Response of Water-Quality Indicators to the Implementation
of Best-Management Practices in the Upper Strawberry
River Watershed, Arkansas
Teresa Ruth Brueggen-Boman, Seo-eun Choi, and Jennifer Louise Bouldin
Southeastern Naturalist, Volume 14, Issue 4 (2015): 697–713
Full-text pdf (Accessible only to subscribers.To subscribe click here.)
Southeastern Naturalist
697
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
22001155 SOUTHEASTERN NATURALIST 1V4o(4l.) :1649,7 N–7o1. 34
Response of Water-Quality Indicators to the Implementation
of Best-Management Practices in the Upper Strawberry
River Watershed, Arkansas
Teresa Ruth Brueggen-Boman1,*, Seo-eun Choi2, and Jennifer Louise Bouldin3
Abstract - This 4-y study monitored 6 sites located in the upper watershed of the Strawberry
River, AR, where multiple types of best-management practices (BMPs) were implemented
in an attempt to alleviate the impacts of cattle grazing on adjacent waterways. The water-
quality variables we assessed included turbidity, total suspended solids (TSS), and
concentrations of NO2
-, NO3
-, PO4
3-, Escherichia coli, and chlorophyll-a. We calculated the
average annual sediment loading for comparison to published acceptable total-maximum
annual load (TMAL). The mean values detected for most parameters that we assessed
were within acceptable state limits and reference-stream values for the Ozark Highlands
ecoregion. Following BMP implementation, all sites showed significant increases in at least
one variable, and the concentration of E. coli for 3 of the sampling locations exceeded the
maximum allowable concentration. The estimated sediment loading was within the accepted
TMAL. We conclude that implementation of BMPs was not effective at improving water
quality during the time-frame of our study. Our results suggest that maintaining desired water
quality in this watershed may require the use of BMPs that are: (a) specifically targeted
to limit the parameters of concern (e.g., E.coli), and (b) implemented in specific locations
of concern rather than dispersed throughout the watershed. We also offer suggestions for
future studies of this type to improve the study design in an effort to more efficiently and
effectively determine the impact of BMP implementation.
Introduction
The United States Environmental Protection Agency (USEPA) (2009) ranked
agriculture as the primary source of non-point-source pollution impairing surface
waters in the US. Streams adjacent to and supporting agricultural practices have
been shown to have greater sediment loads than streams in less-disturbed watersheds
(Wohl and Carline 1996). Poor management of cattle grazing in riparian
zones can lead to significant stream impairment (Davies-Colley et al. 2004, Kauffman
and Krueger 1984, Line et al. 2000), and cattle trampling stream banks may
lead to increased erosion and sedimentation (Braccia and Voshell 2006) resulting in
measured increases in stream turbidity (Stednick 1991).
In addition to increased sedimentation, cattle’s direct defecation and urination
into waterways can cause increases in nutrient concentration, organic loads, and
1Missouri Southern State University, Department of Biology and Environmental Health,
3950 Newman Drive, Joplin, MO 64801. 2Arkansas State University, Mathematics and
Statistics Department, PO Box 70, State University, AR 72467. 3Arkansas State University,
Ecotoxicology Research Facility, 501 Iroquois, State University, AR 72467. *Corresponding
author - boman-t@mssu.edu.
Manuscript Editor: R. Eugene Turner
Southeastern Naturalist
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
698
bacterial contamination (Braccia and Voshell 2006). Larsen et al. (1994) determined
that instream bacterial contamination decreased by 95% when cattle were
prevented from defecating directly into the stream and were moved as little as 2.5
m from a waterway. Cattle walking through or standing in streams also re-suspend
bottom sediments into the water column. The concentration of Escherichia coli
(Migula) Castellani and Chalmers (an enteric bacterium) has been found to be
greater in bottom sediments than in overlying waters, and this bacterium can occur
in greater concentrations in overlying water when bottom sediments are disturbed
(Stephenson and Rychert 1982).
Best management practices (BMPs) may be necessary to limit the impacts cattle
have on water quality and instream communities. Exclusion fencing to keep cattle
from entering streams has been promoted as one method to protect riparian and
aquatic habitats (Platts and Rinne 1985). Other BMPs include providing cattle with
an alternate drinking source, shade sources, forage availability, controlled grazing,
and supplemental feeding (Agouridis et al. 2005).
Implementation of these BMPs, however, has shown mixed results. Several
studies have determined that when cattle are presented with an alternate water
source, the amount of time they spend in or around a stream is reduced (Godwin
and Miner 1996, Miner et al. 1992, Sheffield et al. 1997). Although Sheffield et
al. (1997) determined that significant improvements in water quality and reduced
sediment-loss from streambank erosion occurred after introduction of an alternate
drinking source, they also found an increase in dissolved nutrients. Line
et al. (2000) determined that alternate water sources alone were not enough to
reduce pollutant loads. There is sparse literature available that examines BMPs
involving cattle exclusion and use of riparian buffers or studies assessing the
effectiveness of other BMPs, such as watering facilities or improved forage
(Agouridis et al. 2005).
Cattle production is an important agricultural practice in Arkansas. Many cattleproduction
locations are adjacent to streams, including the Strawberry River and its
tributaries located in north-central AR. Here we report on various water-quality collections
in the Strawberry River Watershed (SRW) that tested for the effects of BMP
implementation. Our objectives were to: (1) monitor chemical and bacteriological
levels before, during, and/or following BMP implementation to assess the impact of
BMPs applied in the upper portion of the watershed; and (2) calculate the average
annual sediment load to compare it to the published total maximum annual loads
(TMAL) for the assessed stream reaches (USEPA 2006). Our overall goal was to
determine whether the implementation of BMPs protected the uppermost portion
of the SRW. We anticipated that eliminating the adverse effects of non-point-source
inputs would afford the opportunity to improve the aquatic conditions in the lower
portions of the watershed.
