows the discharge measurements associated with each site for each sampling event. Discharge measurements were collected using standard USGS protocols as described in Rantz and others (1982), Nolan and Shields (2000), and Turnipseed and Sauer (2010).

Analytical Methods

The analytical methods for E. coli and MST in both water and bed material and suspended sediment analysis in water are discussed in this section.

Data Analysis Methods

The replicate pair variability was analyzed in different ways depending on the analyte. In describing the “two-range model,” Mueller and others (2015) state that over low concentrations, standard deviation (SD) is generally uniform, but at higher concentrations, relative standard deviation (RSD) is more uniform than SD. Paretti and others (2018) reported extensively on quality control and assurance methods for bacteria sampling methods and with a large amount of E. coli replicates, were able to analyze the variability in sampling methods using the bias-corrected log-log regression model (Mueller and others, 2015). Because of the lower number of replicate pairs and relatively higher concentrations for all parameters, RSD was used instead of SD to study variability for all analytes,

For E. coli and suspended sediment, which had some lower concentrations, SD was also calculated. Results for many analytes were censored (that is, less than the reporting level) in one or both replicate samples, which excluded them from replicate analysis.

Instantaneous or mean daily loading rates, also known as flux, for E. coli and MST markers were calculated using a simple approach (Meals and others, 2013) of multiplying concentration times discharge and were reported in units of MPN or copies per day,

The mean daily flux is the instantaneous rate at which the load passes a point in the river. This flux is not equal to the load, which is the mass of a substance passing through a specific point in the river. The load is calculated over a specified timeframe with frequent monitoring. Because of the infrequency of the sampling and targeting of dry season, lower-flow events, accurate loads cannot be calculated.

To categorically compare fluxes of different microbiological constituents at a single site and across multiple sites, the mean daily fluxes were divided into five bins (Bushon and others, 2017). For each constituent, regardless of flow conditions, all results were combined, ranked, and binned. All nondetects were assigned to the 0 bin. Bins 1–4 contained an equal number of sample results per bin, with the highest fluxes placed in bin 4. Table 3 lists the ranges of daily fluxes per bin for each constituent. Once the bins were established, each sample result for a given constituent was assigned a bin number that was then separately averaged for each site for dry and wet season samples. Average bin scores can be used to compare the fluxes of a single constituent across sites and to compare the fluxes of different constituents at a single site.

Statistical analyses were conducted in the R statistical computing environment (R Core Team, 2022). Because a high percentage of MST marker data were censored values, summary statistics were calculated for censored data using the “censtats” function in the Nondetects and Data Analysis (NADA) software package, and the summary statistics using the regression on order statistics method were reported for this study. Boxplots were created by using the “cenboxplot” function from the NADA software package (Lee, 2020) in the R statistical computing environment (R Core Team, 2022). Boxplots for analytes with censored data were grouped by sample type; percentiles were estimated by using the robust regression on order statistics method (Helsel, 2012). Outlier data points on boxplots were defined for this study as greater than 1.5 times the interquartile range. To evaluate correlations, Pearson’s r and Kendall’s tau were calculated for uncensored data, and only tau was calculated for censored data. All E. coli concentration data were log10 transformed before conducting the Pearson’s r calculations because the bacteria data were not normally distributed. Strong linear correlations of r = 0.9 or above typically correspond to tau values of about 0.7 or above (Helsel and others, 2020). For uncensored data, the base R function “cor.test” was used to determine correlations using both Pearson’s r and Kendall’s tau, and for censored data the “cenken” function was used from the NADA package to determine Kendall’s tau (Lee, 2020). When evaluating differences between the datasets, Kruskal-Wallis rank sum test was used for uncensored data using base R function “kruskal.test,” and when a probability value (p-value) was less than 0.05, the differences were found to be significant. If significant differences were found, then to determine if three-point sampling locations within sites differed from others for the uncensored data, Dunn’s test using the Benjamini & Hochberg correction method was used using the R package “PMCMRplus” and its “kwAllPairsDunnTest” function (Pohlert, 2021). To evaluate differences between three-point sampling locations within sites for uncensored data, the “cen1way” function from the NADA2 package was used (Julian and Helsel, 2021), which performs the Peto-Peto nonparametric test of differences.

