See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/318417742

Phosphorus, Iron, and Aluminum Losses in Runoff from a Rotationally Grazed

Pasture in Georgia

Article · January 2017

DOI: 10.13031/trans.12053

CITATION READS
1 124

6 authors, including:

Dinku M. Endale Harry Schomberg

United States Department of Agriculture United States Department of Agriculture
97 PUBLICATIONS   1,626 CITATIONS    122 PUBLICATIONS   3,969 CITATIONS   

SEE PROFILE SEE PROFILE

Dwight Starnes Fisher Michael Jenkins

AZ Statistics, Tucson, Arizona United States Department of Agriculture
132 PUBLICATIONS   2,101 CITATIONS    104 PUBLICATIONS   2,480 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Impact of land-management of soil microbial communities View project

Conservation Practice Effects in the SE Coastal Plain, USA View project

All content following this page was uploaded by Dwight Starnes Fisher on 07 May 2018.

The user has requested enhancement of the downloaded file.

 PHOSPHORUS, IRON, AND ALUMINUM LOSSES IN RUNOFF
FROM A ROTATIONALLY GRAZED PASTURE IN GEORGIA
D. M. Endale, H. H. Schomberg, D. S. Fisher, L. B. Owens, M. B. Jenkins, J. V. Bonta

ABSTRACT. Well-managed grazing systems can provide valuable ecosystem services, such as reducing sediment and phospho-
rus (P) loading to nearby waterways. However, the available long-term data to fully support this hypothesis are limited. In this
article, we describe flow-weighted concentrations (FWCs) and loads for dissolved reactive P (DRP), total P (TP), iron (Fe),
and aluminum (Al) over 11 years (1999-2009) from a 7.8 ha rotationally grazed pasture (W1) near Watkinsville, Georgia. The
region is characterized by Fe and Al rich and acidic Ultisols. Cattle numbering 21 to 224 (mean 91) grazed W1 on 69 occasions
for 1 to 71 d (mean 19.2). Of 74 runoff events, 20 occurred when the monthly rainfall was below the long-term average (deficit
period) and 54 occurred during non-deficit periods. Samples were collected from 43 of 74 runoff events for nutrient analyses.
Event FWC (mg L-1) ranged from 0.38 to 7.07 for DRP (mean 1.91), from 0.36 to 7.60 for TP (mean 2.43), from 0.03 to 0.55
for Fe (mean 0.23), and from 0.43 to 553 g L-1 for Al (mean 65 g L-1). Event load (kg ha-1) ranged from 0.00 to 0.45 for
DRP (mean 0.10), from 0.00 to 0.55 for TP (mean 0.12), from 0.00 to 0.11 for Fe (mean 0.02), and from 0.00 to 0.10 for Al
(mean 0.01). The total load (kg ha-1) was 4.12 for DRP, 5.12 for TP, 0.71 for Fe, and 0.25 for Al. DRP accounted for 80%
of the TP FWC and load. Cattle presence increased sediment load, but the difference was not statistically significant. There
was high correlation between Fe and DRP loads (r = 0.87), a likely indicator of erosion-induced losses due to cattle tread-
ing. Cattle presence increased FWCs but not loads for DRP and TP. The FWCs for DRP and TP were not different between
deficit and non-deficit periods, but mean loads were 3-fold to 4-fold greater during non-deficit periods. Means from the six
largest P loss events were 3-fold greater for FWC and 7-fold greater for load than the remaining 37 events. These six large
events accounted for 53% of the total P load. Less than 1% of the inorganic P applied and redeposited through manure was
lost in runoff. The study demonstrated that hydrologic transport processes were the dominant drivers of pollutant fluxes and
highlighted the possible mitigation of pollutant fluxes through grazing management that includes maintenance of good grass
cover, effective rotational grazing, and limited fertilization.
Keywords. Calving, Cattle, Dissolved reactive phosphorus, Drought, Eutrophication, Manure, Runoff, Total phosphorus,
Water quality.

I
n the eastern half of the U.S., cattle production (grazing
and haying) is generally limited to areas of less produc-
tive land (Philipp et al., 2015). Eighty percent of these
predominantly cow-calf operations have less than
The authors have paid for open access for this article. This
work is licensed under a Creative Commons Attribution-
NonCommercial-NoDerivatives 4.0 International License https://creative
commons.org/licenses/by-nc-nd/4.0/.
Submitted for review in August 2016 as manuscript number NRES
12053; approved for publication by the Natural Resources & Environmental
Systems Community of ASABE in February 2017.
This article has been reviewed and approved for publication per the USDA
internal publication approval process. However, it may not reflect official policy
of the USDA nor of the Agricultural Research Service. Mention of company
or trade names is for description only and does not imply endorsement by the
USDA. The USDA is an equal opportunity provider and employer.
The authors are Dinku M. Endale, ASABE Member, Agricultural
Engineer, USDA-ARS Southeast Watershed Research Laboratory, Tifton,
Georgia; Harry H. Schomberg, Ecologist, USDA-ARS Sustainable
Agricultural Systems Laboratory, Beltsville, Maryland; Dwight S. Fisher,
Research Leader (Retired) USDA-ARS J. Phil Campbell Sr. Natural
Resources Conservation Center, Watkinsville, Georgia; Lloyd B. Owens,
Soil Scientist (Retired), USDA-ARS Appalachian Experimental
Watershed, Coshocton, Ohio; Michael B. Jenkins, Microbiologist, USDA-
ARS National Sedimentation Laboratory, Oxford, Mississippi; James V.
Bonta, ASABE Member, Hydraulic Engineer (Retired), USDA-ARS
National Sedimentation Laboratory, Oxford, Mississippi. Corresponding
author: Dinku M. Endale, 2316 Rainwater Road, P.O. Box 748, Tifton, GA
31793; phone: 229-386-3893; e-mail: Dinku.Endale@ars.usda.gov.
50 cows per farm (Short, 2001, Franzluebbers, 2007; Philipp
et al., 2015). When aggregated, the extent of grasslands that
are used as pastures and hayfields is large: 22,075 km2 in the
Southern Coastal Plain (8%) and 18,355 km2 of the Southern
Piedmont (11%) (USDA-NRCS, 2006). Foraging cattle re-
distribute phosphorus (P) across the pasture landscapes at
spatial densities dependent on where they graze, water, con-
gregate, and camp (Schomberg et al., 2000; Bellows, 2001).
Heavy-use areas can act as source-areas of P and other nu-
trients because they receive greater manure and urine inputs.
They also experience increased treading and loss of vegeta-
tion, which increases soil physical degradation and contrib-
utes to increased risk of runoff and soil erosion (Russell et
al., 2001; McDowell et al., 2003; Haan et al., 2006; Cour-
nane et al., 2010). The highly weathered, iron (Fe) and alu-
minum (Al) rich, and acidic Ultisols of the southeastern U.S.
can hold P tightly in chemical bonds with Fe and Al (Zaimes
and Schultz, 2002). Runoff losses of Al and Fe could be in-
dicative of the portion of P coming from eroding soils versus
cattle or fertilizer.
Release of P from agricultural lands into surface and sub-
surface waters contributes to eutrophication and impairment
in streams, rivers, lakes, or coastal waters (Ryden et al.,
1973; Sharpley et al., 1994; USEPA, 1996; Correll, 1998;
Carpenter et al., 1998; Parry, 1998; Sharpley et al., 2003;
McDowell et al., 2007; Carpenter, 2008; Jarvie et al., 2012;

