Subscriber access provided by UNIV OF W AUSTRALIA
Article
Predicting Sources of Dissolved Organic Nitrogen to an Estuary from an Agro-Urban Coastal Watershed Christopher L. Osburn, Lauren T. Handsel, Benjamin L Peierls, and Hans Paerl Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.6b00053 • Publication Date (Web): 12 Jul 2016 Downloaded from http://pubs.acs.org on July 25, 2016
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 39
1 2
Environmental Science & Technology
Predicting Sources of Dissolved Organic Nitrogen to an Estuary from an Agro-Urban Coastal Watershed
3 4
Osburn, Christopher L.1*, Handsel, Lauren T.1#, Peierls, Benjamin L.2, and Paerl, Hans W.2
5 6 7 8 9
1
Department of Marine, Earth, and Atmospheric Science, North Carolina State University, Raleigh, North Carolina, 27695 USA 2
Institute of Marine Sciences, University of North Carolina at Chapel Hill, Morehead City, North Carolina, 28557 USA
10
*
11
#
corresponding author, +1-919-515-0382;
[email protected] now at Cardno, Inc., Raleigh, NC
12 13
Version: 12-Jul-16
14
1 ACS Paragon Plus Environment
Environmental Science & Technology
15 16
Page 2 of 39
ABSTRACT Dissolved organic nitrogen (DON) is the nitrogen (N)-containing component of dissolved
17
organic matter (DOM) and in aquatic ecosystems is part of the biologically-reactive nitrogen
18
pool that can degrade water quality in N-sensitive waters. Unlike inorganic N (nitrate and
19
ammonium) DON is comprised of many different molecules of variable reactivity. Few methods
20
exist to track the sources of DON in watersheds. In this study, DOM excitation-emission matrix
21
(EEM) fluorescence of eight discrete DON sources was measured and modeled with parallel
22
factor analysis (PARAFAC) and the resulting model (“FluorMod”) was fit to 516 EEMs
23
measured in surface waters from the main stem of the Neuse River and its tributaries, located in
24
eastern North Carolina. PARAFAC components were positively correlated to DON
25
concentration. Principle components analysis (PCA) was used to confirm separation of the eight
26
sources and model validation was achieved by measurement of source samples not included in
27
the model development with an error of 70% of DON was attributed to natural
29
sources, non-point sources, such as soil and poultry litter leachates and street runoff, accounted
30
for the remaining 30%. This result was consistent with changes in land use from urbanized
31
Raleigh metropolitan area to the largely agricultural Southeastern coastal plain. Overall, the
32
predicted fraction of non-point DON sources was consistent with previous reports of increased
33
organic N inputs in this river basin, which are suspected of impacting the water quality of its
34
estuary.
2 ACS Paragon Plus Environment
Page 3 of 39
35
Environmental Science & Technology
INTRODUCTION
36
Eutrophication, the increase in organic matter supply to aquatic ecosystems, is a
37
widespread problem often linked to anthropogenic nutrient enrichment in estuaries, especially
38
nitrogen (N), since it is the primary nutrient limiting algal production1-4. Unlike inorganic N (
39
nitrate, nitrite, ammonium), dissolved organic N (DON) in aquatic systems is likely made up of
40
diverse compounds (e.g., amino acids, urea, humic substances) with varying reactivity,
41
bioavailability, and concentration, which, along with inorganic N, supports the growth of
42
phytoplankton and bacteria5-10, and may differentially favor certain phytoplankton taxa,
43
including harmful algal bloom (HAB) species11-13. Prior work suggests that riverine fluxes of
44
DON will increase in the future due to anthropogenic and climactic factors14-17. In rivers, DON
45
can comprise a substantial fraction of the total dissolved N (TDN, or DON + DIN; the latter
46
being the sum of nitrate, nitrite, and ammonium), load from forested catchments, yet in urban
47
and agricultural catchments, this ratio typically is smaller18.