Physical Parameters and Suspended Sediment Characterization in Water

Physical parameters, including pH, specific conductance, dissolved oxygen, and turbidity, were evaluated at each sample site and three-point sampling location (fig. 6). There was minimal fluctuation in values between each of the three-point sampling locations at the Bridge Blvd and Rio Bravo sites but also between the Bridge Blvd and Rio Bravo sampling groups. As expected, the effluent outfall affected several of the physical parameters after it entered from the east bank upstream from the Above Tijeras site and continued downstream to the Valle de Oro site. Specific conductance increased from a median of 333 microsiemens per centimeter at 25 degrees Celsius (µS/cm at 25 °C) at the west bank to 587 µS/cm at the east bank at Above Tijeras, whereas the median ranged from 331 to 334 µS/cm at Bridge Blvd and Rio Bravo upstream from the effluent outfall. pH decreased at the east bank at Above Tijeras to a median of 7.5 as compared to the median from the upstream sites at Bridge Blvd and Rio Bravo, which ranged from 8.2 to 8.3. Median dissolved oxygen concentrations ranged from 8.1 to 8.3 mg/L upstream at Bridge Blvd and Rio Bravo versus the east bank of Above Tijeras where the median dissolved oxygen concentration was 7.8 mg/L. The median turbidity ranged from 53 to 64 formazin nephelometric units (FNU) at Bridge Blvd and Rio Bravo, and the median turbidity at east bank at Above Tijeras was 39 FNU. The influence of the effluent on the east bank samples continued to be visible at the downstream sites Below Tijeras and Valle de Oro, with medians of 49 and 75 FNU, respectively, and lower than the other three-point sampling locations at those sites.

Suspended sediment concentrations (fig. 7A) were similar at Bridge Blvd and Rio Bravo and at the west and center locations at Above Tijeras, with median values ranging from 138 to 177 mg/L. At the east bank at Above Tijeras, suspended sediment concentrations decreased to a median value of 103 mg/L. Suspended sediment concentrations increased along the west bank at Below Tijeras, with a median of 185 mg/L and a maximum value of 2,990 mg/L. Valle de Oro suspended sediment concentrations showed more consistent mixing in the cross section, ranging from median values of 136 to 198 mg/L.

Escherichia coli Characterization in Water

E. coli concentrations vary between the dry season and the rest of the year. Table 6 contains summary statistics for E. coli, including mean, median, maximum, and minimum values of all the data and only the dry season data. Figure 8A–E shows time series plots of E. coli and MST marker concentrations for each site and sampling location. Figures 9 and 10 show boxplots for all data collected at each site and sampling location and data collected during the dry season only, respectively. Median E. coli values ranged from 42 to 160 MPN/100 mL for all samples, and from 32 to 120 MPN/100 mL for dry season samples (table 6). The west sample location had the highest concentrations collected at Above Tijeras (during the dry season, fig. 10) and Below Tijeras (all data, fig. 9), which were an order of magnitude higher in concentration than those of other samples. These higher values during the dry season could be indicative of an increase of fecal contamination on the west side of the river during the event.

Correlation coefficients were computed to determine how well E. coli concentrations correlated with discharge, suspended sediment concentration, and turbidity (table 7). Relations between E. coli and the other constituents were not strong (Pearson’s r < 0.9 and Kendall’s tau < 0.7), which is likely because the majority of these samples were collected in the dry season, when E. coli concentrations aren’t likely to be due to increased flow, such as during storm events with greater discharge. The strongest correlation is between E. coli in water and turbidity.