Transactions of the ASABE

Vol. 60(3): 861-875 2017 American Society of Agricultural and Biological Engineers ISSN 2151-0032 https://doi.org/10.13031/trans.12053 861
Sharpley et al., 2013). Georgia alone contains 14 river ba-
sins, 52 major watersheds, 70,500 km of perennial streams,
and 38,250 km of intermittent streams (Georgia Adopt-A-
Stream, 2008). These are susceptible to nonpoint-source pol-
lution, including nutrients and sediment, due to high rates of
annual rainfall (1000 to 1500 mm) and surface runoff in the
region. Phosphorus is a primary concern due to its potential
for transport from farm fields and pastures in dissolved form
or attached to sediment and can deposit in streams and lake
bottoms. This P can be resuspended or released (referred to as
“legacy phosphorus”) and is one of the reasons that conserva-
tion and best management practices are not producing de-
sired water quality in some watersheds in the U.S. (Sharpley
et al., 2013; Jarvie et al., 2013). Good pasture management
aims to maximize nutrient cycling efficiency and minimize
nutrient and sediment losses with runoff (Bellows, 2001).
Franzluebbers et al. (2012) articulated the potential of well-
managed pasture systems to provide valuable ecosystem ser-
vices and discussed barriers to adoption of such systems.
The objective of our study was to present concentration
and load data for P, Fe, and Al gathered over 11 years (1999-
2009) that had contrasting weather patterns from a 7.8 ha
pasture used for rotational grazing of cow-calf herds near
Watkinsville, Georgia, and demonstrate the possible mitiga-
tion of pollutant fluxes through proper grazing management
that includes maintenance of good grass cover, avoiding
over-grazing, and limiting fertilizer application.
MATERIALS AND METHODS
EXPERIMENTAL SITE
The study was conducted from January 1999 to Decem-
ber 2009 on a zero-order 7.8 ha watershed (W1; average
slope 4.2%) nested within the larger North Unit watershed at
Figure 1. Location of pasture watershed W1 within the North Unit watershed of the USDA-ARS J. Phil Campbell Sr. Natural Resource Conser-
vation Center near Watkinsville, Georgia.
862
the USDA-ARS J. Phil Campbell Sr. Natural Resource Con-
servation Center near Watkinsville, Georgia (33° 54 N and
83° 24 W) (fig. 1; Endale et al., 2011). The soils are classi-
fied as Cecil (69%) and Pacolet (31%) series (fine, kao-
linitic, thermic Typic Kanhapludults) and generally have
brownish-gray sandy loam to red clay loam surface horizons
overlaying red clayey argillic horizons. The average daily air
temperature ranges from 6°C to 8°C in winter and from 23°C
to 27°C in summer. The mean annual rainfall is 1250 mm
with a mean monthly rainfall variation of 78 to 136 mm. Pe-
riods of drought are common.
The W1 watershed has been used as a rotational-grazing
pasture for Black Angus (Bos taurus) cattle since 1960. The
forage consisted of warm-season grasses, primarily bermu-
dagrass (Cynodon dactylon) with other naturalized grass spe-
cies, for summer grazing and fall over-seeded cereal rye (Se-
cale cerale L.) or wheat (Triticum aestivum) for winter graz-
ing. Supplemental grain and hay were sometimes fed during
the winter. Forage was grazed when it reached approximately
3,500 kg ha-1, and animals were removed when forage was
grazed to approximately 2,000 kg ha-1. Fertilization consisted
of biannual (early spring and late summer) application of ni-
trogen, phosphorus, and potassium (N-P-K) as 34-0-0, 17-17-
17, 21-0-21, or urea with sulfur (table A1 in the Appendix).
There was no inorganic P fertilization after 2003. Soil pH was
maintained between 6.0 and 6.5 by liming as needed. Weed
management consisted use of Grazon P+D (Picloram + 2,4-
D), a broadleaf herbicide, as needed. In addition, mowing was
carried out as needed for weed control and to remove forage
seed heads in late spring and early fall. Grazing was by cows
or cow-calf pairs either alone or with one or two breeding
bulls. Calving season extended from January through early
April except in 2009 when there was no calving.
Soil sampling and analysis conducted at W1 in December
TRANSACTIONS OF THE ASABE
2009 from 14 locations across three parallel longitudinal
(E-W) transects provided the following means (with range)
for soil nutrients for the 0 to 15 cm soil depth: N = 0.24%
(0.16% to 0.34%), C = 2.55% (1.72% to 3.57%), extractable
P (Mehlich I) = 94 kg ha-1 (39 to 186), total P (acid digestion)
= 1,321 kg ha-1 (991 to 1,950), extractable K = 298 kg ha-1
(171 to 186), total K = 2,221 kg ha-1 (1,190 to 4,554), total
Fe = 47,134 kg ha-1 (17,069 to 92,893), and total Al =
41,408 kg ha-1 (14,670 to 78,534). The mean soil pH was 6.0.
RAINFALL AND RUNOFF MEASUREMENT,
SAMPLING, AND PROCESSING
An automated system located at the outlet of W1 was
used to measure runoff through a 1.14 m (3.75 ft) high, 2:1
ratio, concrete broad-crested V-notch weir at 5 min intervals
(Endale et al., 2011). The automated system consisted of a
tipping-bucket rain gauge, a depth-sensing pressure trans-
ducer, and a data logger programmed to convert flow depth
to discharge using the weir calibration curve. A Sigma 900
Max discrete water sampler was used to collect runoff sam-
ples in twenty-four 300 mL glass bottles. Until February
2006, samples were obtained at 8 min intervals through a
runoff event with two consecutive 150 mL samples put in
the same bottle, allowing sampling over 6.5 h. Beginning in
February 2006, the sampling interval was increased to
10 min with three consecutive 100 mL samples put in the
same bottle, allowing sampling over 12 h. Samples for nu-
trient analyses were collected during 43 of 74 recorded run-
off events. Occasional sampling problems precluded collect-
ing samples from all runoff events. Samples were removed
from the sampler within 36 h of each runoff event and
brought to the laboratory for processing.
In the laboratory, samples were composited into approxi-
mate hourly subsamples. From each hourly subsample,
100 mL was filtered through 0.45 m cellulose-nitrate mem-
brane and refrigerated at 4°C for colorimetric analysis of dis-
solved reactive P (DRP) (Murphy and Riley, 1962) within
two to three weeks. Two 100 mL unfiltered subsamples were
frozen. One was kept as spare, and the second was analyzed
for total P, Al, and Fe at the USDA-ARS North Appalachian
Experimental Watershed in Coshocton, Ohio. Total P con-
centration was determined by ion chromatography (EPA
Method 300.0; Pfaff, 1993). Iron and Al concentrations were
determined by atomic absorption spectrophotometry
(USEPA, 1991). When the Al concentration was below the
detection limit of 0.003 mg L-1, this value was taken as the
concentration for the particular sample.
Total sediment solids in runoff was estimated from a total
of 22 events with distribution of one to six events each year,
except 2000, 2005, and 2007 (no samples processed), by
evaporating 100 to 200 mL subsamples at 100°C. Due to vis-
ual observation of limited sediment, not all runoff events
were assessed for sediment content during periods of abun-
dant grass cover.
DATA PROCESSING AND ANALYSIS
Nutrient concentration for a runoff event was considered
as event flow-weighted concentration (FWC) by dividing the
total load by the total runoff volume for each event. The total
event load was estimated by taking the sum of the loads from
60(3): 861-875
each subsample of each event. Subsample loads were calcu-
lated as the product of runoff volume and concentration of
each subsample. Flow-weighted concentrations and loads
for Al, Fe, DRP, and TP from the 43 sampled events were
used as dependent variables to examine management and en-
vironmental influences. To examine the influence of cattle,
variables were grouped into periods when cattle were on or
off W1. To examine the influence of drought, variables were
grouped into categories based on whether monthly rainfall
was less than the long-term average (deficit period) or equal
to or above the long-term average (non-deficit period) (En-
dale et al., 2011). The influence of inorganic fertilization was
considered within factor analysis (see subsequent discus-
sion). The sediment loss data were used to examine possible
links between cattle and soil erosion manifesting as elevated
sediment loss, and correlations among sediment and FWCs
and loads.
For each runoff event, hydrologic parameters were com-
piled as the 1-day (DR1) and 2-day (DR2) rainfall (mm), to-
tal runoff (TR, mm), percentage runoff from 1-day rainfall
(RPR, %), peak runoff discharge rate (PD, L s-1), and 5, 10,
and 15 day rainfall (DR5, DR10, and DR15, mm) used as
pre-runoff antecedent rainfall, as well as days since the last
runoff event (DSLR). Days with no cattle prior to runoff
(DNOCPR) and days since the last fertilization (DSLF) were
considered management variables. Cattle variables were
considered in three groups, all as cattle days (the product of
the number of cattle including cows, calves, and bulls, and
the number of days spent grazing). The first group was cattle
days since the last runoff event (CDSLR). The second group
was continuous cattle days during and immediately prior to
a runoff event (CCDPR) which can last from 1 d to more
than 30 d. The third group was cattle days during the 30 days
immediately prior to a runoff event (CD30DPR). While the
choice of 30 is arbitrary, it is close to the 75th percentile for
the days cattle were on W1 per grazing period. In addition,
animal unit (AU) and animal unit day (AUD) equivalents
were estimated for each grazing period as follows (from
chapter 6 of USDA-NRCS, 2003): 1.0 AU for a mature cow
weighing approximately 454 kg with a calf up to six months
of age, 1.35 AU for a mature bull, 0.6 AU for cattle one year
old, and 0.8 AU for cattle two year old. We assigned 0.7 AU
for cattle that were neither cow-calf pairs nor bulls because
of limited information on cattle age. Estimates of AU and
AUD were not used in statistical analyses but are only given
as summaries where appropriate.
Data analyses were conducted using SAS (ver. 9.4)
within the graphical user interface SAS Enterprise Guide of
SAS for Windows (ver. 7.1, SAS Institute, Inc., Cary N.C.).
Normality of data was checked with the Shapiro-Wilk W sta-
tistic and graphical representation of the data. Non-normal
data were transformed (see below). Parametric tests were
conducted where data normality was established (original or
transformed data), and nonparametric tests were used when
transformation failed to establish normality of data. Signifi-
cant levels were set at   0.05. Factor analysis was con-
ducted using the 13 independent variables (transformations
in parentheses) TR (log10), PD (log10), DR1 (log10), DR2
(log10), DR5 (square root), DR10 (square root), DR15
(square root), DSLR (log10), CDSLR (original), CCDPR
863
(original), CD30DPR (original), DNOCPR (original), and 2013 (Microsoft Corp., Redmond, Wash.) was used to ex-
DSLF (square root) to identify appropriate variables to use amine linear relationships and build linear models between
for developing regression models relating management and variables. SigmaPlot 12 (Systat Software, Inc., San Jose,
runoff to nutrient concentration and load. Factor analysis in- Cal.) was used for graphing.
volved three steps: identifying the number of statistical fac-
tors within the independent variables, extracting loadings for
each factor that link the factors to the independent variables, RESULTS
and rotation of loadings to reduce correlation among the fac- CATTLE, FERTILIZATION, AND HYDROLOGY
tors and clarify the links to the independent variables. Inde- W1 was grazed on 69 occasions using 21 to 224 cattle (11
pendent variables with large loadings within a factor were to 120 AU) from 1 to 71 d per occasion (table 1). Endale et
usually associated with that underlying process. This helped al. (2011) provide additional grazing information, including
us identify the variables that were the best representatives of monthly and annual distributions. Details and summary sta-
each factor and that underlying process in the regressions. tistics for fertilization are given in table A1 (in the Appen-
Pearson’s correlation analysis was performed using trans- dix) and table 1, respectively. Inorganic P and K were ap-
formed data among FWCs and loads of Fe, Al, DRP, and TP. plied from 2000 to 2003, S from 2006 to 2009, and N from
Probability of exceedance was determined for rainfall, run- 2000 to 2009. The interval between fertilization varied from
off, and FWC and load of DRP and TP. Microsoft Excel
Table 1. Summary statistics for cattle and fertilizer management and hydrology at W1 for 1999 to 2009.
Summary Statistics[a] No. of
Variable Mean SE Median Min. Max. Sum[b] Observations
Total head of cattle Cows 56.3 3.1 49.0 11.0 120.0 3881 69
Calves 35.0 4.3 31.0 0.0 112.0 2415 69
Bulls 0.4 0.1 0.0 0.0 2.0 26 69
Sum 91.0 6.7 78.0 21.0 224.0 6278 69
[c]
Animal units 54.4 3.3 46.9 11 120 3545 69
Grazing days[d] Cows 19.2 1.8 15.0 1.0 71.0 1324 69
Calves 11.3 1.8 7.0 0.0 71.0 779 69
Bulls 4.5 1.1 0.0 0.0 46.0 309 69
Sum 19.2 1.8 15.0 1.0 71.0 1326 69
Grazing cattle days (head  days) Cows 936.0 90.7 810.0 31.0 3192.0 64583 69
Calves 463.1 60.0 328.0 0.0 2352.0 31951 69
Bulls 5.3 1.2 0.0 0.0 46.0 366 69
Sum 1404.3 127.4 1134.0 63.0 4704.0 96900 69
Animal unit days[c] 830.5 81.4 616.0 32.4 3192.0 57305 69
Non-grazing days 37.3 3.2 27.0 6.0 114.0 2572 69
Days without cattle pre-runoff 18.9 2.6 7.5 0.0 77.0 NA 74
Cattle days during runoff Cows 413.4 77.5 0.0 0.0 2843.0 30176 73
Calves 135.7 31.9 0.0 0.0 1139.0 9905 73
Bulls 2.0 1.1 0.0 0.0 46.0 146 73
Sum 551.1 101.8 0.0 0.0 3138.0 40227 73
Animal unit days[c] 379.3 70.9 0.0 0.0 2109.0 28068 73
Cattle days for 30 days pre-runoff Cows 494.3 62.7 327.5 0.0 2143.0 36576 74
Calves 207.7 34.8 0.0 0.0 1051.0 15373 74
Bulls 1.3 0.6 0.0 0.0 29.0 99 74
Sum 703.4 87.5 492.0 0.0 2679.0 52048 74
Animal unit days[c] 451.8 58.3 303.5 0.0 1767.0 33429 74
Fertilization
Days in between fertilization 178.5 24.9 150.5 39.0 490.0 NA 22
Days between runoff event and last fertilization 125.3 12.6 108.0 1.0 487.0 NA 73
Rate (kg ha-1) N 75.6 4.1 76.2 56.0 114.2 1664 22
Elemental P 25.0 0.0 25.0 25.0 25.0 200 8
Elemental K 48.6 1.2 47.4 47.4 58.6 438 9
Elemental S 47.1 2.1 42.6 42.6 53.2 330 7
Lime 2100 300 2400 1200 2400 8400 4
Hydrology[e]
1 day rainfall (mm) (runoff day) 45.2 2.9 38.2 8.1 128.8 3346 74
2-day rainfall (mm) (runoff + previous day) 51.3 3.0 44.3 16.3 137.2 3793 74
Total runoff (mm) 5.4 1.0 1.0 0.0 41.1 401 73
Percent runoff (from 1-day rainfall) 8.9 1.4 4.5 0.0 65.5 NA 73
Peak discharge (L s-1) 73.9 14.9 13.9 0.1 565.1 NA 74
5-day antecedent rainfall (5-day prior to runoff) (mm) 26.1 3.7 15.2 0.0 173.0 1931 74
10-day antecedent rainfall (10-day prior to runoff) (mm) 45.8 4.3 35.2 0.0 191.3 3386 74
15-day antecedent rainfall (15-day prior to runoff) (mm) 64.6 4.9 52.1 5.1 226.3 4778 74
Days since last runoff 55.4 10.9 23.5 1.0 651.0 NA 74
[a]
SE = standard error, Min. = minimum, and Max. = maximum.
[b]
NA = not applicable, i.e., values removed because they did not have meaning.
[c]
Computed per USDA-NRCS (2003).
[d]
Number of days between the on and off dates of cattle in W1.
[e]
For runoff and percent runoff, one event excluded. Antecedent refers to pre-runoff dates.

