Long-Term and Seasonal Trend Decomposition ... - ACS Publications

Subscriber access provided by RENSSELAER POLYTECH INST

Article

Long-Term and Seasonal Trend Decomposition of Maumee River Nutrient Inputs to Western Lake Erie Craig A. Stow, YoonKyung Cha, Laura T. Johnson, Remegio Confesor, and R. Peter Richards Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/es5062648 • Publication Date (Web): 13 Feb 2015 Downloaded from http://pubs.acs.org on February 18, 2015

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 33

Environmental Science & Technology

1

Long-Term and Seasonal Trend Decomposition of Maumee River

2

Nutrient Inputs to Western Lake Erie

3 4

Craig A. Stow1, YoonKyung Cha2, Laura T. Johnson3, Remegio Confesor3, R. Peter Richards3 1

5

Laboratory, 4840 South State Road, Ann Arbor, MI 48108, USA

6 7

2

School of Natural Resources and Environment, University of Michigan, 440 Church Street, Ann Arbor, MI 48109, USA

8 9

National Oceanic and Atmospheric Administration Great Lakes Environmental Research

3

Heidelberg University, National Center for Water Quality Research, Tiffin, OH 44883 USA

10 11 12 13 14 15

Keywords:Lake Erie, harmful algal blooms, Great Lakes Water Quality Agreement, phosphorus targets

16

1 ACS Paragon Plus Environment


17

Abstract

18

Cyanobacterial blooms in western Lake Erie have recently garnered widespread attention.

19

Current evidence indicates that a major source of the nutrients that fuel these blooms is the

20

Maumee River. We applied a seasonal trend decomposition technique to examine long-term

21

and seasonal changes in Maumee River discharge, and nutrient concentrations and loads. Our

22

results indicate similar long-term increases in both regional precipitation and Maumee River

23

discharge (1975-2013), though changes in the seasonal cycles are less pronounced. Total and

24

dissolved phosphorus concentrations declined from the 1970s into the 1990s; since then total

25

phosphorus concentrations have been relatively stable while dissolved phosphorus

26

concentrations have increased. However, both total and dissolved phosphorus loads have

27

increased since the 1990s because of the Maumee River discharge increases. Total nitrogen and

28

nitrate concentrations and loads exhibited patterns that were almost the reverse of those of

29

phosphorus, with increases into the 1990s and decreases since then. Seasonal changes in

30

concentrations and loads were also apparent with increases since approximately 1990 in March

31

phosphorus concentrations and loads. These documented changes in phosphorus, nitrogen, and

32

suspended solids likely reflect changing land-use practices. Knowledge of these patterns should

33

facilitate efforts to better manage ongoing eutrophication problems in western Lake Erie.

Page 2 of 33

34


Page 3 of 33

35 36


Introduction Eutrophication problems resulting from excessive nutrient inputs have plagued Lake

37

Erie for many years. In 1965 an article in The Economist declared that Lake Erie was “literally

38

dying” 1. To control eutrophication, the 1978 Great Lakes Water Quality Agreement

39

established target total phosphorus (TP) loads for each of the Great Lakes, including an 11,000

40

metric ton per year goal for Lake Erie 2. Following implementation of the 1978 Agreement,

41

notable phosphorus reductions and corresponding water quality improvements were

42

documented 3. However, the Lake Erie phosphorus target is still occasionally exceeded 4,5 and

43

since the mid-late 1990s symptoms of “re-eutrophication” including increased phytoplankton

44

biomass and a reappearance of widespread hypoxia have been documented6,7. In 2011, despite

45

meeting the TP load target, Lake Erie experienced the most extensive algal bloom on record 8.

46

In August 2014 the city of Toledo drinking water supply was temporally shut down due to high

47

toxin levels resulting from a Microcystis bloom.

48

The resurgence of eutrophication in Lake Erie have prompted the National Oceanic and

49

Atmospheric Administration (NOAA) and partners to develop predictive models for harmful

50

algal bloom occurrence 9-11 and issue regular bulletins for potentially affected stakeholders

51

(http://www.glerl.noaa.gov/res/waterQuality/). Additionally, the updated 2012 Great Lakes

52

Water Quality Agreement mandated a review of the existing total phosphorus target loads with

53

a directive to update the Lake Erie target and develop numerical phosphorus concentration

54

criteria within three years 12.