Escherichia coli Mean Daily Fluxes

Mean daily fluxes were calculated for each E. coli sample collected from the three-point sampling events (fig. 11, table 8). The TMDL for E. coli for the study reach ranges from 5.27 × 10¹² cfu/100 mL/day in high flow conditions (2,960 million gallons per day or 4,580 ft³/s) to 1.89 × 10¹¹ cfu/100 mL/day in low flow conditions (106 million gallons per day or 164 ft³/s; NMEDSWQB, 2019b). Because there were limited E. coli sampling data, true loads could not be calculated, and fluxes (MPN/day) are the rates that have been calculated. Therefore, the TMDL and the fluxes cannot be directly compared.

Table 8.    

Escherichia coli (E. coli) fluxes for each site and event at the three-point sampling locations.

[Data from U.S. Geological Survey (USGS, 2021). Dates shown as month/day/year. Mean daily flux values are shown in E notation. MPN/d, most probable number per day; Blvd, boulevard]

Table 8.    Escherichia coli (E. coli) fluxes for each site and event at the three-point sampling locations.

Site name	Sample date	E. coli mean daily flux (billion MPN/d)
		East location	Center location	West location
Bridge Blvd	4/7/2020	5.70E+02	5.58E+02	3.87E+02
Bridge Blvd	4/27/2020	6.87E+02	6.64E+02	9.38E+02
Bridge Blvd	5/13/2020	1.66E+03	7.65E+02	6.88E+02
Bridge Blvd	6/10/2020	8.34E+02	6.77E+02	5.34E+02
Bridge Blvd	7/23/2020	5.73E+03	2.75E+03	3.05E+03
Bridge Blvd	8/31/2020	1.76E+03	9.13E+02	1.30E+03
Bridge Blvd	2/8/2021	1.07E+03	7.07E+02	9.19E+02
Bridge Blvd	3/3/2021	7.57E+01	1.78E+02	2.10E+02
Bridge Blvd	3/10/2021	1.08E+03	9.78E+02	7.69E+02
Bridge Blvd	5/18/2021	1.00E+03	9.10E+02	8.11E+02
Bridge Blvd	6/23/2021	1.67E+03	1.69E+03	1.08E+03
Bridge Blvd	8/16/2021	2.28E+04	5.89E+03	5.89E+03
Rio Bravo	4/7/2020	2.51E+02	2.73E+02	2.17E+02
Rio Bravo	4/27/2020	5.49E+02	5.03E+02	4.81E+02
Rio Bravo	5/13/2020	7.52E+02	6.50E+02	2.80E+03
Rio Bravo	6/10/2020	3.39E+02	4.04E+02	5.73E+02
Rio Bravo	7/23/2020	1.93E+03	2.53E+03	1.19E+03
Rio Bravo	8/31/2020	1.17E+03	1.04E+03	1.57E+03
Rio Bravo	2/8/2021	2.23E+02	5.12E+02	2.72E+02