864 TRANSACTIONS OF THE ASABE

39 to 490 d. The interval between a runoff event and the pre-
vious fertilization date varied from 1 to 487 d. Details for
event-based hydrologic parameters are presented in tables
A2 and A3 in the Appendix, and the statistics are summa-
rized in table 1. Of the 74 recorded runoff events, 20 events
occurred during the deficit period, and 54 events occurred
during the non-deficit period. The interval between runoff
events varied from 1 to 651 d (mean 55.4). The 1-day rainfall
(mm, day of runoff) explained 46.4% of the runoff variabil-
ity, while the 2-day rainfall (mm, the day of runoff plus the
previous day) explained 54% of the runoff variability (En-
dale et al., 2011). The common logarithm of the probability
of exceedance (PE) for the 1-day rainfall and runoff (%) each
fit a linear model (see fig. 2 for model parameters). There
was approximately a 25% probability of exceeding a 1-day
rainfall of 54 mm and 14% runoff, a 10% probability of ex-
ceeding a 1-day rainfall of 75 mm and 26% runoff, and a 5%
probability of exceeding a 1-day rainfall of 94 mm rainfall
and 35% runoff.
NUTRIENT AND SEDIMENT CONCENTRATIONS
AND LOADS
Dissolved reactive phosphorus (DRP) accounted for ap-
proximately 80% of the TP concentration and load in runoff
(eqs. 1 and 2, respectively):
FWCDRP = 0.771(FWCTP)
(r2 = 0.968; p < 0.001)
LoadDRP = 0.803(LoadTP)
2
(r = 0.984; p < 0.001)
where FWC is the flow-weighted concentration in mg L-1,
and load is in kg ha-1. The common logarithm of the proba-
bility of exceedance for the FWCs of DRP and TP each fit a
linear model (see fig. 3 for model parameters). Similarly, the
common logarithm of the probability of exceedance for the
loads of DRP and TP each fit a linear model (see fig. 4 for
model parameters).
The overall distribution for the FWCs and loads for DRP
and TP are shown in figure 5a and 5b, respectively. For DRP,
FWC varied from 0.38 to 7.07 mg L-1 (mean 1.91 and me-
dian 1.64), and the load varied from 0.16  10-3 kg ha-1 to
0.45 kg ha-1 (mean 0.10 and median 0.03). For TP, the FWC
varied from 0.36 to 7.59 mg L-1 (mean 2.43 and median
1.97), and the load varied from 0.16  10-3 kg ha-1 to 0.55 kg
ha-1 (mean 0.12 and median 0.03). Table 2 provides sum-
mary statistics for the FWC and load for Fe and Al. For Fe,
the FWC varied from 25 to 550 g L-1, and the load varied
from 0.11 to 110 g ha-1. For Al, the FWC varied from 0.43
to 552 g L-1, and the load varied from 2.0  10-3 g ha-1 to
95 g ha-1.
Pearson’s correlation analysis showed that there was no
correlation between the FWCs of Fe and DRP nor between
the FWCs of Fe and TP. There was significant correlation
between the FWCs of Al and DRP (r = 0.351) and between
the FWCs of Al and TP (r = 0.364). There was also signifi-
cant correlation between the loads of Fe and DRP (r = 0.867)
and between the loads of Fe and TP (r = 0.855). The corre-
(1) lation was less between the loads of Al and DRP (r = 0.585)
and between the loads of Al and TP (r = 0.579). The FWC
and load of Al were highly correlated (r = 0.736) but the
(2) FWC and load of Fe less so (r = 0.445). The loads of Al and
Fe were moderately correlated (r = 0.617) but not the FWCs
of Al and Fe. The strong correlation between loads of Fe and
DRP and the loads of Fe and TP, and the moderately high
correlation between the loads of Al and DRP and the loads

Figure 2. Best-fitted distribution for the com-
mon logarithm (Log) of the probability of ex-
ceedance (PE) for (a) 1-day rainfall (mm), and
(b) equivalent runoff (%) (n = 74). Also indi-
cated are model parameters.
Figure 3. Best-fitted distribution for the com-
mon logarithm (Log) of the probability of ex-
ceedance (PE) of the flow-weighted concentra-
tion (FWC) for (a) dissolved reactive phospho-
rus (DRP) and (b) total phosphorus (TP) (n =
43). Also indicated are model parameters.
Figure 4. Best-fitted distribution for the com-
mon logarithm (Log) of the probability of ex-
ceedance (PE) of the load for (a) dissolved re-
active phosphorus (DRP) and (b) total phos-
phorus (TP) (n = 43). Also indicated are model
parameters.