55

Lake Erie’s most severe algal blooms originate in the western basin of the lake, where

56

nutrient inputs from the Maumee River (Figure 1) drive algal growth 6,13. Unlike many Great

57

Lakes tributaries where input data are sparse, making long-term estimation and inference



58

uncertain 14, nearly continuous records of approximately daily concentration data are available

59

for phosphorus, nitrogen, and other constituents in the Maumee River. These data are collected

60

by Heidelberg University and, combined with available United States Geological Survey

61

(USGS) discharge monitoring records, allow detailed analyses of changes in nutrient loads and

62

concentrations since the late 1970s. Previous examination of long-term patterns in the annual

63

phosphorus loads and flow-weighted mean concentrations from the Maumee River revealed

64

substantial increases in bioavailable phosphorus forms since the early 1990s while total

65

phosphorus changed little or decreased slightly 13.

66

Page 4 of 33

In this analysis we build on previous work13 and use an approach to simultaneously

67

examine both long-term and seasonal fluctuations in nutrient concentrations, loads, Maumee

68

River discharge, and regional precipitation. Environmental trend analyses commonly employ

69

linear regression or nonparametric methods based on order statistics, including Kendall’s tau

70

test 15 and variations 16,17. However, these methods are most appropriate for linear or monotonic

71

trends; assessments using them may be inaccurate when the supporting assumptions are not

72

well met or when the data fluctuate with time. Polynomial regression is sometimes applied for

73

nonlinear time-series analysis but this approach uses the entire time-range of observations to

74

estimate nonlinear patterns, whereas nonlinearities are often temporally local features.

75

Generalized additive modeling 18 is a graphical approach to highlight nonlinear patterns in data,

76

based on various localized smoothing techniques (aka smoothers). To highlight seasonal and

77

long-term patterns in Maumee River data we apply a form of generalized additive modeling:

78

Seasonal Trend decomposition using Loess (STL) 19,20 which employs a locally-weighted

79

running-line smoother, and has been previously used to analyze long-term nutrient

80

concentration and river discharge data 21,22. Our results suggest that long-term precipitation


Page 5 of 33


81

increases are partially responsible for discharge and load increases since the mid-1990s, and

82

that the seasonal timing of delivery has also changed for some constituents.

83 84

Methods

85

Water quality data were downloaded from the National Center for Water Quality

86

Research website (http://www.heidelberg.edu/academiclife/distinctive/ncwqr/data). Data from

87

the Maumee River were collected at the Bowling Green Water Treatment Plant, which is

88

located 5.6 km upstream from the USGS stream gage in Waterville, Ohio (USGS #04193500).

89

Samples were collected using an ISCO automated sampler that pulled water from a receiving

90

basin continuously flushed with river water via a submersible pump in the river. Discrete

91

samples were collected either 3 (1974-1988) or 4 (1988-present) times a day and all samples

92

collected during high flow or with high turbidity were analyzed. Otherwise, one sample per day

93

was analyzed. Samples were retrieved weekly and returned to the laboratory for analysis

94

following standard EPA protocols. Dissolved reactive P and total P were both analyzed using

95

molybdate blue colorimetry (EPA method 365.3); total P was pre-treated using persulfate

96

digestion. Nitrate (NO3-N) was analyzed using ion chromatography (EPA method 300.1).

97

Total Kjeldahl N (TKN) was analyzed using phenol colorimetry after pre-treatment with acid

98

digestion (EPA method 351.2). Total N was calculated as the sum of NO3-N and TKN. Finally,

99

total suspended solids (TSS) was analyzed by measuring dry weight after filtration through a

100 101

glass fiber filter (EPA method 160.2). We downloaded average monthly precipitation data for Ohio region 1 from the National

102

Oceanic and Atmospheric Administration (NOAA) National Climatic Data Center, Midwest

103

Regional climate Center web site (http://mrcc.isws.illinois.edu/CLIMATE/index.jsp.).