Rio Bravo	3/3/2021	7.57E+01	1.78E+02	1.45E+02
Rio Bravo	3/10/2021	4.33E+02	5.82E+02	8.76E+02
Rio Bravo	5/18/2021	1.06E+03	9.24E+02	9.77E+02
Rio Bravo	6/23/2021	1.45E+03	1.54E+03	1.25E+03
Rio Bravo	8/16/2021	4.13E+04	1.54E+04	2.75E+04
Above Tijeras	4/7/2020	9.78E+02	5.62E+02	5.73E+02
Above Tijeras	4/27/2020	1.92E+03	1.32E+03	1.32E+03
Above Tijeras	5/13/2020	1.76E+03	7.69E+02	5.92E+02
Above Tijeras	6/10/2020	2.00E+03	7.00E+02	1.87E+03
Above Tijeras	7/23/2020	2.83E+03	2.47E+03	3.34E+03
Above Tijeras	8/31/2020	1.54E+03	1.20E+03	8.03E+02
Above Tijeras	2/8/2021	6.26E+02	3.80E+02	1.88E+02
Above Tijeras	3/3/2021	7.21E+02	2.93E+02	7.47E+01
Above Tijeras	3/10/2021	9.50E+02	6.08E+02	4.98E+02
Above Tijeras	5/18/2021	2.08E+03	7.61E+02	3.12E+02
Above Tijeras	6/23/2021	1.97E+03	1.18E+03	1.18E+04
Above Tijeras	8/16/2021	1.86E+04	5.67E+04	2.69E+04
Below Tijeras	4/7/2020	1.08E+03	5.51E+02	4.05E+02
Below Tijeras	4/27/2020	1.80E+03	1.15E+03	8.64E+02
Below Tijeras	5/13/2020	3.03E+03	1.51E+03	1.64E+03
Below Tijeras	6/10/2020	1.50E+03	4.12E+02	5.12E+02
Below Tijeras	7/23/2020	3.34E+03	1.82E+03	2.47E+03
Below Tijeras	8/31/2020	1.40E+03	2.01E+03	1.07E+03
Below Tijeras	2/8/2021	1.07E+03	9.08E+02	8.87E+02
Below Tijeras	3/3/2021	7.68E+02	5.77E+02	5.78E+02
Below Tijeras	3/10/2021	2.41E+03	1.45E+03	6.48E+02
Below Tijeras	5/18/2021	3.11E+03	1.84E+03	1.03E+03
Below Tijeras	6/23/2021	2.42E+03	1.97E+03	1.39E+03
Below Tijeras	8/16/2021	1.58E+04	2.36E+04	9.25E+04
Valle de Oro	4/7/2020	6.30E+02	4.61E+02	2.25E+02
Valle de Oro	4/27/2020	1.07E+03	9.24E+02	6.48E+02
Valle de Oro	5/13/2020	1.51E+03	1.64E+03	1.26E+03
Valle de Oro	6/10/2020	7.87E+02	7.12E+02	4.50E+02
Valle de Oro	7/23/2020	7.27E+02	3.41E+02	2.69E+02
Valle de Oro	8/31/2020	1.67E+03	1.34E+03	1.14E+03
Valle de Oro	2/8/2021	6.78E+02	6.66E+02	5.04E+02
Valle de Oro	3/3/2021	9.09E+02	4.33E+02	1.47E+02
Valle de Oro	3/10/2021	1.83E+03	1.46E+03	1.33E+03
Valle de Oro	5/18/2021	3.10E+03	1.60E+03	1.31E+03
Valle de Oro	6/23/2021	1.90E+03	1.40E+03	9.48E+02
Valle de Oro	8/16/2021	5.04E+04	5.34E+04	2.26E+04