60(3): 861-875 865

Figure 5. Box plots for the original (non-transformed) data for (a) flow-weighted concentration (FWC) and (b) load for dissolved reactive phos-
phorus (DRP) and total phosphorus (TP) when cattle were in and out of W1, as well as for all the data, and (c) FWC and (d) load for DRP and
TP for deficit (Def) and non-deficit (Non-def) periods. The dashed and solid lines inside the boxes represent means and medians, respectively.
Whiskers represent the 10th and 90th percentiles, and dots represent outliers. Different letters above boxes indicate significant effect of the pres-
ence of cattle or deficit and non-deficit periods (p ≤ 0.05) based on analysis using transformed data.

Table 2. Descriptive statistics for flow-weighted concentration (FWC) and load for iron (Fe) and aluminum (Al) across 43 runoff events, for events
where cattle were on and off W1, for deficit and non-deficit periods, and sediment concentration and load with cattle on or off W1.
Percentiles
Variable Period[a] Mean SE Median Min. Max. 10th 25th 75th 90th
FWC (g L-1) Fe All (n = 43) 230.0 17.7 200.0 25.4 550.4 103.5 166.0 282.5 394.3
Cattle (n = 19) 247.1 26.7 225.8 119.7 550.4 136.6 183.6 275.1 542.0
No cattle (n = 24) 216.3 23.9 196.0 25.4 500.0 92.8 158.5 296.0 392.9
Deficit (n = 12) 227.3 29.0 200.0 119.7 500.0 135.7 172.4 271.9 439.6
Non-deficit (n = 31) 231.0 22.1 201.6 25.4 550.4 107.0 161.6 286.7 397.2
Al All 65.0 18.2 3.0 0.4 552.5 3.0 3.0 76.4 243.3
Cattle 51.8 18.5 3.0 0.4 245.1 3.0 3.0 84.9 240.5
No cattle 75.4 29.4 3.0 0.5 552.5 3.0 3.0 69.5 325.7
Deficit 35.6 14.6 6.7 3.0 171.9 3.0 3.0 61.0 143.3
Non-deficit 76.36 24.45 3.00 0.11 109.82 3.00 3.00 119.08 286.46
Load (g ha-1) Fe All 16.42 3.84 2.62 0.11 109.82 0.20 0.97 21.48 59.81
Cattle 15.93 5.18 2.84 0.14 62.36 0.21 0.71 37.29 58.11
No cattle 16.81 5.61 2.42 0.11 109.82 0.86 1.03 21.23 64.67
Deficit 7.04 5.08 0.99 0.11 62.36 0.20 0.37 5.10 46.11
Non-deficit 20.05 4.83 8.52 0.43 552.50 0.83 1.14 35.58 60.38
Al All 5.75 2.82 0.21 0.00 95.42 0.01 0.02 0.72 14.89
Cattle 3.88 3.48 0.22 0.00 66.47 0.02 0.08 0.46 3.05
No cattle 7.23 4.28 0.11 0.00 95.42 0.01 0.02 0.84 29.32
Deficit 0.23 0.09 0.06 0.00 0.91 0.00 0.01 0.34 0.87
Non-deficit 7.89 3.86 0.32 0.00 95.42 0.02 0.03 0.85 37.50
-1
Concentration (mg L )
Sediment[b] Cattle (n = 7) 237.80 101.65 90.00 20.00 747.00 ND 52.86 90.00 ND
No cattle (n = 12) 121.10 32.81 75.63 30.00 434.50 32.00 50.00 184.00 237.80
Load (kg ha-1)
Sediment Cattle 4.58 1.65 2.91 0.04 12.09 ND 1.81 7.27 ND
No cattle 2.01 0.62 1.52 0.01 7.27 0.11 0.55 2.01 2.02
[a]
n = number of observations.
[b]
From 19 of 22 events when there was FWC and load data for P, Fe, and Al; ND = not determined by Excel.

of Al and TP indicate that P losses were likely associated SEDIMENT LOSS AND IMPLICATIONS
with transport of colloidal clays and sediment, which are Sediment concentration in the 22 assessed events varied
high in Al and Fe in this highly weathered soil. from 18 to 747 mg L-1 with a mean and a median of 157 and