104

Before analyzing the data we first averaged the Maumee River concentration data by

105

day (when multiple measurements were available) and then averaged the daily data by month to

106

obtain monthly mean concentrations. We estimated loads first on a daily basis by multiplying

107

the average daily concentration by average daily discharge from the USGS stream gage (USGS

108

#04193500), and then on a monthly basis by summing the estimated daily loads. We then used

109

the STL procedure on the average monthly concentrations and total monthly loads. Missing

110

values early in the time series were imputed using long-term monthly means (Supplemental

111

Information – Figure 1).

112

Page 6 of 33

STL applies repeated loess fitting 23 to decompose monthly time-series data into a

113

smoothed trend, seasonal, and residual components. STL uses one continuous loess line for the

114

smoothed long-term component and 12 month-specific loess lines for the seasonal component.

115

Fitting is done on each component iteratively until the resulting trend and seasonal components

116

no longer differ from those of the previous iterations. There are two smoothing parameters in

117

the model, representing the window widths of the seasonal and long-term components. We

118

chose window widths of 48 months and 250 months, respectively, to visually reveal trends.

119

Generally, window-widths are selected to produce relatively smooth long-term patterns that

120

highlight potentially important underlying changes, while simultaneously reducing the residuals

121

to white noise. Attaining both of these goals is a balancing act; sometimes the residuals reveal

122

noteworthy signals 24. However, in this analysis no signals were evident in the residuals, thus,

123

our results focus on the smooth long-term and seasonal patterns.

124 125

Results


Page 7 of 33

126


To illustrate the STL approach we first applied it to the entire available period (1895-

127

2013) of monthly average precipitation data (Figure 2). Over this time-span the monthly data

128

appeared noisy with no obvious visual pattern (Figure 2a). The STL procedure breaks the time-

129

series into three components, a long-term, smoothed trend (Figure 2b, left), a seasonal pattern

130

that changes with time (Figure 2b, middle), and residuals (Figure 2b, right).

131

On the scale of the original monthly variability the long-term, smoothed pattern was

132

almost linear with no obvious trend (Figure 2b, left). Superimposed over the original monthly

133

time-series, the long-term, smoothed pattern went approximately through the middle of the

134

observations, and captured the subtle fluctuations in the data (Figure 2a). When resolution on

135

the vertical scale was expanded an overall increasing pattern was apparent, which deviated

136

from a least squares linear fit through the data (Figure 2c). Through the 1900s a periodicity of

137

approximately 30 years was distinguishable, similar in timing to a corresponding cycle

138

previously reported in Great Lakes water levels 24. A comparison with the least squares line

139

emphasizes an uptick in precipitation beginning in the early 1990s.

140

A view of the seasonal periodicity over the entire period of record was not of sufficient

141

resolution to highlight clear patterns; however, this depiction did indicate that amplitude

142

fluctuations in the cycle have occurred over this time (Figure 2b, middle). When the resolution

143

was increased, and seasonal patterns were depicted by month, several changes became more

144

apparent (Figure 2d). The overall seasonal pattern displayed a minimum in February and a peak

145

in June, and most months display only slight trends or apparent random fluctuations about the

146

overall monthly average. January and March are exceptions, with both months showing a

147

decrease over the period of record.



148

When we focus on data from 1975-2013, the period for which nutrients and sediment

149

measurements are available for the Maumee River, the long-term, smoothed patterns for

150

precipitation and Maumee River discharge were similar. Both precipitation and discharge

151

were relatively stable from approximately 1975 to the late 1980s, declined slightly into the

152

early 1990s, and steadily increased since the early to mid-1990s, with overall increases of

153

approximately 12-16% (Figure 3a,c). By contrast, the seasonal patterns for precipitation and

154

discharge differed (Figure 3b,d). Highest precipitation was from April-August and minimum

155

from January-February, whereas maximum discharge was in March and minimum from

156

August-October (Figure 3b,d). Changes in the seasonal pattern over time were not pronounced,

157

except in March where precipitation and discharge have increased since about 1990, a time

158

approximately coincident with the increase in the long-term, smoothed precipitation and flow

159

(Figure 3a,c).