Field-site Description
The SRW is situated in north-central AR, with the headwaters located in
southern Fulton County (Fig. 1). The Arkansas Pollution Control and Ecology
Southeastern Naturalist
699
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
Commission (APCEC 2014) designated the Strawberry River as (1) an
extraordinary resource water; (2) an ecologically sensitive waterbody; and (3) a
natural and scenic waterway that is used for (1) primary contact; (2) secondary
contact; (3) domestic, industrial, and agricultural water supply; and (4) fisheries.
The Strawberry River supports a diverse community of aquatic macroinvertebrates
including 3 federally endangered and several ranked freshwater mussels
(Harp and Robison 2006). Additionally, one of the most diverse fish populations
in AR occurs in this system—over 100 species of fish have been recorded (Robison
and Buchanan 1992), including the endogenous Etheostoma fragi Distler
(Strawberry River Orange-throat Darter).
The upper portion of the SRW where this study took place is ~19,700 ha, and
the land cover is reflective of the watershed as a whole—predominately forest
(47%) and grassland (45%) (AWIS 2006). Forested-land cover in the upper SRW
decreased and anthropogenic land-use (e.g., pasture, cropland, urban development)
increased from 1999 to 2006 (AWIS 2006, Brueggen-Boman and Bouldin 2012).
In 2008, 2 headwater reaches were listed as not supporting aquatic life because of
excess turbidity from surface erosion, and were placed on Arkansas’ 303d list of
impaired waterbodies (ADEQ 2008). Uunpaved roads, streambank erosion, and
runoff from adjacent pastureland contribute to excess turbidity (ADE Q 2008).
Best-management practice implementation
Implementation of BMPs occurred throughout the upper watershed (Fig. 1). The
Fulton County Conservation District (FCCD) provided the locations—township,
range, and section—where BMPs were implemented. We converted these locations
to universal transverse mercator (UTM) coordinates using Earth Point software
Figure 1. Strawberry River watershed location in Arkansas (Arkansaswater.org 2010). The
inset shows the approximate locations of farms where best-management practices (BMPs)
were implemented, sampling sites, and confined animal feeding operations within the upper
watershed of the Strawberry River. BMP implementation took place December 2008–June
2011. LSR = Little Strawberry River, SR = Strawberry River, UP = upper sampling location,
MID = middle sampling location, and LO = lower sampling location.
Southeastern Naturalist
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
700
(Earth Point 2012). We defined the implementation point as the center-point of
the section identified and we accurately mapped the location point to within ~1.1
km. The exact points were not made available because of privacy concerns of the
BMP implementers.
At least 81 farms in the upper SRW implemented 1 or more BMPs (FCCD
2011). The BMPs implemented included installing cross-fencing and fencing to
exclude livestock from streams; planting pastures with native grasses; providing
livestock with watering tanks or ponds as alternate drinking sources in heavy-use
protection areas; undertaking prescribed grazing; and brush, pest, and nutrient
management (Table 1). If necessary, wells were installed to provide water as alternate
drinking sources. Pest management involved mowing fields to remove
undesirable plant species.
Monitoring sites
Members of the ANRC, Arkansas State University Ecotoxicology Research
Facility (ASU-ERF), and the FCCD selected the sampling sites (Fig. 1). The sampling
locations were chosen based on their accessibility and assumed proximity to
future BMP implementations. We monitored these sampling sites to determine the:
(1) baseline conditions and water-quality changes during implementation of BMPs,
and (2) water quality changes following BMP implementation.
Two sampling locations were located in the Little Strawberry River (LSR): LSR
upper (UP) and LSR lower (LO), separated by a stream distance of 2.74 km. We
selected 3 sites in the Strawberry River (SR): SRUP, SR middle (MID), and SRLO.
The stream distance between SRUP and SRMID was 3.17 stream km, and SRMID
and SRLO were separated by 8.86 stream km. The 6th sampling site was located
in an unnamed tributary (UT) that enters the SR between the SRMID and SRLO
sampling sites. The UT site was 2.01 stream km above the confluence with the SR
(Fig. 1).
Table 1. Extent of best-management practices (BMPs) implemented within the upper watershed of
the Strawberry River. Implementation took place between December 2008 and June 2011. Names
provided for implemented BMPs were designated by the US Department of Agriculture and Natural
Resource Commission (FCCD 2011).
BMP Total implementation Number of farms
Fencing 27,301 m 53
Exclusion fencing 14,621 m 9
Brush Management 1317 ha 42
Pasture Establishment 170 ha 20
Watering Tank 29 22
Pond 2 2
Well 3 3
Heavy-use impact area 0.08 ha 22
Prescribed grazing 5307 ha 81
Pest management 3472 ha 63
Nutrient management 7789 ha 63
Southeastern Naturalist
701
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
Methods
Sampling regime
We collected water samples semiweekly beginning 6 months before BMP implementation
(June–December 2008) and at monthly intervals during the BMP
implementation phase (January 2009–June 2011). We conducted 1 year of post-
BMP semiweekly sampling from July 2011 to June 2012.
Water-quality variables
We obtained samples within the water column in Nalgene® containers prepared
according to American Public Health Association (APHA 2005) protocols. At each
site, we collected 1 sample in a sterile bottle with headspace for bacteriological
analysis, filled 1 bottle without headspace for the chemical analyses., and filtered
another water sample through a 45-μm filter and froze it for the analysis of dissolved
nutrients.
We tracked the following water-quality measures over the full course of the
study: total suspended solids (TSS), turbidity, and dissolved nutrients (NO2
-, NO3
-,
PO4
3-). We measured turbidity using a Hach 2100P turbidimeter (Hach, Loveland,
CO) and dissolved nutrients with a LACHAT Quikchem 8500 Flow Injection Analysis
(FIA) automated nutrient analyzer (Lachat Instruments, Loveland, CO). We
assessed E. coli during BMP- and post-BMP implementation phases, and measured
chlorophyll-a (chl-a) in the post-BMP phase. All analyses were conducted according
to standard APHA methods (Table 2).