48
Tracking these sources of N (and other organic and inorganic nutrients) is of key
49
importance for computing load estimates from streams and rivers to receiving waters such as
50
lakes, reservoirs and estuaries. Sources of OM have been tracked by using their fluorescence
51
properties (e.g., chromophoric dissolved organic matter, CDOM) (e.g., refs. 19-22). For
52
example, the indole moiety directly imparts characteristic protein-like fluorescence properties to
53
organic matter, including N bound to aromatic macromolecular material (so-called “humic
54
substances”), making this technique relevant for measurements of DON23.24. Fluorescence of
55
filtered water samples is relatively easy to measure and is highly informational, and can be
56
evaluated using multi-way analysis techniques, especially parallel factor analysis
57
(PARAFAC25,26). EEM-PARAFAC models have been developed for a number of natural and
3 ACS Paragon Plus Environment
Environmental Science & Technology
58
artificial systems ranging from rivers, lakes, and estuaries to recycled water and water treatment
59
systems and a searchable database (OpenFluor, http://www.openfluor.org) exists where new
60
models can be compared to existing models27,28.
61
Substantial challenges exist to utilizing EEM-PARAFAC models as water quality
62
proxies. First, the signals (components) from a PARAFAC model are not unique to specific
63
compounds but rather are thought to be specific to classes of molecules. Thus, PARAFAC
64
models of organic matter do not identify discrete fluorophores23,29. This means empirical
65
relationships between fluorescent components and water quality parameters must be developed.
66
Second, any water quality model developed using organic matter fluorescence potentially is
67
system specific. In other words, a model developed for one watershed may not be directly
68
applicable to other watersheds.
69
Page 4 of 39
Rivers impacted by different land uses exhibit distinct fluorescence patterns30. EEM
70
fluorescence of farm wastes (swine and cattle slurry, sheep barn waste) and rivers influenced by
71
sewage and by wastewater treatment facility (WWTF) effluences are enriched in protein-like
72
fluorescence resembling the amino acids tryptophan and tyrosine31. These differences are distinct
73
from natural organic matter (NOM) in streams and have important implications for organic
74
matter metabolism in and health of aquatic ecosystems32.
75
The aim of this study was to use organic matter fluorescence as a means to track potential
76
sources of DON in the Neuse River Basin (NRB), a coastal river watershed in eastern North
77
Carolina, whose estuary (the Neuse River Estuary, NRE) is a major tributary to the Pamlico
78
Sound, the nation’s second largest estuarine complex. Fluorescence was measured on a series of
79
eight potential sources of DON to this river system and modeled with PARAFAC. We
80
hypothesized that PARAFAC model components determined for the eight sources would be 4 ACS Paragon Plus Environment
Page 5 of 39
Environmental Science & Technology
81
correlated to source DON concentrations. Further, we expected that when applied to streams and
82
the Neuse River proper, fluorescence would be dominated by natural background but reflect
83
contributions from urban and agricultural sources of organic matter in response to land cover and
84
land use. The output from PARAFAC was used in a mixing model, termed FluorMod, to
85
estimate the relative amounts of DON originating from the eight sources at three locations along
86
the Neuse River proper and three tributaries to the river. Each site was located along a land use
87
gradient from the urbanized Raleigh-Durham metropolitan area in the Piedmont to the rural and
88
agricultural-Atlantic coastal plain of the US. We discuss the development, calibration, and
89
validation of this modeling approach and its utility for monitoring nutrient water quality.
5 ACS Paragon Plus Environment
Environmental Science & Technology
90
METHODS AND MATERIALS
91
Sample Collection.
92
Page 6 of 39
The Neuse River Basin is approximately 16,000 km2 in size and the river itself is about
93
320 km in length, originating as outflow from Falls Lake, a reservoir near the Raleigh
94
metropolitan area, and flowing southeasterly from the northern Piedmont through the Atlantic
95
coastal plain through an increasing rural and agriculturally dominated landscape. Nutrient
96
loading is primarily from nonpoint sources to tributaries and the Neuse River proper33.
97
Concentrated animal feeding operations (swine, poultry) are heaviest in the Middle Neuse near
98
the city of Kinston, NC. The lower Neuse River and upper estuary have experienced nutrient
99
loadings sufficient enough to cause water quality issues such as algal blooms and hypoxia4,34. A
100
total maximum daily load for total N has been implemented for this watershed35 (see Supporting
101
Information). Monthly samplings were conducted in the NRB at 10 sites (Table S1). Samples
102
were collected from just below surface using water samplers lowered into streams from bridge
103
overpasses approximately mid-stream, kept on ice or cold, in the dark, until returned to the
104
laboratory, generally within 24 hours. In the laboratory, samples were filtered through 0.7 µm
105
porosity glass fiber filters (pre-combusted at 450 °C for 6 hours) to collect particulates (POM),
106
the filtrate (DOM) was collected in detergent-washed and ultrapure water-rinsed polycarbonate
107
or glass vials and kept refrigerated and in the dark for up to one week until analysis.