Fluxes were consistently under 1.0 × 10⁴ billion MPN/day for the bulk of the study except for one event (fig. 11), which was a large storm event. That last storm event, which was in August 2021, had the greatest fluxes ranging from 5.89 × 10³ billion MPN/day to 9.25 × 10⁴ billion MPN/day.

Characterization of Escherichia coli and Particle Size Analysis in Bed Material

Samples for bed material analyzed for E. coli and particle size analysis were only collected during the dry season. The mean values for E. coli at all the sites ranged from 0.5 to 10 MPN/g_DW (Table 6). Unlike the dry-season E. coli concentrations in water where the maximum value was observed at the Above Tijeras West location, the maximum concentration for E. coli in bed material was observed at Above Tijeras East with 59 MPN/g_DW. There is little variation of E. coli in bed material between the sites (fig. 12). Six percent of bed material samples were below the detection limit. Above Tijeras showed the most variation between the west and east locations.

Approximately 78 percent of the samples (n=72) had no fine particles (USGS, 2021). Bridge Blvd and Rio Bravo showed no fine material, but the downstream sites had a small percentage (4 percent or less) of fine material. Particle sizes in sediment have been identified as explaining some of the E. coli variability in rivers. Pachepsky and Shelton (2011) found more studies reporting an increase in E. coli concentrations when there was an increased amount of clay and silt bed material present. Coarse sediments may not provide as much protection from the sunlight or protozoan grazing for E. coli to survive as would finer sediments (Pachepsky and Shelton, 2011). Because E. coli concentrations in bed material were detected at low concentrations and the bed material is primarily a sandy substrate during the dry season, it is likely that the substrate is not a suitable habitat that would allow for bacterial growth during the dry season. Sampling particle size in both bed sediment and suspended sediment and E. coli concentrations in bed material during or directly after a storm would provide insight about whether more fine-grained material is present in bed material that could contribute to a higher E. coli concentration in the water.

This sand/fine particle ratio was compared to historical bed sediment data collected at the Albuquerque gage (USGS site no. 08330000) to understand if this finding is similar to historical data in the river. The gage is upstream from the sampling sites and was used for comparative purposes, because bed sediment used for particle-size analysis has been collected for more than 20 years and provides a good description of the bed particle size distribution. The bed sediment data collected from 2019 to 2020 (USGS, 2021) for 0.125-, 0.25-, 0.5-, 1-, 2-, 4-, and 8-mm particle size analysis were analyzed, and the first sediment particle size to reach 50 percent of the sample was considered the average size. For the entire date range, the average particle size during these years only ranged from 0.5 to 1 mm. As defined by the Wentworth Scale, 0.5 to 1 mm is considered “coarse sand” (Wentworth, 1922).

E. coli concentrations in bed material had little to no correlation between E. coli in water, turbidity, and suspended sediment concentration on the basis of Pearson’s r and (or) Kendall’s tau (table 7). Of the two, Kendall’s tau is a more appropriate indicator of correlation for these parameters because the relation between E. coli in bed material and E. coli in water, turbidity, and suspended sediment concentration is not linear. The highest correlation observed was between E. coli in bed material and E. coli in water, with a Kendall’s tau of 0.24; however, it is well below the value of 0.7 and so does not indicate a strong correlation.

Microbial Source Tracking in Water and Bed Material

The human marker was detected in all but 3.7 percent of the 108 water samples collected (figs. 8–10). The human marker concentrations in water vary by location, increasing downstream from the effluent outfall (observed mainly in center and east locations of the Above Tijeras and Below Tijeras sites) and do not vary by season. Median values in water samples ranged from 5,100 to 75,000 copies/100 mL for all data collected (fig. 9, table 9), and 2,000 to 100,000 copies/100 mL for concentrations during the dry season (fig. 10). The eastern sample location at Above Tijeras had the highest concentrations of up to 200,000 copies/100 mL, and that sample had a corresponding E. coli concentration in water of 160 MPN/100 mL.

No statistically significant correlation was found between E. coli concentrations in water and the human marker concentrations in water for this study, which was focused on dry season sampling (table 7). Although elevated human marker concentrations could indicate increases in fecal contamination, that was not observed during this sampling. The qPCR methods used in this study do not distinguish between viable and nonviable Bacteroides bacteria from which the human marker originates, and it has been found that dead bacteria can be detected by the method up to 4 weeks after it has been killed (Nielsen and others, 2007). Therefore, a high human marker concentration does not always equate with a high viable E. coli concentration, as was the case downstream from the effluent outfall, where a large amount of nonviable human-associated markers were found with smaller concentrations of E. coli.

For bed material samples, human marker median values ranged from 48 to 330 copies per gram dry weight (copies/g_DW) (fig. 12, table 9). The maximum human marker observed in the bed material was 5,100 copies/g_DW at Below Tijeras East, with an associated E. coli concentration of 4 MPN/g_DW. There was also little correlation found between the E. coli concentrations in bed material and human markers in water (table 7).