866 TRANSACTIONS OF THE ASABE

86 mg L-1, respectively. Sediment load varied from 0.01 to
65.38 kg ha-1 with a mean and a median of 5.54 and 1.90 kg
ha-1, respectively. Sediment load was generally low, as 54%
of the events had less than 2.0 kg ha-1, 23% of the events had
between 2.0 and 5.0 kg ha-1, and 14% of the events had be-
tween 5.0 and 10.0 kg ha-1. There were FWC and load data
for DRP, TP, Fe, and Al for 19 of the 22 events. Cattle were
present in the pasture in 7 of the 19 events (means of
46.7 cows, 12.7 calves, and 17.7 d; 36.5 AU and 633 AUD).
For these 19 events, there were no significant differences be-
tween observations with cattle compared to observations
without cattle (table 2). However, the differences may have
been too small to detect in this particular data set.
A regression analysis between sediment concentration
(SedimentCONC) and FWC for DRP and TP (all in mg L-1)
produced equations 3 and 4:
FWCDRP = 0.687(SedimentCONC) + 0.006 (3)
(r2 = 0.672; p < 0.02)
FWCTP = 0.771(SedimentCONC) + 0.009 (4)
(r2 = 0.850; p < 0.02)
Similarly, a regression analysis between sediment load
(SedimentLOAD) and DRP and TP load (all in kg ha-1) pro-
duced equations 5 and 6. One pair of data was excluded as
an outlier because it differed by orders of magnitude. In ad-
dition, the intercept was set as zero because of the physical
relationship between loads and because initial regressions
showed neither intercept was significantly different from
zero:
LoadDRP = 0.011(SedimentLOAD) (5)
(r2 = 0.703; p < 0.0001)
LoadTP = 0.013(SedimentLOAD) (6)
(r2 = 0.791; p < 0.0001)
A regression analysis between sediment concentration
and FWC of Fe or Al did not produce a significant model,
and the r2 was very low at 0.1 or less. However, regression
between sediment load and Fe and Al load showed signifi-
cance when two data pairs were removed as outliers (high sed-
iment, very low Fe or Al). Neither intercept was significant. A
regression with zero intercept produced equations 7 and 8:
LoadFe = 0.006(SedimentLOAD) (7)
(r2 = 0. 701; p < 0.0001)
LoadAl = 0.009(SedimentLOAD) (8)
(r2 = 0.634; p < 0.0001)
The grass sod was well established (since 1960), and the
rotational grazing system with limited cattle numbers pro-
tected the pasture from excessive cattle-induced erosion. Ar-
eas of the pasture subjected to more traffic from cattle near
fence lines, close to the weir (totally fenced in), along the
southern and southwestern edge, and at feeding and watering
points exhibited exposed and disturbed soil. While these ar-
eas were only a small percentage of the total pasture area,
the medium to high associations between the variables
60(3): 861-875
discussed above may have been related to the resulting soil
disturbance and erosion in these heavy-use areas.
IMPACT OF CATTLE
The FWC and load data for DRP and TP were log10 trans-
formed prior to analysis for cattle impacts. The overall dis-
tribution for FWCs and loads for DRP and TP when cattle
were on and off W1 are shown in figures 5a and 5b, respec-
tively. Cattle increased the FWCs of DRP and TP. The non-
transformed mean and median FWCs for DRP when cattle
were on W1 (n = 19) were 2.48 and 2.08 mg L-1, respec-
tively. The equivalent values when cattle were off W1 (n =
24) were 1.46 and 1.27 mg L-1, respectively (figs. 5a and 5b).
The mean and median FWCs for TP were 3.14 and 2.72 mg
L-1, respectively, when cattle were on W1 and 1.87 and
1.69 mg L-1, respectively, when they were off.
In contrast, there were no differences in loads for DRP
and TP whether cattle were on or off W1. The mean and me-
dian loads for DRP were 0.11 and 0.03 kg ha-1, respectively,
when cattle were on and 0.09 and 0.02 kg ha-1, respectively,
when cattle were off (fig. 5b). The mean and median loads
for TP were 0.14 and 0.04 kg ha-1, respectively, when cattle
were on and 0.11 and 0.03 kg ha-1, respectively, when they
were off. The total loads for DRP and TP were 2.00 and
2.59 kg ha-1, respectively, when cattle were on W1 and 2.12
and 2.52 kg ha-1, respectively, when cattle were off W1. The
few values shown as dots in figures 5a and 5b (>90th per-
centile) occurred at high rainfall, runoff, and peak discharge
rates and are discussed below.
The presence of cattle had no effects on the FWC of Fe
nor on the loads of Fe and Al based on analysis using the
transformed data (square root for FWC of Al and log10 for
loads of Fe and Al). The presence of cattle had no effect on
the FWC of Al based on the non-parametric Wilcoxon rank
sum test. This test was used for Al FWC because several
transformation methods did not produce normally distrib-
uted data. Table 2 provides summary statistics for the non-
transformed data for the FWCs and loads of Fe and Al.
IMPACT OF DROUGHT
The FWCs for DRP and TP when compared between def-
icit (n = 12) and non-deficit (n = 31) periods were not differ-
ent. The relatively fewer observations during the deficit pe-
riod may have limited the statistical power of this test. For
the non-transformed data, the mean and median FWCs for
DRP were 1.84 and 1.57 mg L-1, respectively, for the non-
deficit period and 2.08 and 2.14 mg L-1, respectively, for the
deficit period (figs. 5c and 5d). The mean and median FWCs
for TP were 2.29 and 1.87 mg L-1, respectively, for the non-
deficit period and 2.77 and 2.55 mg L-1, respectively, for the
deficit period. Drought did not affect the FWCs of Fe or Al.
In contrast, there were differences in the DRP and TP
loads between the two periods (fig. 5d). The mean and me-
dian loads for DRP were 0.12 and 0.06 kg ha-1, respectively,
for the non-deficit period and 0.03 and 0.01 kg ha-1, respec-
tively, for the deficit period. The mean and median loads for
TP were 0.15 and 0.10 kg ha-1, respectively, for the non-def-
icit period and 0.04 and 0.02 kg ha-1, respectively, for the
deficit period. The total loads were 3.70 kg ha-1 for DRP and
4.61 kg ha-1 for TP during the non-deficit period and 0.42
867
and 0.51 kg ha-1, respectively, during the deficit period.
Drought also had an effect on loads of Fe and Al (table 2).
Transport processes during non-deficit periods favored
greater flux delivery. The 1-day mean rainfall was 55 mm
during the non-deficit period but 37 mm during the deficit
period. The mean runoff was 11.6% during the non-deficit
and 7.3% during the deficit period. The mean peak discharge
was 102 L s-1 during the non-deficit and 38 L s-1 during the
deficit period. The few values shown as dots in figures 5c
and 5d (>90th percentile) occurred at high rainfall, runoff,
and peak discharge rates and are discussed below.
HIGH FWC AND LOAD-PRODUCING EVENTS FOR P
Five of the six events with the greatest FWCs for DRP
(probability of exceedance <15%) occurred during calving
season (February to March). There were cows but no calves
during the sixth event (November) (table 3). The number of
cows ranged from 40 to 69 for 6 to 40 d. The number of
calves ranged from 10 to 61 for 6 to 34 d. Animal units (AU)
varied from 40 to 61(mean 53), and animal unit days (AUD)
varied from 366 to 2127 (mean 1135). The mean FWC for
DRP for the six events was 4.45 mg L-1, compared with
1.47 mg L-1 for the remaining 37 events.
Four of these six events also were among the six events
that had the greatest FWCs for TP (probability of exceed-
ance <15%; table 3). Cattle were present during these four
events (AU = 46 to 61 and AUD = 366 to 1990). There were
no cattle during the other two of the six events with the great-
est FWCs for TP (19 Jan. 2001 and 30 Mar. 2002). However,
23 days prior to the 19 January 2001 event, 54 cattle were
taken off W1 after grazing for 15 d, and 19 days prior to the
30 March 2002 event, 71 cattle were taken off W1 after graz-
ing for 45 d. The mean FWC for TP for the six events was
5.67 mg L-1, compared with 1.88 mg L-1 for the remaining
37 events. The mean FWC for TP for four of the six events
with cattle present was 6.21 mg L-1, compared with 4.58 mg
L-1 for the remaining two events with no cattle present. Alt-
hough the number of observations were too small for statis-
tical comparison with the overall average (table 3), in two of
the six events with the greatest FWC for DRP, the runoff
(16% and 25%) exceeded that of the average. In one each of
the six events with the greatest FWCs for DRP and TP, the
peak discharge rate (173 and 112 L s-1, respectively) ex-
ceeded the overall average peak discharge rate. Values for
other hydrologic parameters were close to or below the over-
all average (table 3).
The six largest loads for DRP and TP (probability of ex-
ceedance <15% for both) occurred during the same six events:
three in the fall, two in midsummer, and one in midwinter
(table 3). The mean loads (kg ha-1) from the six events were
0.37 for DRP and 0.46 for TP, compared with 0.05 for DRP
and 0.06 for TP for the remaining 37 events. These top six
loads accounted for 53% of the total load for each of DRP and
TP. Cattle were present on W1 for one midsummer event (23
cows and 11 calves for 47 d prior to the runoff event), the mid-
winter event (40 cows and 16 calves for 28 d), and one fall
event only (36 cows and no calves for 46 d). For the three
events with cattle, AU varied from 23 to 40 and AUD varied
from 1136 to 1160. The mean loads (kg ha-1) were 0.37 for
DRP and 0.49 for TP for the three events with cattle present,
compared with 0.37 (again) for DRP and 0.42 for TP for the
three events with no cattle present. Compared with the overall
average (table 3), the 1-day rainfall was greater in five of the
six events (55 to 102 mm), the runoff was greater in all six
events (18% to 43%), and the peak discharge rate was also
greater in all six events (173 to 565 L s-1).
The concurrence of cattle with hydrologic events that fa-
vor transport processes (high rainfall, runoff, peak flow,
etc.), led to the six greatest FWCs and loads for DRP and TP.
The pattern was similar for Fe and Al (data not shown).
Some of the hydrologic events causing this were associated
with severe weather (19 Jan., 3 Mar., and 12 Mar. 2001),
some of the state’s wettest periods (1 July 2003), wet
weather as a direct or indirect effect of remnant tropical
storms (hurricanes Frances and Ivan on 22 and 23 Nov. 2004,
and Cindy on 7 July 2005), and stalled weak frontal bounda-
ries (28 Mar. and 2 Apr. 2009). A significant weather phenom-
enon was the fall 2009 El Nino/Southern Oscilla-tion, which
led to very wet conditions across the southeastern U.S. In con-
secutive months from September through December 2009,
the monthly rainfall at W1 and surrounding areas exceeded
the average by 183, 146, 44, and 134 mm, respectively.

Table 3. Details for runoff events that produced the six largest flow-weighted concentrations (FWC) and loads for dissolved reactive phosphorus
(DRP) and total phosphorus (TP).
FWC (mg L-1) Load (kg ha-1)
Date DRP Date TP Date DRP Date TP
Runoff events ranked by FWC and load[a] 1 3 Mar. 2001 7.07 3 Mar. 2001 7.59 12 Oct. 2009 0.45 12 Oct. 2009 0.55
2 12 Mar. 2001 5.53 12 Mar. 2001 7.12 22 Nov. 2004 0.41 1 July 2003 0.51
3 6 Feb. 2002 4.20 6 Feb. 2002 5.31 23 Nov. 2004 0.40 23 Nov. 2004 0.50
4 19 Nov. 2003 3.73 19 Jan. 2001 4.90 1 July 2003 0.34 22 Feb. 2006 0.42
5 21 Feb. 2005 3.26 19 Nov. 2003 4.81 22 Feb. 2006 0.31 22 Nov. 2004 0.40
6 22 Feb. 2006 2.91 30 Mar. 2002 4.27 7 July 2005 0.29 7 July 2005 0.36
Means for hydrology and cattle parameters[b]
1-day rainfall (mm) (45.2) 39.7 - 41.4 - 76.6 - 76.6 -
5-day antecedent rainfall (mm) (26.1) 6.8 - 6.4 - 21.8 - 21.8 -
10-day antecedent rainfall (mm) (45.8) 22.7 - 24.1 - 34.7 - 34.7 -
Runoff (mm) (5.4) 3.2 - 1.1 - 21.2 - 21.2 -
Percent runoff (8.9) 7.7 - 2.2 - 28.6 - 28.6 -
Peak discharge (L s-1) (74.4) 45.3 - 26.3 - 302.4 - 302.4 -
Cattle days during runoff (551) 1751 - 1002 - 825 - 825 -
Mean animal unit days during runoff[c] (379) 1135 - 635 - 576 - 576 -
[a]
Cattle were present during all events except those shown in bold.
[b]
Means from all runoff events are shown in parentheses.
[c]
Computed per USDA-NRCS (2003).