We determined stream-discharge volumes at the sampling sites in the fall
and spring, beginning in May 2009 and ending in May 2012 using a Marsh and
McBirney Model 2000 portable flow meter (Frederick, MD). We recorded bank
width every 5 m over a 50-m stream reach that we considered representative of the
sampling site, and made 10 equidistant depth measurements across a representative
section of the sampling location to determine average stream depth, The average
width and depth of the channel were used to create a stream pro file.
Data analysis
We excluded from analysis samples from the UT site because dry conditions
prevented us from collecting adequate samples there. The NO2
- results were below
Table 2. Parameters and American Public Health Association (APHA) method used to determine
effectiveness of best-management practices implemented in the upper Strawberry River Watershed
between May 2008 and June 2012.
Parameter APHA method Detection limit
Total suspended solids (mg/L) 2540D
Turbidity (NTU) 2130B
NO2- (mg/L) 4500-NO2–B 0.10
NO3- (mg/L) 4500-NO3–I 0.10
PO43- (mg/L) 4500-PO4-3-PG 0.01
Escherichia coli (CFU/100ml) 9222B
Chlorophyll-a (mg/m3) 10200H
Southeastern Naturalist
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
702
the detection limit (DL) for all but 1 sample; thus, we also excluded them from
our analyses. We used 0.05 for NO3
- mg/L, and 0.005 mg/L for PO4
3 when nutrient
values fell below the DL, which is a conventional method of replacing half-values
of the DL (Table 2; Kayhanian et al. 2002).
We tested for the significance of BMP effect and location effect for each river
using a 2-way multivariate analysis of variance (MANOVA). The model employed
took into account the sample-site location and the phase of BMP implementation
(pre- vs. post-implementation), as well as the interaction between the 2 factors and
4 variables (NO3
-, PO4
3-, TSS, and turbidity). We used these 4 variables because
they had been collected before and after BMP implementation. The model was as
follows:
NO3
- + PO4
3- + TSS + turbidity = overall mean + BMP effect + location +
interaction between BMP and location
We performed additional analysis if the location and/or BMP were determined
to be significant for a stream segment. We used the univariate 2-sample t-test (α =
0.05) or analysis of variance (ANOVA) (α = 0.05) for the LSR and SR sites, respectively,
for each variable. If we detected a statistical difference, we performed
a post-hoc Tukey’s comparison to determine which pairwise comparison(s) were
significantly different.
We compared the 4 water-quality variables to the acceptable/reference limits for
the Ozark highlands ecoregion (OHE) or the SRW (ADPCE 1987; APCEC 2014;
USEPA 2000, 2006) and previously published values for this area of the SRW
(ADEQ 2003). Exceedance of these values indicated system impairment. We compared
the turbidity values to the published all-flow (all month samples combined)
and base-flow (June–October samples) means, but could not compare them to the
allowable percent exceedance of these values because our methodology did not
meet the 24 monthly sample qualification (APCEC 2014).
We performed the univariate 2-sample t-test (α = 0.05) on the log-transformed
E. coli data to compare the results between during-BMP and post-BMP implementation
for each sampling location. We used the univariate 2-sample t-test (α = 0.05)
and ANOVA (α = 0.05) to compare sampling sites for the LSR and SR sites, respectively.
If there was a statistical difference, then we performed a post-hoc Tukey’s
comparison to determine which pairwise comparison(s) were significantly different.
We calculated the geometric means and 95% confidence intervals for reporting
purposes (Altman et al. 1983).
We calculated the percentage of E. coli samples that exceeded the maximum allowable
concentration (MAC) using individual sample criteria for the during-BMP
sampling phase by separating the primary (1 May—30 September) and secondary
(1 October–30 April) contact seasons. In this way, we were able to assess if samples
exceeded the MAC by >25% of the time, which is the allowable limit as designated
by the APCEC (2014). We conducted the same analyses for the post-BMP sampling
phase except that we used the MAC geometric-mean criteria (APCEC 2014). E. coli
values exceeding a 25% limit indicated that BMPs were not effective at limiting this
source of impairment to the systems.
Southeastern Naturalist
703
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
We employed a univariate 2-sample t-test (α = 0.05) and ANOVA (α = 0.05) to
compare mean concentrations of chl-a at sampling sites for the LSR and SR sites,
respectively. If there was a statistical difference, then we performed a post-hoc
Tukey’s comparison to determine which pairwise comparison(s) were significantly
different. A site with a significantly greater chl-a concentration indicated greater
primary productivity and nutrient availability. If chl-a concentrations were above
values for reference-stream data for the OHE, then it indicated atypical primary
productivity for the ecoregion and indirectly indicated elevated nutrient levels.
We used the univariate 2-sample t-test (α = 0.05) to compare discharge by
season—fall and spring—to determine if there were significant effects due to
season. If there were significant differences, we tested data for significant differences
among sites for fall and spring separately. If there were no effects of season,
we combined data and compared sites to identify significant differences. We used
the univariate 2-sample t-test (α = 0.05) and ANOVA (α = 0.05) to compare the
LSR and SR sampling sites, respectively. We used the discharge data to generate a
conservative estimate of annual sediment loading at the varying sample locations.
Total suspended solids values represented sediment concentration. We calculated
the available total-maximum daily loads (TMDLs) for the SR and LSR to obtain a
total-maximum annual load (TMAL). If the total annual loading at monitoring sites
exceeded the calculated TMAL, then we assumed that BMPs were not effective at
limiting sediment loading in the systems.
We transformed data for any variables that were not normally distributed. We
performed a nonparametric test on any variable not normalized through transformation.