160

Page 8 of 33

Long-term, smoothed total phosphorus (TP) concentrations declined steadily from 1975

161

until the late 1990s, and since that time have been relatively stable or slightly increasing

162

(Figure 4a). The long-term, smoothed TP load declined similarly to TP concentrations through

163

the late 1990s, but has risen steadily since approximately 2000, reflecting the concurrent

164

increase in discharge, and is currently at levels comparable to those of the mid-1970s (Figure

165

4b). Long-term, smoothed dissolved reactive phosphorus (DRP) concentrations also declined

166

from 1975 until approximately the early 1990s but have increased since then (Figure 4c). The

167

long-term, smoothed DRP load decline from 1975 until approximately 1990 tracked that of the

168

DRP concentration, but since approximately 1990 DRP loads have increased more rapidly than

169

concentrations, with current levels slightly exceeding those of 1975 (Figure 4d).


Page 9 of 33

170


Long-term, smoothed Maumee River nitrogen trajectories were almost the reverse of

171

the phosphorus trends, with increases from 1975 until the early to mid-1990s and declines since

172

that time (Figure 4e,f,g,h). Long-term, smoothed total nitrogen (TN) concentrations increased

173

from 1975 until the early 1990s and have decreased steadily since then (Figure 4d) while the

174

long-term, smoothed TN load increased and remained fairly stable until the early 2000s, but has

175

decreased slightly since then (Figure 4f). Long-term, smoothed nitrate concentration and load

176

patterns over time were similar to that of TN, nitrate being a substantial proportion of total

177

nitrogen (Figure 4g,h).

178

Long-term, smoothed suspended solids (SS) concentrations were relatively stable or

179

slowly declining from 1975 until approximately 1990 and have declined continuously since the

180

early 1990s, though the rate of decline has been relatively slow since approximately 2000

181

(Figure 4i). The overall long-term, smoothed SS load pattern was similar to the concentration

182

pattern except that the load has increased slightly since 2000 (Figure 4j).

183

Phosphorus, nitrogen and suspended solids all followed regular seasonal patterns,

184

although their seasonal concentration cycles differed from one-another (Figure 5a,c,e,g,i). TP

185

concentrations were generally lowest in August-November, rising in December and remaining

186

at relatively high levels into July (Figure 5a). DRP concentrations tended to drop earlier than

187

TP, declining in March and remaining at seasonal lows through November (Figure 5c). Total

188

nitrogen and nitrate both followed a well-defined seasonal pattern with the lowest

189

concentrations in August and September and high, relatively steady concentrations from

190

December-June (Figure 5d,e). Suspended solids concentrations generally peaked in March-

191

April and drop to seasonal lows in August-November (Figure 5f).



192

While these seasonal concentration patterns have been relatively stable over time there

193

are a few changes that, in the context of the ongoing problems in Lake Erie, are worth noting.

194

Both TP and DRP have increased since the 1990s from approximately December-March, the

195

period in which vegetative ground cover is minimal. During much of the rest of the year slight

196

decreases have occurred over this time period (Figure 5a,c). The biggest observable change in

197

the seasonal pattern of total nitrogen and nitrate concentrations was a steady decrease in July

198

(Figure 5e,g). July SS concentrations also steadily declined, while January concentrations have

199

steadily increased and March concentrations have increased since the 1990s, approximately

200

consistent with the patterns for TP (Figure 5i).

201

Page 10 of 33

Constituent load seasonal patterns tracked the discharge seasonal pattern with peaks in

202

March and minima running from roughly July-October (Figures 3d and 5b,d,f,h,j right).

203

Notably, all loads decreased during March until ~1990 and have since increased (Figure

204

5b,d,f,h,j) approximately matching the March discharge trajectory (Figure 3d). This decline

205

and reversal was particularly pronounced for TP, DRP, and SS, all of which also exhibit similar

206

curves in their respective concentrations (Figure 5a,c,i).

207

In addition to changes in both concentrations and loads, we examined differences in

208

DRP:TP and TN:TP ratios over time and seasonally (Figures 6 and 7). The long-term,

209

smoothed DRP:TP ratio declined from 1975 through the early 1990s, and subsequently

210

increased since then (Figure 6a), similar to both the TP and DRP long-term, smoothed

211

concentration patterns (Figure 4a,c). However, the most recent DRP:TP ratio was currently

212

somewhat below that of the beginning of the time-series. Seasonally, this ratio was lowest

213

from March-October and somewhat higher from November – February (Figure 6b). Changes


Page 11 of 33


214

over time during March-October have been minimal or more recently slightly declining, while

215

during November-February the ratio has been recently increasing.