868 TRANSACTIONS OF THE ASABE

FACTOR ANALYSIS ponent) seemed more logical for representing the effects of
Four factors were identified, each of which represented a recent grazing activity. Under factor 4, CDSLR (cattle days
different underlying process (table 4). The variables PD, TR, since the last runoff) was selected because it has a greater
DR1, and DR2 clustered in factor 1 and are identified with loading than DSLR (days since last runoff) and, as above,
hydrologic transport processes of nutrients. The variables the two are correlated (r = 0.671). Equation 9 shows the lin-
DR5, DR10, and DR15 clustered in factor 2 and represent ear regression model adopted on the basis of these consider-
soil antecedent moisture conditions. The variables ations, and table 5 lists the coefficients:
DNOCPR, CCDPR, and CD30DPR clustered in factor 3 and
Y = a + bTR + cPD + dDR10 + eCD30DPR + fCDSLR (9)
represent the level of pasture use by cattle that influences the
degree of nutrient availability. The variables DSLR and where Y is the FWC (mg L-1) or load (kg ha-1) for DRP, TP,
CDSLR clustered in factor 4 and may represent an underly- Fe, and Al.
ing process related to an interaction of environmental and The high r2 and very low p-values for the models (table 5)
cattle influences on nutrient availability. Five of the inde- suggest that these five independent variables serve as appro-
pendent variables were selected from the four factors to de- priate representatives of the drivers within this system. The
velop regression models to predict variations in the FWCs coefficients suggest TR followed by PD as strong drivers for
and loads of DRP, TP, Fe, and Al. The variables TR and PD loads of DRP, TP, Fe, and Al. The r2 values of the models
were selected from factor 1 because FWCs and loads are for the FWCs for DRP and TP were smaller than those for
very much related to TR (used to calculate loads), while PD the loads (~0.44). The coefficient for TR (b) was negative,
is considered a surrogate for runoff intensity. Although the while that for PD (c) was positive. This would be expected,
two are highly correlated (r = 0.874), selecting one without as greater TR leads to dilution. Greater PD suggests greater
the other seemed an incomplete look at the transport process. potential for erosion-related transport of nutrients. Neither
The variables DR10 and CD30DPR were selected from fac- the overall model nor the coefficients were significant for the
tors 2 and 3, respectively, because of their greater loading FWCs for Fe and Al, with the exception of the coefficient
under these factors. In addition, the choice of CD30DPR for CD30DPR (e) for the FWC of Al. The coefficients for
(cattle days 30 days prior to runoff) over CCDPR (continu- CD30DPR (e) were significant for the FWCs and loads for
ous cattle days prior to runoff, of which CD30DPR is a com- DRP and TP, but the magnitudes were only a very small frac-
tion of those for TR and PD. The coefficient for DR10 (d)
Table 4. Matrix table from factor analysis showing the numbers of was significant only for the FWCs and loads of DRP and TP.
retained factors and their loadings after orthogonal varimax The coefficient for CDSLR (f) was significant only for Fe
prerotation.
Factor
load. Overall, it appears that hydrologic transport processes
Variable 1 2 3 4 had a much stronger influence than nutrient sources as driv-
PD 0.8401 0.0712 0.1413 0.2717 ers of nutrient loss.
TR 0.8393 0.0371 0.2266 0.3085 This model was used to estimate FWCs and loads for the
DR1 0.8701 0.1762 0.1871 0.1515 events for which we were not able to collect samples or con-
DR2 0.8482 0.0935 0.1871 0.1705
DR5 0.0083 0.8080 0.0175 0.0054 duct analyses. While the distributions of the data appear to
DR10 0.0400 0.8635 0.0555 0.1914 have a similar pattern as those from the measured values, the
DR15 0.0096 0.8591 0.0605 0.1723 results suggest that cattle and drought impacted P loads ra-
DSLR 0.0307 0.3344 0.1695 0.7547 ther than concentrations during those particular events (data
DNOCPR 0.0336 0.2065 0.7181 0.2074
CDSLR 0.0728 0.0554 0.1302 0.8172 not shown). Total load was estimated as 1.67 kg ha-1 for DRP
CCDPR 0.0229 0.0548 0.8341 0.0835 and 2.07 kg ha-1 for TP.
CD30DPR 0.1298 0.0626 0.8944 0.0184
DSLF 0.1154 0.0539 0.1851 0.0585

Table 5. Parameters for a five-variable linear model (eq. 9) developed from factor analysis for estimating flow-weighted concentration (FWC) and
load for dissolved reactive phosphorus (DRP), total phosphorus (TP), iron (Fe), and aluminum (Al).[a]
Model Statistics Model Coefficients ( 10-3)
Dependent Variable Parameter p-Value r2 a b c d e f
FWC (mg L-1) DRP Estimate 0.001 0.435 84 -117 111 -30 0.17 0.03
p-value 0.518 0.110 0.093 0.062 0.000 0.113
TP Estimate 0.000 0.459 192 -154 147 -41 0.19 0.02
p-value 0.176 0.056 0.041 0.019 0.000 0.164
Fe Estimate 0.428 0.120 13970 159 968 -281 0.80 0.26
p-value 0.000 0.900 0.398 0.313 0.289 0.347
Al Estimate 0.450 0.116 117 23 -13 0.2 -0.04 -0.01
p-value 0.105 0.559 0.707 0.980 0.096 0.391
Load (kg ha-1) DRP Estimate <0.001 0.937 -2177 629 362 -32 0.17 0.02
p-value 0.000 0.000 0.000 0.061 0.001 0.277
TP Estimate <0.001 0.934 -2071 599 396 -41 0.17 0.02
p-value 0.000 0.000 0.000 0.022 0.001 0.333
Fe Estimate <0.001 0.849 -3141 668 315 3 0.02 0.08
p-value 0.000 0.000 0.005 0.900 0.772 0.005
Al Estimate <0.001 0.480 -3625 947 -136 13 -0.22 -0.04
p-value 0.000 0.002 0.602 0.842 0.215 0.488
[a]
Y = a + bTR + cPD + dDR10 + eCD30DPR + fCDSLR, where Y is the FWC (mg L-1) or load (kg ha-1) for DRP, TP, Fe, and Al.