We employed the Mann-Whitney U test (α = 0.05) for the LSR river
data and the Kruskal-Wallis test and the Dunn’s multiple comparison test for
SR data (α = 0.05).
Results
The results of 2-way MANOVA indicated that the interaction between the BMP
and location was not significant for either stream segment. Individually, location
was significant (P < 0.001; P < 0.047) and BMP was significant (P < 0.001; P less than
0.001) for the LSR and SR, respectively. The statistical tests based on sample location
as well as comparing pre- and post-BMP implementation periods indicated that
there were significant differences for most water-quality variables (Fig. 2).
Nutrients
The mean NO3
- concentration was significantly greater at the LSRUP site than
at the LSRLO site for all sampling phases. The mean NO3
- concentration was also
greater in the LSR compared to the SR. We detected no significant differences
between sites in the SR. We found significant increases in the mean NO3
- and PO4
3-
concentrations at 3 of the 5 sites (LSRUP, SRUP, and SRLO) pre- and post-BMP
implementation. The mean PO4
3- concentration increased significantly at the LSRLO
site, while the mean NO3
- concentration increased significantly at the SRMID site.
Most nutrient mean-concentration values fell within previously published criteria
Southeastern Naturalist
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
704
Figure 2. Mean ± standard error (SE) of nutrients, total suspended solids (TSS), and turbidity
prior to implementation of best-management practices (BMPs) (Pre), May–Dec 2008,
and following BMP implementation (Post), July 2011–June 2012, for the various sampling
locations in the upper Strawberry River Watershed. The geometric mean ± 95% confidence
interval of Escherichia coli during the implementation of BMPs (During), January 2009 to
June 2011, and post-BMP implementation. Values followed by different letters indicate significant
differences (P ≤ 0.05) between or among sampling locations within the same stream
segment only (i.e., the Little Strawberry River [LSR] or Strawberry River [SR]). * indicates
a significant difference (P ≤ 0.04) between BMP sampling time-frames in individual sampling
locations, univariate 2-sample t-test (P ≤ 0.05). † indicates that the post-BMP median
was significantly different (P ≤ 0.01) from the pre-BMP median, the original mean and SE.
UP = upper, MID = middle, and LO = lower sampling location.
Southeastern Naturalist
705
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
limits, but the mean NO3
- concentration we detected at the LSRUP in the post-
BMP phase exceeded the reference values for the OHE site (ADPCE 1987, USEPA
2000). Additionally, the mean NO3
- concentrations of pre- and post-BMP samples at
LSR sites and the post-BMP samples at SR sites exceeded the previously published
means of the sample sites in the upper SR W (ADEQ 2003).
TSS and turbidity
For the pre-BMP-implementation portion of the study, the means of all-flow and
base-flow turbidity values in the LSR were significantly greater at the LSRLO site
than at the LSRUP site. Conversely, on the SR, the mean all-flow and base-flowturbidity
values were significantly greater at the SRLO than at the SRUP. The mean
TSS values at LSRUP and LSRLO were not significantly different. The mean TSS
value was significantly lower at the SRLO site compared to the o ther 2 SR sites.
Following BMP implementation, the mean TSS value was significantly greater
at the LSRLO site than at the LSRUP site, and the mean TSS value at SR was
significantly lower than at SRMID or SRLO. Following BMP implementation the
all-flow and base-flow-turbidity values were significantly lower at SRUP than at
SRMID or SRLO. The mean turbidity values were significantly lower at SRMID
than at SRLO.
The results of the univariate 2-sample t-tests for pre- and post-BMP measurements
at individual sites indicated significant increases in the mean TSS
and turbidity at 3 of the 5 sites (LSRUP, LSRLO, and SRLO) in the post-BMP
implementation phase (compared to pre-BMP). The mean TSS values increased
at the SRUP site between pre- and post-BMP implementation, although we
detected no significant change in turbidity for this site. We found a significant
decrease in the mean base-flow turbidity at the SRMID site, although the
all-flow turbidity and mean TSS values were not significantly different. While
the mean TSS and turbidity values generally fell within previously published
criteria limits and/or reference-stream data, the 2 mean TSS values in the post-
BMP sampling phase (at LSRLO and SRLO) exceeded a previously published
reference value for the OHE (ADPCE 1987). The LSRLO post-BMP TSS mean
exceeded a previously published mean from a sample site located in the LSR in
the upper SRW (ADEQ 2003).
E. coli
We detected no significant differences in the geometric means of the E. coli
concentrations at site locations during the BMP-implementation phase. However,
in the post-BMP phase, the density of E. coli was significantly lower at LSRLO
compared to LSRUP and also at SRMID compared to SRLO. A significant increase
in E. coli density during-BMP to post-BMP sampling intervals occurred at 3 sites
(LSRUP, LSRLO, and SRUP).
During-BMP implementation, which occurred during the primary contact season,
33% of samples collected from LSRUP and 44% of the samples collected
from LSRLO exceeded the MAC (298 CFU/100 ml) (APCEC 2014). Twenty-eight
percent of samples from LSRLO and 29% from SRUP exceeded the MAC (126
Southeastern Naturalist
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
706
CFU/100 ml) (APCEC 2014) in the post-BMP sampling phase. All sites were within
acceptable limits during the secondary contact season.
Chl-a
The statistical analysis of the chl-a concentrations indicated that there were
significant differences among sampling locations in the post-BMP phase (Fig. 3).
The chl-a concentrations at LSRLO and SRUP were significantly lower than those
measured at LSRUP and SRLO, respectively. We detected the highest monthly
means in August for LSRUP (4.13 ± 0.94 mg/m3), SRLO (1.06 ± 0.35 mg/m3),
SRUP (0.15 ± 0.10 mg/m3), and SRMID (1.13 ± 0.44mg/m3). We measured the
highest monthly mean value for LSRLO (1.67 0.41mg/m3) in September 2011. The
only site monthly mean that reached or exceeded reference-stream criteria (2.67 ±
0.53 mg/m3) was the LSRUP site in the months of July (2.91 ± 1.38 mg/m3), August
(4.13 ± 0.94 mg/m3), and September (2.67 ± 4.80 mg/m 3) 2011.