216

The long-term, smoothed TN:TP ratio pattern was nearly the reverse of the DRP:TP

217

pattern. The long-term, smoothed pattern increased through the late 1990s and decreased since

218

then (Figure 7a). Seasonally, this ratio was highest from December-June and lowest in August-

219

October (Figure 7b). In contrast to the DRP:TP ratio, the TN:TP ratio has recently been steady

220

or slightly declining during the seasonal high period (December-June), and increasing during

221

the seasonal low period (August-October) (Figure 7b).

222 223 224

Discussion Establishing the causes of recurring HABs in western Lake Erie is essential to develop

225

effective management strategies. While it would be misleading to ascribe too much meaning to

226

every wiggle in the long-term or seasonal STL results a few signals seem noteworthy. Our

227

results reaffirm previously documented concentration and load increases in the more

228

bioavailable phosphorus fraction 13, but also indicate that discharge increases associated with

229

increased regional precipitation are influencing loads of other important constituents. The

230

contrasts between long-term, smoothed concentration and load patterns revealed by our results

231

underscore the influence of a long-term discharge increase (Figure 3c) on differences in

232

Maumee River loads. For example, smoothed, long-term TP concentrations show a slight

233

increase since the late 1990s, whereas the smoothed, long-term TP load increase since then has

234

been more pronounced, returning approximately to late 1970s levels (Figure 4a,b). Similarly,

235

the recent smoothed, long-term DRP load increase exceeds the corresponding concentration

236

increase (Figure 4d and c, respectively), while smoothed, long-term nitrogen concentration



237

decreases are more pronounced than the corresponding load decreases (Figure 4e,f,g,h).

238

Concurrently, the smoothed, long-term SS concentration decrease since approximately 2000

239

corresponds to a load increase (Figure 4i,j); in this instance the discharge increase has more

240

than offset the concentration decrease.

241

Page 12 of 33

Increases in March river discharge (Figure 3d) and both TP and DRP concentrations

242

and load (Figure 5a,b,c,d) since approximately 1990 may be important factors promoting

243

increased cyanobacterial blooms. Recent studies have suggested that March-June phosphorus

244

inputs are most strongly related to cyanobacterial bloom size in Lake Erie 10,11. The observation

245

that peak seasonal discharge (March) generally precedes peak precipitation (~April-August),

246

suggests that the March discharge is attributable to runoff from melting snowpack and high

247

antecedent soil moisture conditions. Thus, the runoff in March probably reflects activities on

248

the land that have occurred during the preceding winter. Investigating whether increasing

249

phosphorus concentrations during March can be attributed to changing land-uses during the

250

winter months may offer clues for more effective phosphorus management.

251

Baker et al. 13 emphasized the importance of increasing bioavailable P inputs, and while

252

our results confirm recent DRP increases (Figure 4c,d) we also note that during the March-June

253

critical period, both the DRP concentrations (Figure 5c) and DRP:TP concentration ratios

254

(Figure 6b) are at seasonal lows. These seasonal dips likely reflect the observation that SS

255

concentrations are generally highest from March-June (Figure 5i), thus the spring phosphorus

256

pulse includes a relatively high proportion of particulate-associated phosphorus.

257

DRP load has been increasing since the early 1990s (Figure 4d) because both

258

components of load, discharge and concentration, have been rising (Figure 3c and Figure 4c,

259

respectively). We suggest that both of these factors may be contributing to recent increases in


Page 13 of 33


260

HAB occurrence. Baker et al. 25 observed that storm-related discharge pulses pushed DRP

261

further offshore than particulate phosphorus. Similarly, long-term discharge increases may be

262

expanding phosphorus distribution over a larger portion of the lake than would occur by

263

concentration increases alone. Higher discharge can lower hydraulic residence time, and

264

modify local circulation patterns, possibly increasing both TP and DRP delivery efficiency over

265

a broader area. Further investigation using fine-scale hydrodynamic models may be useful to

266

evaluate the extent to which increased discharge is influencing phosphorus availability and

267

resultant primary productivity 26.