60(3): 861-875 869

DISCUSSION
Aluminum and iron phosphates are common inorganic
forms of P in acid soils (Zaimes and Schultz, 2002).
Schroeder et al. (2004) found that, in Cecil soil, inclusion of
site-specific Fe and Al data can improve the relationship be-
tween soil test P and runoff P. Analyzing for Al and Fe con-
centrations provided possible surrogates for assessing cattle-
induced soil erosion and release of soil-bound P to solution
phase and for identification of pathways of P transport. The
treading of soil and defecation within pastures by grazing
cattle, particularly in heavy-use areas, is expected to release
soil-bound nutrients such as Fe, Al, and P and increase their
transport in runoff because of reduced vegetation cover and
increased soil disturbance (Warren et al., 1986; Russell et al.,
2001; McDowell et al., 2003; Hubbard et al., 2004; Haan et
al., 2006; Cournane et al., 2010). The high correlation found
between loads of Fe and DRP and TP, and a moderately high
correlation found between loads of Al and DRP and TP, ap-
pear to indicate mobilization of these nutrients by dissolution
and transport in runoff as a result of the grazing process. This
would largely have occurred in heavy-use areas, which cov-
ered only a very small percentage of the total pasture area,
whereas grass cover protected much of the pasture from ero-
sion and reduced the potential negative grazing impact.
Of the 13 independent variables considered in factor anal-
ysis, three hydrologic variables (TR, PD, and DR10) and two
cattle-related variables (CD30DPR and CDSLR, surrogates
for nutrient input) were identified as representing the major
influencers of P, Fe, and Al losses (table 5). The variable TR
followed by PD had the greatest influence on P, Fe, and Al
loads compared with the cattle-related variables. The overall
results demonstrate that hydrologic transport processes were
the dominant drivers of pollutant fluxes and highlight the im-
portance of managing grazing to maintaining adequate for-
age cover on pastures.
The FWCs and loads observed in the present study were
in the ranges reported by others for plot and field grazing
studies (Edwards et al., 1996; Nash et al., 2000; Owens and
Shipitalo, 2006; Dougherty et al., 2008; table 6). Results
from those studies indicate that large manure or inorganic
fertilizer application rates, especially shortly before runoff
Table 6. Summary of study description and results for flow-weighted concentration (FWC) and load for dissolved reactive and total phosphorus
(DRP and TP) from 4 cited literature and the current study.
Reference Study Description
Edwards et al. Four 0.6 to 1.5 ha pastures; 31 consecutive months
(1996) (Sept. 1991 to Apr. 1994). Animal manure used prior for DRP for inorganically fertilized fields = 1.7 (range 0.6 to 3.8). Mean total
to study. During the study, two fields continued on
manure, and two fields received inorganic fertilizer. tractible soil P = 177 to 364 mg kg-1, high due to history of manure use.
Nash et al. 34 runoff events (1994-1996) from a 3.6 ha pasture
(2000) managed as part of a producer’s larger farm in Dar-
-1
num, Victoria, Australia; 100 kg P ha , which was
more than three times the recommended rate.
Owens and Rotationally grazed pastures under high and low fer- Mean FWC for total DRP = 0.5 to 2.5. Mean load for total DRP = 0.03 to
Shipitalo tility management near Coshocton, Ohio; 14 years
(2006) (May 1974 to Apr. 1988).
Dougherty 3.5-year monitoring of small (50 m  25 m) grazed
et al. (2008) dairy pastures at Camden, New South Wales, Aus-
tralia. P rates of 0, 20, 40, and 80 kg ha-1 year-1.
This study Zero-order, 7.8 ha, rotationally grazed, beef cattle
pasture near Watkinsville, Georgia, 1999-2009 (n =
-1
43). P fertilizer application: mean of 25 kg ha (n =
8). Estimated P from cattle manure: 41 kg ha-1 year-1.
870
events, are the primary drivers for FWCs and loads. Peak
FWCs for P reported in those studies where fertilizer appli-
cation rates were 1.4 to 3.5 times greater than the values ob-
served in the present study. In one case where runoff oc-
curred the day of fertilizer application, the peak FWC for
DRP was 12 times greater than observed in the present study.
The mean yearly loads reported by Nash et al. (2000), where
the P application rate was 100 kg ha-1 year-1, were orders of
magnitude greater than the peak loads reported in the present
study and in the other studies. The mean extractable soil P
reported by Edwards et al. (1996) was 4 to 9 times greater
than the mean observed in the 2009 sampling at W1 (42 mg
kg-1). The two fields that continued under manure fertiliza-
tion in the Edwards et al. (1996) study had 1.8 times the
mean and 4 to 8 times the maximum FWC for DRP com-
pared to the two fields that were switched to inorganic ferti-
lization, and one manure-fertilized field had 5 times the
mean and maximum DRP loads compared to another field
that was switched to inorganic fertilization.
In the present study, a total of only ~1,550 kg of inorganic
P was applied and only during 2000 to 2003 (table A1). The
average interval between fertilizer application and a runoff
event was 125 d. These two facts likely contributed to the
finding from factor analysis that inorganic fertilization had
no impact on P FWC and load. While the adopted manage-
ment strategy (forage, cattle, fertilization) appears to have
kept nutrient losses at comparatively low levels, the concur-
rence of cattle with hydrologic events that favor transport
processes (high rainfall, runoff, peak flow, etc.) increased
DRP and TP flux. There was P addition from cattle manure
over time at W1. Using manure and manure P production
rates by cattle given in table 10 of Zaimes and Schultz
(2002), a total manure P deposition of 3,517 kg was esti-
mated over the 11 years, amounting to an average of about
320 kg P year-1 or 27 kg P month-1. The total P lost in runoff
at W1 (~40 kg) was a small fraction (~0.80%) of the inor-
ganic P applied and redeposited through manure. The results
presented and discussed above highlight the possible mitiga-
tion of pollutant fluxes through proper grazing management
that includes maintenance of good grass cover and effective
rotational grazing and fertilizer application.
Flow-Weighted Concentration (FWC, mg L-1) and Load (kg ha-1)
Mean FWC for DRP for manure fields = 2.9 (range 0.6 to 24.4). Mean FWC
DRP load = 0.05 to 0.26; greater numbers associated with manure fields. Ex-
Mean FWC for DRP and TP were similar (4.6 to 14.2). Mean yearly loads
were also similar (1.7 to 9.7).
1.23. Ten events in high-fertility pastures had FWC >10 for total DRP. The
greatest FWC for total DRP (85.7) was from runoff occurring the same day as
fertilization. Few large events accounted for most of the total DRP loss.
Mean FWC for TP from zero-fertilizer plot was 1.9. Range of FWC for TP =
0.9 to 11.1.
Mean FWC for DRP = 1.91 (range 0.38 to 7.07). Mean FWC for TP = 2.43
(range 0.36 to 7.59). Mean load for DRP = 0.10 (range 0.00 to 0.45). Mean
load for TP = 0.12 (range 0.00 to 0.55).
TRANSACTIONS OF THE ASABE
CONCLUSIONS
Well-managed grazing systems can provide valuable eco-
system services, including protecting the soil from erosion
and reducing pollutant flux to nearby waterways, but long-
term data are limited to fully support this hypothesis. In this
article, we described flow-weighted concentration and load
data for DRP, TP, Fe, and Al gathered over 11 years (1999-
2009) that had contrasting weather patterns from a 7.8 ha
pasture used for rotational grazing of cow-calf herds near
Watkinsville, Georgia:
 The study demonstrated the possible mitigation of pol-
lutant fluxes through proper grazing management that
includes maintenance of good grass cover, rotational
grazing that minimized over-grazing, and limited fer-
tilizer application. The total P lost in runoff was <1.0%
of the inorganic P applied and redeposited through cat-
tle manure.
 The results demonstrated that surface (and possibly
subsurface) transport processes were the dominant
drivers of pollutant fluxes and highlighted the im-
portance of managing grazing to maintain adequate
forage cover on pastures.
 The concurrence of cattle with hydrologic events that
favor transport processes (high rainfall, runoff, peak
flow, etc.) led to elevated levels of pollutant flux as
demonstrated by the six largest P load events (proba-
bility of exceedance <15%), which accounted for 53%
of the total P loss from all 43 monitored events.
 Drought periods created unfavorable conditions for
hydrologic transport processes and limited pollutant
fluxes.
 The study provided data that can be used to calibrate,
test, and validate water quality models. These data are
particularly useful because they come from a relatively
long-term collection, include periods of variable
weather, and represent a management approach simi-
lar to that of typical producer operations in grazed ar-
eas of the southeastern U.S.
 Nutrient losses observed during the present study are
considered edge-of-field losses.
 There likely were shortcomings in the study, such as
unbalanced distribution of observations between some
of the contrasted variables that might have influenced
the statistical outcomes and subsequent conclusions
about the differences between contrasted variables.
Limitations associated with sample collection
(method, instrumentation, programming, etc.) are also
recognized in arriving at the pollutant loading esti-
mates.
ACKNOWLEDGEMENTS
We appreciate the competent technical support of Ste-
phen Norris and the students and other personnel at the re-
search station who assisted him. We appreciate Clara Parker,
farm manager, and members of the Land Herd Management
Team for management at the pasture and for providing us
with well-compiled data on cattle and pasture management.
We also are grateful for laboratory support from Robert Mar-
60(3): 861-875
tin at Watkinsville, Georgia, and Joyce Alloway, Phyllis Di-
eter, and Don Lightell at Coshocton, Ohio. We appreciate
the contribution of two anonymous reviewers in improving
the clarity and readability of the manuscript.
REFERENCES
Bellows, B. (2001). Nutrient cycling in pastures: Livestock systems