Discharge
We employed a nonparametric approach to compare spring and fall discharge.
The median discharge was 0.39 m3/s lower in the fall, and was significantly different
from the spring discharge. We detected no significant differences among the
sampling locations in each river for either fall or spring median discharges. The
estimated annual sediment load was greatest at SRLO (Table 3). All estimated loading
values fell below the TMAL determined for the upper portion of the Strawberry
River and the Little Strawberry River.
Figure 3. Mean
Chlorophyll-a concentration
± standard
error (SE) for
the post-BMP implementation,
July
2011–June 2012.
The different letters
indicate the significant
differences (P
< 0.001) between/
among sampling locations
within the
same stream segment
only, for the
Little Strawberry
River (LSR) or the
Strawberry River
(SR). UP = upper,
MID = middle, and
LO = lower sampling
location.
Southeastern Naturalist
707
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
Discussion
The quantity and/or type of BMPs selected for implementation in this study
were not effective at improving water quality during the time frame within which
we conducted our assessment. We measured numerous water-quality variables
across multiple site locations and and found that values for many of them exceeded
published recommendations or historical values. It should be noted, however, that
the majority of the detected differences, measured over the entirety of this study,
were within limits deemed acceptable for the OHE.
The mean turbidity levels and the mean TSS concentrations were significantly
greater at the downriver sampling locations compared to in the upriver sample sites
within the same river. Our results agree with those of a previous study in the SRW
that found an increasing trend in a downstream direction for these 2 parameters
(ADEQ 2003). The findings of both studies suggest that attempts to decrease sediments
may benefit from BMP implementation because TSS and turbidity results
indicated a cumulative effect with tributary input and increased river size, but not
site-specific inputs. The mean concentration of TSS and turbidity at these lower
sampling sites exceeded the historic and/or reference-stream values (ADEQ 2003,
ADPEC 1987), and there were significant increases from the pre- to the post-BMP
sampling phase. Additional implementation of BMPs could benefit this region of
the SRW to alleviate the cumulative downstream effect.
Table 3. Median instantaneous discharge, median instantaneous-total suspended solids (TSS),
and estimated annual sediment load for sampling locations in the upper Strawberry River Watershed
from 2009 to 2012. LSR= Little Strawberry River, SR = Strawberry River, UP = upper,
MID = middle, and LO = lower sampling locations. Base-flow total maximum daily load (TMDL)
for Strawberry River 11010012-011 = 0.49 tons TSS/day (USEPA 2006); base-flow TMDL for
Little Strawberry River 11010012-010 = 0.23 tons TSS/day (USEPA 2006).
Total maximum
Median discharge Median TSS Median annual TSS annual load
Site (m3/sec) (mg/L) load (metric tons) (metric tons)
Little Strawberry River 85.10
LSRUP
Median 0.13 5.30 17.18
Range 0.01–0.51 2.1–15.6
LSRLO
Median 0.25 4.40 14.95
Range 0.01–0.72 0.5–19.6
Upper Strawberry River 178.24
SRUP
Median 0.26 2.80 17.58
Range 0.04–0.57 0.7–6.1
SRMID
Median 0.09 5.63 11.26
Range 0.01–1.22 0.87–10.9
SRLO
Median 0.85 6.30 72.01
Range 0.13–1.18 0.03–32.6
Southeastern Naturalist
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
708
We noted throughout the study that cattle continued to have unrestricted access
at 3 sampling sites. We saw hoof prints, excrement, or cattle in water at the LSRUP,
SRLO, and SRUP sites. These locations had the greatest estimated annual TSS
loading as well as the greatest E. coli concentrations in the post-BMP sampling
phase. Both the LSRUP and SRUP had geometric E. coli means that exceeded the
AR acceptable limits during the primary-contact season and had significant increases
in their TSS concentrations from the pre- to post-BMP sampling phase.
Turbidity also increased significantly at LSRUP. SRLO had both the greatest mean
TSS and significant increases in all variables when we compared pre- and post-
BMP implementation. The unrestricted cattle access at these sites may have
contributed to the elevated concentrations of E. coli and increased sediment loads
compared to the other sampling sites without cattle access. The potential for further
decreases in water quality at these upper-watershed sites are of concern because
they could impact water quality in lower portions of the SRW. Parkyn et al. (2003)
concluded that achieving water-quality goals for a watershed may not be possible
when the upstream areas and tributaries remain impacted.
Although some impacts observed during our study seem to be directly related
to anthropogenic sources (e.g., cattle access to waterways), others may have been
related to climatic events and are, therefore, not directly related to BMP implementation.
Various severe climatic events after the pre-BMP data-collection phase
may have impacted the water-quality results by causing natural damage to the
stream banks, potentially increasing sediment input into the rivers. For example,
a severe ice storm in the area in late January 2009 (NWS 2010) resulted in trees
on unstable banks falling into streams with their full root wads. This event led
to the increased potential for sedimentation in waterways because banks were
less stable due to loss of soil-holding vegetation, and the sediments from root
wads were subjected to stream flow. Various studies have noted that increased
stream-bank erosion, and hence, increases in TSS and turbidity, are possible when
streamside trees fall (Harmon et al. 1986, Keller and Swanson 1979, Trimble
1997). In some areas of the stream segments sampled in this study, trees were
later removed to prevent log-jams (D. Hall, Fulton County Conservation District,
AR, pers. comm.). This tree-removal potentially exposed even more bank surface,
and released sediment built up behind this coarse woody debris. In 2011,
the wettest month on record occurred in April causing widespread flooding in the
area (NOAA 2012) and bank collapse at multiple sampling sites (T.R. Brueggen-
Boman, 15 May 2011 pers. observ.).