268

Our results indicate, over approximately the same period that phosphorus has been

269

increasing, nitrogen has been decreasing (Figure 4e,f,g,h). Thus the TN:TP ratio has also been

270

decreasing (Figure 7a). Reavie et al. 27 reported a decline in the N:P ratio in the western and

271

central basins of Lake Erie, possibly reflecting this trend in the river. The TN:TP ratio is lowest

272

during the summer (Figure 7b) and may underlie reported summer nitrogen limitation in

273

portions of the western Lake Erie basin 28-30, though the overall decline has probably not been

274

sufficient, so far, to effect changes in the dominant phytoplankton species or a shift to nitrogen

275

limitation. We note that the early 1990s onset of the long-term nitrogen decline is

276

approximately coincident with nitrogen drops reported in rivers in the eastern United States,

277

which have been attributed to reduced atmospheric nitrogen deposition resulting from various

278

emission control programs 31. From our results the cause of the decline cannot be ascertained.

279

Because previous studies have implicated changing agricultural practices13,32 as the main cause

280

of the observed phosphorus increase, it seems plausible that regional changes in agriculture

281

might also be responsible for the simultaneous nitrogen decline. However, the possibility that



282

larger-scale processes may also be influencing the observed nitrogen patterns invites further

283

exploration.

284

Page 14 of 33

Declining nitrogen inputs may also influence the toxicity of blooms in Lake Erie; recent

285

studies have highlighted a positive relationship between nitrogen and microcystin

286

production33,34. The influence of the changing ratio of phosphorus and nitrogen on the relative

287

toxicity of Microcystis in western Lake Erie may also be important. Davis et al. 35 reported that

288

toxic Microcystis strains have a greater need for both nitrogen and phosphorus than non-toxic

289

strains, and that toxic strains are preferentially stimulated by inorganic forms of both nutrients.

290

The effect of concurrent phosphorus increases and nitrogen declines is unclear, but merits

291

additional research.

292

The 1978 GLWQA established the use of annual target loads as an accounting

293

procedure to evaluate the effectiveness of actions taken to reduce nutrient pollution. This

294

approach has been widely adopted under the United States Clean Water Act for a range of

295

pollutants under the Total Maximum Daily Load (TMDL) program. While it is generally

296

acknowledged that targets may be exceeded during years of unusually high precipitation and

297

tributary discharge, the use of load targets remains a common management tool. However

298

Milly et al. 36 highlighted the growing recognition that, for variables such as tributary

299

discharge, the assumption of stationarity, in an era of uncertain climate change, poses

300

management challenges. Our results, indicating progressive precipitation and discharge

301

increases in the Maumee River basin (Figure 3) and concurrent phosphorus input increases to

302

Lake Erie (Figure 4a, b, c, d) suggest that imposing fixed load targets may require phosphorus

303

concentrations to be persistently lowered to compensate for increasing discharge, if the targets

304

are to be achieved. As phosphorus load targets are re-evaluated pursuant to the updated 2012


Page 15 of 33


305

GLWQA, it may be appropriate to address the possibility that continued discharge increases

306

into the future may affect target attainment even if phosphorus reduction strategies are

307

successful. Establishing phosphorus “substance objectives” (concentration targets) as required

308

by the 2012 GLWQA37 may be useful to complement load targets as a means to evaluate the

309

effectiveness of phosphorus reduction management actions in the watershed.

310 311 312



313

Acknowledgments

314 315

This work was supported, in part, by a grant from the Great Lakes Restoration Initiative. Cathy Darnell provided assistance with graphics. GLERL contribution number 1746.

Page 16 of 33

316 317


Page 17 of 33


318

References

319

1. The Economist. Goodbye to Lake Erie? 1965, 216: 782.

320

2. International Joint Commission, 1978. Great Lakes Water Quality Agreement of 1978, with

321

Annexes and Terms of Reference Between the United States and Canada, Signed at Ottawa,

322

November 22, 1978, Windsor, Ontario, Canada.

323

3. DePinto, J. V.; Young, T. C.; Mcilroy, L. M., Great-Lakes Water-Quality Improvement - the

324

Strategy of Phosphorus Discharge Control Is Evaluated. Environ Sci Technol 1986, 20, (8),

325

752-759.