guide. Butte, MT: National Center for Appropriate Technology,
Appropriate Technology Transfer for Rural Areas. Retrieved
from http://www.attra.ncat.org.
Carpenter, S. R. (2008). Phosphorus control is critical to mitigating
eutrophication. Proc. Natl. Acad. Sci., 105(32), 11039-11040.
https://doi.org/10.1073/pnas.0806112105
Carpenter, S. R., Caraco, N. F., Correll, D. L., Howarth, R. W.,
Sharpley, A. N., & Smith, V. H. (1998). Nonpoint pollution of
surface waters with phosphorus and nitrogen. Ecol. Appl., 8(3),
559-568. https://doi.org/10.1890/1051-
0761(1998)008[0559:NPOSWW]2.0.CO;2
Correll, D. L. (1998). The role of phosphorus in the eutrophication
of receiving waters: A review. J. Environ. Qual., 27(2), 261-266.
https://doi.org/10.2134/jeq1998.00472425002700020004x
Cournane, F. C., McDowell, R. W., & Condron, L. M. (2010).
Effects of cattle treading and soil moisture on phosphorus and
sediment losses in surface runoff from pasture. New Zealand J.
Agric. Res., 53(4), 365-376.
https://doi.org/10.1080/00288233.2010.509903
Dougherty, W. J., Nicholls, P. J., Milham, P. J., Havilah, E. J., &
Lawrie, R. A. (2008). Phosphorus fertilizer and grazing
management effects on phosphorus in runoff from dairy
pastures. J. Environ. Qual., 37(2), 417-428.
https://doi.org/10.2134/jeq2007.0049
Edwards, D. R., Daniel, T. C., Murdoch, J. F., & Moore Jr., P. A.
(1996). Quality of runoff from four northwest Arkansas pasture
fields treated with organic and inorganic fertilizer. Trans. ASAE,
39(5), 1689-1696. https://doi.org/10.13031/2013.27686
Endale, D. M., Fisher, D. S., Owens, L. B., Jenkins, M. B.,
Schomberg, H. H., Tebes-Stevens, C. L., & Bonta, J. V. (2011).
Runoff water quality during drought in a zero-order Georgia
Piedmont pasture: Nitrogen and total organic carbon. J. Environ.
Qual., 40(3), 969-979. https://doi.org/10.2134/jeq2010.0411
Franzluebbers, A. J. (2007). Integrated crop-livestock systems in the
southeastern U.S. Agron. J., 99(2), 361-372.
https://doi.org/10.2134/agronj2006.0076
Franzluebbers, A. J., Paine, L. K., Winsten, J. R., Krome, M.,
Sanderson, M. A., Ogles, K., & Thompson, D. (2012). Well-
managed grazing systems: A forgotten hero of conservation. J.
Soil Water Cons., 67(4), 100A-104A.
https://doi.org/10.2489/jswc.67.4.100A
Georgia Adopt-A-Stream. (2008). Getting to know your watershed.
Atlanta, GA: Georgia Department of Natural Resources,
Environmental Protection Division. Retrieved from
http://adoptastream.georgia.gov/sites/adoptastream.georgia.gov/f
iles/related_files/document/WS_Intro.pdf
Haan, M. M., Russell, J. R., Powers, W. J., Kovar, J. L., & Benning,
J. L. (2006). Grazing management effects on sediment and
phosphorus in surface runoff. Rangeland Ecol. Mgmt., 59(6),
607-615. http://dx.doi.org/10.2111/05-152R2.1
Hubbard, R. K., Newton, G. L., & Hill, G. M. (2004). Water quality
and the grazing animal. J. Animal Sci., 82(13S), E255-E263.
https://doi.org/10.2527/2004.8213_supplE255x
Jarvie, H. P., Sharpley, A. N., Scott, J. T., Haggard, B. E., Bowes,
M. J., & Massey, L. B. (2012). Within-river phosphorus
retention: Accounting for a missing piece in the watershed
phosphorus puzzle. Environ. Sci. Tech., 46(24), 13284-13292.
871
 https://doi.org/10.1021/es303562y
Jarvie, H. P., Sharpley, A. N., Spears, B., Buda, A. R., May, L., &
Kleinman, P. J. A. (2013). Water quality remediation faces
unprecedented challenges from legacy phosphorus. Environ. Sci.
Tech., 47(16), 8997-8998. https://doi.org/10.1021/es403160a
McDowell, R. W., Drewry, J. J., Muirhead, R. W., & Paton, R. J.
(2003). Cattle treading and phosphorus and sediment loss in
overland flow from grazed cropland. Soil Res., 41(8), 1521-
1532. https://doi.org/10.1071/SR03042
McDowell, R. W., Nash, D. M., & Robertson, F. (2007). Sources of
phosphorus lost from a grazed pasture receiving simulated
rainfall. J. Environ. Qual., 36(5), 1281-1288.
https://doi.org/10.2134/jeq2006.0347
Murphy, J., & Riley, J. P. (1962). A modified single-solution
method for the determination of phosphate in natural waters.
Anal. Chim. Acta, 27, 31-36. http://dx.doi.org/10.1016/S0003-
2670(00)88444-5
Nash, D. M., Hannah, M., Halliwell, D., & Murdoch, C. (2000).
Factors affecting phosphorus export from a pasture-based
grazing system. J. Environ. Qual., 29(4), 1160-1166.
https://doi.org/10.2134/jeq2000.00472425002900040017x
Owens, L. B., & Shipitalo, M. J. (2006). Surface and subsurface
phosphorus losses from fertilized pasture systems in Ohio. J.
Environ. Qual., 35(4), 1101-1109.
https://doi.org/10.2134/jeq2005.0402
Parry, R. (1998). Agricultural phosphorus and water quality: A U.S.
Environmental Protection Agency perspective. J. Environ.
Qual., 27(2), 258-261.
https://doi.org/10.2134/jeq1998.00472425002700020003x
Pfaff, J. D. (1993). U.S. EPA Method 300.0: Methods for the
determination of inorganic substances in environmental samples.
EPA-600/R-93-100, NIST PB-94-121811. Washington, DC:
U.S. Environmental Protection Agency.
Philipp, D., Popp, M. P., Rumley, E. R., Savin, M. C., & Coffey, K.
P. (2015). Regulatory, production, and consumer-based
challenges of forage-based cattle production systems in the
southeastern United States. Animal Frontiers, 5(4), 24-31.
Russell, J. R., Betteridge, K., Costall, D. A., & MacKay, A. D.
(2001). Cattle treading effects on sediment loss and water
infiltration. J. Range Mgmt., 54(2), 184-190.
https://doi.org/10.2307/4003181
Ryden, J. C., Syers, J. K., & Harris, R. F. (1974). Phosphorus in
runoff and streams. Advan. Agron., 25, 1-45.
http://dx.doi.org/10.1016/S0065-2113(08)60777-4
Schomberg, H. H., Stuedemann, J. A., Franzluebbers, A. J., &
Wilkinson, S. R. (2000). Spatial distribution of extractable
phosphorus, potassium, and magnesium as influenced by
fertilizer and tall fescue endophyte status. Agron. J., 92(5), 981-
986. https://doi.org/10.2134/agronj2000.925981x
Schroeder, P. D., Radcliffe, D. E., Cabrera, M. L., & Belew, C. D.
(2004). Relationship between soil test phosphorus and
phosphorus in runoff. J. Environ. Qual., 33(4), 1452-1463.
https://doi.org/10.2134/jeq2004.1452
Sharpley, A. N., Chapra, S. C., Wedepohl, R., Sims, J. T., Daniel, T.
C., & Reddy, K. R. (1994). Managing agricultural phosphorus
for protection of surface waters: Issues and options. J. Environ.
Qual., 23(3), 437-451.
https://doi.org/10.2134/jeq1994.00472425002300030006x
872
Sharpley, A. N., Daniel, T., Sims, T., Lemunyon, J., Stevens, R., &
Parry, R. (2003). Agricultural phosphorus and eutrophication
(2nd Ed.). ARS-149. Beltsville, MD: USDA Agricultural
Research Service.
Sharpley, A. N., Jarvie, H. P., Buda, A. R., May, L., Spears, B., &
Kleinman, P. J. (2013). Phosphorus legacy: Overcoming the
effects of past management practices to mitigate future water
quality impairment. J. Environ. Qual., 42(5), 1308-1326.
https://doi.org/10.2134/jeq2013.03.0098
Short, S. D. (2001). Characteristics and production costs of U.S.
cow-calf operations. USDA 2001. SB-974-3. Washington, DC:
USDA Economic Research Service.
USDA-NRCS. (2003). National range and pasture handbook.
Washington, DC: USDA Natural Resources Conservation
Service.
USDA-NRCS. (2006). Land resource regions and major land
resources areas of the U.S., the Caribbean, and the Pacific
Basin. USDA Handbook 296. Washington, DC: USDA Natural
Resources Conservation Service.
USEPA. (1991). Methods for the determination of metals in
environmental samples. EPA/600/4-91/010. Cincinnati, OH:
U.S. Environmental Protection Agency, Environmental
Monitoring Systems Laboratory.
USEPA. (1996). Environmental indicators of water quality in the
U.S. USEPA Report R41-R-96-002. Washington, DC: U.S.
Environmental Protection Agency, Office of Water.
Warren, S. D., Thurow, T. L., Blackburn, W. H., & Garza, N. E.
(1986). The influence of livestock trampling under intensive
rotational grazing on soil hydrologic characteristics. J. Range
Mgmt., 39(6), 491-495.
Zaimes, G. N., & Schultz, R. C. (2002). Phosphorus in agricultural
watersheds: A literature review. Ames, IA: Iowa State
University, Department of Forestry.
NOMENCLATURE
CCDPR = continuous cattle days prior to a runoff event
(sum of number of cattle  days spent in pasture)
CDSLR = cattle days since the last runoff event
CD30DPR = cattle days 30 days prior to a runoff event
DNOCPR = days with no cattle prior to a runoff event
DR1 and DR2 = 1-day and 2-day rainfall (mm) leading to
a runoff event
DR5, DR10, and DR15 = antecedent daily rainfall 5, 10,
and 15 days prior to a runoff event
DRP = dissolved reactive phosphorus
DSLF = days since the last fertilization before a runoff
event
DSLR = days since the last runoff event
FWC = flow-weighted concentration (mg L-1)
PD = peak discharge rate (L s-1)
RPR = percent of the 1-day rainfall partitioned to runoff
TP = total phosphorus
TR = total runoff (mm)
TRANSACTIONS OF THE ASABE
 APPENDIX
Table A1. Fertilizer application schedule at W1 from December 1999
to July 2009.
Fertilizer Rate Elemental (kg ha-1)[b]
Date Type[a] (kg ha-1) N P K S
15 Dec. 1999 Lime 2242 - - - -
17 Feb. 2000 34-0-0 224 76.2 - - -
27 Mar. 2000 17-17-17 336 57.1 25.0 47.4 -
3 Aug. 2000 34-0-0 224 76.2 - - -
13 Sept. 2000 Lime 2242 - - - -
5 Nov. 2000 17-17-17 336 57.1 25.0 47.4 -
15 Mar. 2001 17-17-17 336 57.1 25.0 47.4 -
13 June 2001 17-17-17 336 57.1 25.0 47.4 -
15 Jan. 2002 Lime 1121 - - - -
14 Mar. 2002 17-17-17 336 57.1 25.0 47.4 -
19 July 2002 21-0-21 336 70.6 - 47.4 -
24 Sept. 2002 17-17-17 336 57.1 25.0 47.4 -
25 Feb. 2003 17-17-17 336 57.1 25.0 47.4 -
7 Oct. 2003 17-17-17 336 57.1 25.0 47.4 -
5 Apr. 2004 UAN[c] 174 56.0 - - -
7 July 2004 34-0-0 280 95.2 - - -
4 Nov. 2004 Lime 2242 - - - -
1 Jan. 2005 34-0-0 224 76.2 - - -
5 Mar. 2005 34-0-0 336 114.2 - - -
8 Mar. 2006 Urea with sulfur 280 103.6 - - 53.2
20 Aug. 2006 Urea with sulfur 224 82.9 - - 42.6
1 Jan. 2007 Urea with sulfur 224 82.9 - - 42.6
1 Mar. 2008 Urea with sulfur 280 103.6 - - 53.2
1 Aug. 2008 Urea with sulfur 224 82.9 - - 42.6
1 Mar. 2009 Urea with sulfur 280 103.6 - - 53.2
1 July 2009 Urea with sulfur 224 82.9 - - 42.6
[a]
P as P2O5 and K as K2O.
[b]
Dash means not applicable.
[c]
Plus 2.34 L ha-1 Grazon.

60(3): 861-875 873

874 TRANSACTIONS OF THE ASABE
 60(3): 861-875 875

View publication stats