The nutrient concentrations we measured at our sample sites were below threshold
levels for impairment outlined in AR water-quality regulations (APCEC 2014).
It should be noted, however, that multiple sites had significant increases in nutrients
over the course of the study and that these values exceeded historic and referencestream
values (ADEQ 2003, ADPCE 1987, USEPA 2000). We would expect that
nutrient levels in the upper SRW would be low because there is low urban-land usage,
minimal crop-land cover, and, with the exception of 1 confined animal-feeding
operation (CAFO) near SRUP, no major point-source dischargers (ADEQ 2003,
Southeastern Naturalist
709
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
AWIS 2006). BMPs, such as cattle exclusion, were designed to affect nutrient
concentrations in waterways, but thus far the results are mixed. For example, Sheffield
et al. (1997) determined that there was an increase in dissolved nutrients with
reduced cattle presence in a stream, but they compared measurements made during
winter pre-BMP implementation to those made during summer post-BMP implementation.
Other studies have determined that there were no significant differences
in NO3
- and PO4
3- concentrations when comparing reaches with and without cattlegrazing
(Campbell and Allen-Diaz 1997, Ranganath et al. 2009).
Chl-a concentration is used to estimate phytoplankton concentration and is an
indirect indicator of nitrogen and phosphorous concentrations (YSI Environmental
2005). The majority of the levels we detected in this study fell below historically
published values and support the conclusion that NO3
- and PO4
3- concentrations
were not at levels of concern at most sampling locations. Both nutrient and chl-a
concentrations indicated that agricultural activities in the upper portion of the SRW
were not causing excessive nitrogen and phosphorous concentrations in the waterways
at most sampling locations.
The BMPs implemented within this study area were not specifically targeted
to properties adjacent to waterways and were implemented on a strictly volunteer
basis. Thus, it is difficult to determine if a single type and/or locat ion of BMP was
primarily responsible for the results we obtained. However, multiple findings in
this study indicate that future site-specific or goal-specific BMPs to target variables
that impact water-quality should be implemented to maintain and/or improve water
quality in this portion of the SRW. Modeling of agricultural watersheds indicates
that sensitive areas characterized by erosive soils and steep slopes contribute excess
sediments and nutrients to adjacent streams (Mostaghimi et al. 1997, Wright
2015). Wright (2015) reported that implementation of conservation practices on as
little as 5% of sensitive areas in a watershed can significantly decrease non-pointsource
contaminant inputs. Identifying and targeting for management such areas in
the SRW may be key to the success of future water-quality initiatives.
It is possible that the threshold of BMP implementation necessary to limit
increases in sediment and bacterial impairment of these sub-watersheds was not
met, or that our study was not long enough to truly assess the impact of the implemented
BMPs because of lag time from implementation to measurable effect.
Spooner (1991) suggested that a minimum of 2–3 y of pre- and post-BMP samples
should be collected to account for the annual variability in environmental conditions.
Other studies have determined that multiple years of data are necessary to
detect the effects of physical changes, such as stream bank stability, to occur from
BMP implementation because there may be a lag before the resulting changes in
water-quality variables, such as TSS, and turbidity are detectable (Agouridis et al.
2005, Ranganath et al. 2009). Other studies have determined that there are positive
impacts of BMP implementation in less than 4 y (Edwards et al. 1997, Larsen et al. 1994,
Miller et al. 2009). We suggest that the occurrence of severe climatic events could
have affected the variability in our results beyond what can be accounted for by
BMP implementation alone.
Southeastern Naturalist
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
710
Our study design was unique among published studies on BMP effects because
the BMPs were not implemented only at sites of concern or near streams, and
our sampling sites were not directly adjacent to BMP-implementation areas. Our
approach yielded results that are likely more reflective of the reality of current
BMP implementation in which scattered and limited sections of a watershed are
improved. This study method potentially allows for a better understanding of what
may occur in stream segments not directly adjacent to implemented BMPs and
which are a closer representation of streams in the majority of the watershed.
Overall, we conclude that BMPs as implemented at our sites did not result in
improved water quality in the time-frame assessed. While the BMPs implemented
in this study may not have been effective at decreasing concentrations of various
measured variables, they may have been a positive preemptive measure that prevented
increases in water-quality variables to levels of concern over a 4-y time
period. Meals et al. (2010) noted that it is common to find little or no improvement
in water quality in most watershed nonpoint-source projects that have taken
place over the past 4 decades. They suggest that variability in weather, low landowner-
participation, inappropriate BMPs selected for implementation, inadequate
number of BMPs implemented or poor distribution of BMPs, and lag time from
implementation of BMPs may impede researchers’ ability to detect measurable effects
of BMPs.
Our study suggests that additional BMPs specifically designed to limit the concentration
of the parameters of concern (e.g., E. coli) and implementation of those
BMPs in targeted locations may be required to maintain the water quality desired in
this watershed. We recommend that additional monitoring in the upper SRW would
likely allow land managers to gain a better understanding of the long-term impact
of the implemented BMPs.
We also recommend that future studies attempting to determine the effectiveness
of BMPs perform a minimum-detectable–change analysis to account for environmental
variability in the specific area being studied and to determine the sampling
frequency necessary to obtain statistically sound results for the specific parameters
being studied (Spooner et al. 2011). Nonpoint-source models should be used to identify
critical areas to maximize BMP effectiveness and reduce costs by focusing on
areas indicated as large contributors of pollutants to the waterways (Mostaghimi et
al. 1997). Lag time (which can vary from <1 to >50 years depending on the size of the
study area), BMPs implemented, and the parameter of concern should be taken into
serious consideration when planning additional studies (Meals et al. 2010). Meals et
al. (2010) noted that lag time is an unfortunate fact of watershed management, but
must be recognized so that vital restoration efforts do not become discouraged due
to the inability to document improvements in water quality. Following these recommendations
would save time and money, and assist land managers in determining
the most effective BMPs to be implemented to achieve the established water-quality
improvement goals. Our results and those of others like it should help guide watershed-
monitoring programs that aim to implement watershed-conservation actions
that will result in measurable results within a short time period.