The canine markers in water do not appear to vary by location and slightly decrease during the dry season (figs. 9 and 10; table 10). Of the 108 samples collected, 21.3 percent were below the detection limit. Median concentrations ranged from 4,800 to 7,400 copies/100 mL for all data collected (fig. 9) and ranged from 3,700 to 8,000 copies/100 mL during the dry season (fig. 10). The west sample locations at Above Tijeras and Below Tijeras had the highest concentrations (32,000 and 29,000 copies/100 mL, respectively) and corresponding E. coli concentrations of 4,100 and 14,000 MPN/100 mL, respectively.

Summary statistics could not be calculated for the canine marker in bed material samples using the NADA package, because more than 80 percent of the data were censored (Lee, 2020). Maximum and minimum values in samples where the canine marker was above the detection limit were variable. The concentrations ranged from 180 to 680 copies/g_DW. The maximum concentration of 680 copies/g_DW was observed at Below Tijeras West and had a corresponding E. coli concentration in bed material of 0.9 MPN/g_DW (USGS, 2021).

The waterfowl markers in water do not appear to vary by location and slightly increase during the dry season, during the months when the migratory bird population increases (figs. 8 and 9). Seventy-five percent of the 108 samples were below the detection limit. Median values ranged from 33 to 480 copies/100 mL for all data collected and increased to 100–780 copies/100 mL during the dry season (Table 11). The western sample locations at Below Tijeras had the highest concentration (51,000 copies/100 mL) and a corresponding E. coli concentration in water of 82 MPN/100 mL.

The waterfowl marker was not detected in 82 percent of bed material samples, and summary statistics could not be calculated using the NADA package (Lee, 2020). Maximum and minimum concentrations of the waterfowl marker that were above the detection limit were 1,100 and 130 copies/g_DW, respectively. Out of this range, the maximum concentration was the only value reported that was not qualified as an estimate with “b” (extrapolated below the limit of quantification but above the limit of detection). This sample, having a concentration of 1,100 copies/g_DW, was collected at the Below Tijeras site from the west bank location in February 2021 and had a corresponding E. coli concentration for this sample of 0.9 MPN/g_DW (USGS, 2021).

Mean daily fluxes were calculated for each MST marker in water collected from the three-point sampling events (table 12). These fluxes were then ranked and placed in bins and those bins were used to compare fluxes of different microbiological constituents at a single site and across multiple sites. The flux ranks are discussed below, and scores were used to compare results across sites.

Spatial and Temporal Evaluation

Data were evaluated to determine if there were any temporal (dry versus wet season) and spatial (sample site and location) differences using the Kruskal-Wallis and Peto-Peto test statistics (table 13). These statistical tests look for differences between centers of independent groups (Helsel, 2012;Helsel and others, 2020). The most notable seasonal differences were observed in E. coli concentrations in water, suspended sediment concentrations, dissolved oxygen, and turbidity, which is consistent with the wet season samples collected after storm events, which can contribute E. coli and sediment to the system. Seasonal variation was not evaluated in bed material because all bed material samples were collected in the dry season. When evaluating the most statistically significant differences between sampling locations (each three-point sampling location at a sample site was grouped and evaluated), specific conductance, pH, and the human marker had the most statistically significant differences, most likely caused by the influence of the effluent along the eastern bank of the river downstream from the effluent outfall (fig. 4).

The pH was significantly lower, and the specific conductance was significantly higher at several sites and locations downstream from the effluent outfall (fig. 6). The pH and specific conductance at Above Tijeras East were statistically significantly different from all of the other locations (pH p-values ranged from 0.0003 to 0.0128 and specific conductance p-values ranged from 0.0002 to 0.0078). This statistically significant difference continued downstream but began to dissipate through the cross sections at the Below Tijeras and Valle de Oro sites (fig. 6).