108
Optical analyses. Absorbance (200 – 800 nm) was measured on filtered samples using a Varian
109
Cary 300UV spectrophotometer in 1-cm, 5-cm, or 10-cm quartz cells and diluted if absorbance
110
was greater than 0.4 at 240 nm (1-cm cells only). Ultrahigh purity laboratory water (18.2 MΩ
111
resistivity) was used as a blank, and the blank-corrected absorbance values were converted to
112
Napierian absorption coefficients (aλ) using the following equation: 6 ACS Paragon Plus Environment
Page 7 of 39
Environmental Science & Technology
aλ = 2.303 ×
113
Aλ ,meas − Aλ ,blank L
114
where Aλ,meas is the measured Absorbance (aka optical density “OD”; unitless) of a sample at
115
wavelength, λ; Aλ,blank is the measured Absorbance of the Milli-Q water blank and L is the
116
pathlength in meters.
(1)
117
Fluorescence was measured on samples, diluted to match the absorbance measurements,
118
if necessary, on a Varian Eclipse spectrofluorometer. Excitation (Ex) was measured from 240 to
119
450 nm, in 5 nm increments, and emission (Em) was measured from 300 to 600 nm at 2 nm
120
intervals. Slit widths of 5 nm were used in both Ex and Em modes and scanning speed was set to
121
9600 nm/min with an integration time of 0.0125 s. Corrections for lamp intensity, detector
122
response, inner-filtering effects, and dilution were applied20 (see Supporting Information). Final
123
values were calibrated in quinine sulfate units (QSU, where 1 QSU = 1 ppb quinine sulfate).
124
FluorMod design – We took a forward modeling approach wherein probable sources of organic
125
N were measured and modeled and then fit to stream water samples. Fluorescence was measured
126
on discrete point and non-point sources of DOM to the NRB (Fig. 1). Details on source
127
acquisition and preparation are in the Supporting Information. Two ‘Reference’ streams
128
represented the natural background source of ON in unimpacted streams within the NRB. These
129
streams are classified as “outstanding source waters (OSW)” by the North Carolina Department
130
of Environment and Natural Resources (NCDENR)35. ‘Influent’ was raw sewage inflow to
131
WWTF that are permitted discharges into the Neuse River and several of its tributaries.
132
‘Effluent’ was treated water discharged from those WWTF. ‘Swine’ was surface samples of
133
swine lagoons collected by farmers and sent to the North Carolina Department of Agriculture
134
(NCDA) laboratory for nitrogen analysis. ‘Poultry’ was water-soluble extracts (“leachates”) 7 ACS Paragon Plus Environment
Environmental Science & Technology
Page 8 of 39
135
leached from turkey, hen, and broiler litters. ‘Street’ samples were runoff collected roadside near
136
storm drains in Raleigh, NC, during and after rain events. ‘Septic’ was samples collected from
137
residential community septic discharge ditches around the City of Durham, NC. ‘Soil’ samples
138
were water-soluble extracts prepared in similar fashion to Poultry leachates.
139
All source samples were scanned for absorbance and fluorescence and processed as
140
described for the stream and river samples. Some soil leachates and all of the swine lagoon and
141
poultry leachates were highly concentrated and required substantial dilution with MilliQ water
142
(up to 1:1000). These dilution values were recorded and used to correct final values of
143
absorbance and fluorescence. Street runoff and WWTF Effluent contained residual natural DOM
144
signals that may obscure modeling. Mean normalized fluorescence intensity (I) of the Reference
145
stream measurements was thus subtracted from mean-normalized Effluent and Street
146
fluorescence. The residual was rescaled, negative values zeroed, and used as the discrete
147
fluorescence fingerprint of each source (Fig. S1).
148
A total of 180 source EEMs were modeled with PARAFAC to create FluorMod and
149
resulted in a 9 component model that was validated through analysis of leverages (using a
150
criteria 70% (Table S5). Maximum Reference contribution was 91% at Contentnea Creek and
302
minimum Reference contribution was