Southeastern Naturalist
711
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
Acknowledgments
We thank the US Environmental Protection Agency for a 319 grant and the Arkansas
Water Resource Center for a 104-B US Geological Service grant. Our gratitude also goes
to all the Arkansas State University Ecotoxicology Research Facility staff and student
employees, Arkansas State University students, and Fulton County, AR, Conservation District
employees who helped with data collection. We also thank Katie Kilmer who assisted
in the editing process of this manuscript.
Literature Cited
Agouridis, C.T, S.R. Workman, R.C. Warner, and G.D. Jennings. 2005. Livestock-grazing
management impacts on stream-water quality: A review. Journal of American Water
Resources Association 41:591–606.
Altman, D.G., S.M. Gore, M.J. Gardner, and S.J. Pocock. 1983. Statistical guidelines for
contributors to medical journals. British Medical Journal 286:1 489–1493.
American Public Health Association (APHA). 2005. Standard Methods for the Examination
of Water, and Wastewater. 21st Edition. American Public Health Association, Washington,
DC. 1325 pp.
Arkansas Department of Environmental Quality (ADEQ). 2003. Physical, chemical, and
biological assessment of the Strawberry River watershed. QA-03-12-01. Arkansas Department
of Environmental Quality, Little Rock, AR. 282 pp.
ADEQ. 2008. Integrated water-quality monitoring and assessment report. Prepared pursuant
to Section 305(b) of the Federal Water Pollution Act. Little Rock, AR. 350 pp.
Arkansas Department of Pollution Control and Ecology (ADPCE). 1987. Physical, chemical,
and biological characteristics for least-disturbed reference streams in the Arkansas’
ecoregions Volume II: Data analysis. Little Rock, AR. 148 pp.
Arkansas Pollution Control and Ecology Commission (APCEC). 2014. Regulation no. 2.
Regulation establishing water-quality standards for surface water of the state of Arkansas.
Little Rock, AR. 125 pp.
Arkansas Watershed Information System (AWIS). 2006. A module of the Arkansas automated
reporting and mapping system. Available online at http://watersheds.cast.uark.
edu/ viewhuc.php? hucid=1101001202. Accessed 23 January 2012.
Arkansaswater.org. 2010. Strawberry River watershed-11010012. Available online at http://
arkansaswater .org/index.php?option=com_content&task=view&id=34. Accessed 12
Oct 2010.
Braccia, A., and J.R. Voshell Jr. 2006. Environmental factors accounting for benthic macroinvertebrate-
assemblage structure at the sample scale in stream subjected to a gradient
of cattle grazing. Hydrobiologia 573:55–73.
Brueggen-Boman, T.R., and J.L. Bouldin. 2012. Characterization of temporal and spatial
variation in subwatersheds of the Strawberry River, AR, prior to implementation of agricultural
best-management practices. Arkansas Academy of Science 66:41–49.
Campbell, C.G., and B. Allen-Diaz.1997. Livestock grazing and riparian habitat water
quality: An Examination of oak-woodland springs in the Sierra Foothills of California.
Pp. 339–346, In N.H. Pillsbury, J. Verner, and W.D. Tietje (Technical Coordinators).
Proceedings of a symposium on oak woodlands: ecology, management, and urban interface
issues; 19–22 March 1996, San Luis Obispo, CA 1997. USDA Forest Service
PSW-GTR-160. USDA Pacific Southwest Forest and Range Research Station, Albany,
CA. 738 pp.
Southeastern Naturalist
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
712
Davies-Colley, R.J., J.W. Nagels, R.A. Smith, R.G. Young, and C.J. Phillips. 2004. Water
quality impact of a dairy-cow herd crossing a stream. New Zealand Journal of Marine
and Freshwater Research 38:569–576.
Earth Point. 2012. Township and range-search by description. Available online at http://
www.earthpoint.us/TownshipsSearchByDescription.aspx. Accessed 20 April 2012.
Edwards, D.R., T.C. Daniel, H.D. Scott, P.A. Moore, J.F. Murdoch, and P.F. Vendrall. 1997.
Effect of BMP implementation on storm-flow quality of two northwestern Arkansas
streams. Transactions of American Society of Agricultural Engineers 40:1311–1319.
Fulton County Conservation District (FCCD). 2011. Final report Strawberry River subwatershed
project FY 09-500. Fulton County Conservation District, Salem, AR.14 pp.
Godwin, D.C., and J.R. Miner. 1996. The potential of off-stream livestock watering to reduce
water-quality impacts. Bioresource Technology 58:285–290.
Harmon, M.E., J.F. Franklin, F.J. Swanson, P. Sollins, S.V. Gregory, J.D. Lattin, N.H.
Anderson, S.P. Cline, N.G. Aumen, J.R. Sedell, G.W. Lienkaemper, K. Cromack Jr.,
and K.W. Cummins. 1986. Ecology of coarse woody debris in temperate ecosystems.
Advances in Ecological Research 15:133–302.
Harp, G.L., and H.W. Robison. 2006. Aquatic macroinvertebrates of the Strawberry
River system in north-central Arkansas. Journal of the Arkansas Academy of Science
60:46–91.
Kauffman, J.B., and W.C. Krueger. 1984. Livestock impacts on riparian ecosystems
and streamside management implications: A review. Journal of Range Management
37:430–438.
Kayhanian, M., A. Singh, and S. Meyer. 2002. Impact of non-detects in water-quality data
on estimation of constituent-mass loading. Water Science and Technology 45:219–225.