The human marker increases significantly at the effluent outfall (fig. 9). Above Tijeras East and Center differed significantly from the upstream sites and western sites downstream, with p-values ranging from 0.0051 to 0.0311. Although this difference was notable, there was no statistically significant increase of E. coli input into the system at the effluent outfall. As noted in the previous section, the analytical method does not distinguish between viable and nonviable bacteria (Nielsen and others, 2007). E. coli concentrations are a measurement of viable bacteria, but there was no statistically significant increase of E. coli concentrations in water noted throughout the study area since the study largely sampled during the dry season. Because the E. coli concentrations throughout the study reach did not have statistically significant differences, this would indicate that the effluent outfall was not a significant contribution of E. coli in the study area, and therefore the source of bacteria could be originating upstream from these sites.

The E. coli concentrations in bed material were statistically significantly greater at Above Tijeras East (p-values = 0.048), as compared to the other locations at Above Tijeras (fig. 12). The concentrations at Valle de Oro Center were significantly greater than those at the Above Tijeras Center and Above Tijeras West locations (p-values = 0.048) and those at Below Tijeras West (p-value = 0.048).

Strength of Bacteria and Potential Sources

Mean daily fluxes were calculated for E. coli and MST markers in water (tables 8 and 12). These fluxes were used to compare different microbiological constituents across multiple sites. The fluxes were ranked by site and placed into four equally sized bins with flux ranges for each bin listed in table 3 and the average bin scores listed in table 14 and figure 13. All results below the detection limit were placed into the 0 bin. The average bin scores listed in table 14 and illustrated in figure 13 can be used to compare the fluxes of a single constituent across sites and to compare the fluxes of different constituents at a single site.

The results of bin rankings indicate that for all samples, the E. coli mean daily fluxes were greatest at the Below Tijeras, Above Tijeras, and Valle de Oro sites (fig. 13). Average bin scores for the human marker were also highest at these sites (table 14), implying they are near a source of human fecal contamination. The average bin scores for the canine marker were highest at Bridge Blvd and Below Tijeras. Within each site and among the three MST markers, the canine marker had the highest bin score at Bridge Blvd and Rio Bravo, and the human marker had the highest bin score at Above Tijeras, Below Tijeras, and Valle de Oro. The waterfowl marker was detected less frequently than the human and canine markers during both seasons, and bins scores were lower for all samples and the dry season. During the dry season, the average bin scores for E. coli and the human markers were highest at the same sites as those for all samples. The canine markers had the highest bin scores at Bridge Blvd and Below Tijeras. The waterfowl marker was found more often during the dry season than the wet season, and bin scores were highest at Bridge Blvd and Rio Bravo. Within each site and among the three MST markers, the canine marker remained the highest bin score at Bridge Blvd and Rio Bravo, and the human marker had the highest bin score at Above Tijeras, Below Tijeras, and Valle de Oro.

The top potential sources of fecal bacteria are human and canine (table 15) based on a combination of known potential sources of fecal bacteria, percentage of detections of the source markers in both water and bed sediment samples, and analysis of the average bin scores for daily mean flux (table 15). Given that the analysis of known-source fecal sample from the wastewater influent from the WWTP resulted in detection of the canine marker in addition to the human marker, this could indicate that the canine marker could at least in part be from wastewater influence. This occurrence is not unexpected because fecal bacteria may be transferred between host species living in close contact (Field and Samadpour, 2007; Stewart and others, 2007; Roslev and Bukh, 2011). With the presence of many different potential sources (fig. 4), the high incidence of septic use (LaMotte, 2018) and general park and trail use (City of Albuquerque Parks and Recreation Department Open Space Division, 2014) may be nonpoint sources contributing to the reach during the dry season because there is little correlation between E. coli concentrations and discharge. Also, the lack of statistically significant spatial differences in E. coli concentrations between sampling locations may also indicate that there are no significant contributions of E. coli within the study area and contributions could be coming from one or more sources upstream from the study area.