Keller, E.A., and F.J. Swanson. 1979. Effects of large organic material on channel form and
fluvial processes. Earth Surface Processes 4:361–380.
Larsen, R.E., J.R. Miner, J.C. Buckhouse, and J.A. Moore. 1994. Water-quality benefits
of having manure deposited away from streams. Bioresource Technology 48:113–118.
Line, D.E., W.A. Harman, G.D. Jennings, E.J. Thompson, and D.L. Osmond. 2000. Nonpoint-
source pollutant-load reductions associated with livestock exclusion. Journal of
Environmental Quality 29:1882–1890.
Meals, D.W., S.A. Dressing, and T.E. Davenport. 2010. Lag time in water-quality response
to best-management practices: A review. Journal of Environmental Quality 39:85–96.
Miller, J., D. Chanasyk, T. Curtis, T. Entz, and W. Willms. 2009. Influence of streambank
fencing with a cattle crossing on riparian health and water quality of the Lower Little
Bow River in Southern Alberta, Canada. Agricultural Water Management 97:247–258.
Miner, J.R., J.C. Buckhouse, and J.A. Moore. 1992. Will a water trough reduce the amount
of time hay-fed livestock spend in the stream (and therefore improve water quality)?
Rangelands 14:35–38.
Mostaghimi, S., S.W. Park, R.A. Cooke, and S.Y. Wang. 1997. Assessment of management
alternatives on a small agricultural watershed. Water Resources 31:1867–1878.
National Climatic Data Center- National Oceanic and Atmospheric Administration (NCDCNOAA).
2012. Climate data online. Available online at http://www.ncdc.noaa.gov/ cdoweb/
review. Accessed 25 August 2012.
National Oceanic and Atmospheric Administration (NOAA). 2012. Storm data and unusual
weather phenomena–April 2011. Available online at http://www.srh.noaa.gov/images/
lzk/pdf/lzkapr11.pdf. Accessed 25 August 2012.
National Weather Service (NWS). 2010. Ice storm on January 26–28, 2009. Available online
at http://www.srh. noaa.gov/lzk/?n=win0109a1yr.htm. Accessed 25 August 2012.
Southeastern Naturalist
713
T.R. Brueggen-Boman, S. Choi, and J.L. Bouldin
2015 Vol. 14, No. 4
Parkyn, S.M., R.J. Davies-Colley, N.J. Halliday, K.J. Costley, and G.F. Croker. 2003.
Planted riparian-buffer zones in New Zealand: Do they live up to expectations? Restoration
Ecology 11:436–447.
Platts, W.S., and J.N. Rinne. 1985. Riparian and stream-enhancement research in the Rocky
Mountains. North American Journal of Fisheries Management 5:115–125.
Ranganath, S.C, W.C. Hession, and T.M. Wynn. 2009. Livestock-exclusion influences on
riparian vegetation, channel morphology, and benthic-macroinvertebrate assemblages.
Journal of Soil and Water Conservation 64:33–42.
Robison, H.W., and T.M. Buchanan. 1992. Fishes of Arkansas. The University of Arkansas
Press, Fayetteville, AR. 536 pp.
Sheffield, R.E., S. Mostaghimi, D.H. Vaughan, E.R. Collins Jr., and V.G. Allen. 1997. Offstream
water sources for grazing cattle as a stream-bank stabilization and water-quality
BMP. Transactions of American Society of Agricultural Engineers 40:595–604.
Spooner, J. 1991. Associating changes in water-quality monitoring data with nonpointsource
pollution-control programs. Technical note in North Carolina Quality Group
Newsletter 50. Available online at http://www.bae.ncsu.edu/programs/extension/wqg /
issues/50.html. Accessed 5 September 2012.
Spooner, J., S.A. Dressing, and D.W. Meals. 2011. Minimum detectable-change analysis.
Available online at www.bae.ncsu.edu/programs/extension/wqg/319monitoring /tech_
notes.htm. Accessed 30 July 2015.
Stednick, J. 1991. Wildland Water Quality Sampling and Analysis. Academic Press, San
Diego, CA. 217 pp.
Stephenson, G.R., and R.C. Rychert. 1982. Bottom sediment: A reservoir of Escherichia
coli in rangeland streams. Journal of Range Management 35:119–123.
Trimble, S.W. 1997. Stream-channel erosion and change resulting from riparian forests.
Geology 25:467–469.
US Environmental Protection Agency (USEPA). 2000. Ambient water-quality criteria recommendations.
Office of Research and Development, Washington, DC.
USEPA. 2006. TMDLs for turbidity in the Strawberry River Basin, AR. Prepared by FTN
Associates, Ltd., Little Rock, AR. 29 pp.
USEPA. 2009. National water-quality inventory: Report to Congress, 2004 reporting cycle.
Available online at http://water.epa.gov/lawsregs/guidance/cwa/305b/upload/2009_05_
20_305b_2004report_report2004pt3.pdf. Accessed 23 July 2012.
Wohl, N.E., and R.F. Carline. 1996. Relations among riparian grazing, sediment loads,
macroinvertebrates, and fishes in three central Pennsylvania streams. Canadian Journal
of Fisheries and Aquatic Sciences 53:260–266.
Wright, R.E. 2015. Economic and environmental modeling of an agricultural watershed in
the Mississippi Delta: Crop selection using an LP model and water-quality impacts of
bioenergy crops using the AnnAGNPS model. Ph.D. Dissertation. Arkansas State University,
Jonesboro, AR. 345 pp.
YSI Environmental. 2005. The basics of chlorophyll measurement. YSI Incorporated, Yellow
Springs, OH. Available online at http://www.ysi.com/media/pdfs/T606-The-Basicsof
Chlorophyll-Measurement.pdf. Accessed 15 October 2012.