326 327 328 329 330

4. Dolan, D. M.; McGunagle, K. P., Lake Erie total phosphorus loading analysis and update: 1996-2002. J Great Lakes Res 2005, 31, 11-22. 5. Dolan, D. M.; Chapra, S. C., Great Lakes total phosphorus revisited: 1. Loading analysis and update (1994-2008). J Great Lakes Res 2012, 38, (4), 730-740. 6. Kane, D. D.; Conroy, J. D.; Richards, R. P.; Baker, D. B.; Culver, D. A., Re-eutrophication

331

of Lake Erie: Correlations between tributary nutrient loads and phytoplankton biomass. J

332

Great Lakes Res 2014, 40, (3), 496-501.

333

7. Scavia, D.; Allan, J. D.; Arend, K. K.; Bartell, S.; Beletsky, D.; Bosch, N. S.; Brandt, S. B.;

334

Briland, R. D.; Daloglu, I.; DePinto, J. V.; Dolan, D. M.; Evans, M. A.; Farmer, T. M.; Goto,

335

D.; Han, H. J.; Hook, T. O.; Knight, R.; Ludsin, S. A.; Mason, D.; Michalak, A. M.;

336

Richards, R. P.; Roberts, J. J.; Rucinski, D. K.; Rutherford, E.; Schwab, D. J.; Sesterhenn, T.

337

M.; Zhang, H. Y.; Zhou, Y. T., Assessing and addressing the re-eutrophication of Lake Erie:

338

Central basin hypoxia. J Great Lakes Res 2014, 40, (2), 226-246.

339 340

8. Michalak, A. M.; Anderson, E. J.; Beletsky, D.; Boland, S.; Bosch, N. S.; Bridgeman, T. B.; Chaffin, J. D.; Cho, K.; Confesor, R.; Daloglu, I.; DePinto, J. V.; Evans, M. A.; Fahnenstiel,



Page 18 of 33

341

G. L.; He, L. L.; Ho, J. C.; Jenkins, L.; Johengen, T. H.; Kuo, K. C.; LaPorte, E.; Liu, X. J.;

342

McWilliams, M. R.; Moore, M. R.; Posselt, D. J.; Richards, R. P.; Scavia, D.; Steiner, A. L.;

343

Verhamme, E.; Wright, D. M.; Zagorski, M. A., Record-setting algal bloom in Lake Erie

344

caused by agricultural and meteorological trends consistent with expected future conditions.

345

P Natl Acad Sci USA 2013, 110, (16), 6448-6452.

346

9. Wynne, T. T.; Stumpf, R. P.; Tomlinson, M. C.; Schwab, D. J.; Watabayashi, G. Y.;

347

Christensen, J. D., Estimating cyanobacterial bloom transport by coupling remotely sensed

348

imagery and a hydrodynamic model. Ecol Appl 2011, 21, (7), 2709-2721.

349 350 351

10. Stumpf, R. P.; Wynne, T. T.; Baker, D. B.; Fahnenstiel, G. L., Interannual Variability of Cyanobacterial Blooms in Lake Erie. Plos One 2012, 7, (8). 11. Obenour, D.R.; Gronewold, A.D.; Stow, C.A.; Scavia, D., Using a Bayesian hierarchical

352

model with a gamma error distribution to improve Lake Erie cyanobacteria bloom forecasts.

353

Water Resources Research 2014, 50, 7847-7860.

354

12. Great Lakes Water Quality Protocol. 2012. Protocol Amending the Agreement between

355

Canada and the United States of America on Great Lakes Water Quality. Signed at

356

Washington, D.C. September 7, 2012.

357

13. Baker, D. B.; Confesor, R.; Ewing, D. E.; Johnson, L. T.; Kramer, J. W.; Merryfield, B. J.,

358

Phosphorus loading to Lake Erie from the Maumee, Sandusky and Cuyahoga rivers: The

359

importance of bioavailability. J Great Lakes Res 2014, 40, (3), 502-517.

360

14. Cha, Y.; Stow, C. A.; Reckhow, K. H.; DeMarchi, C.; Johengen, T. H., Phosphorus load

361

estimation in the Saginaw River, MI using a Bayesian hierarchical/multilevel model. Water

362

Res 2010, 44, (10), 3270-3282.

363

15. Kendall, M. G. Rank Correlation Methods; Charles Griffin:London, 1975.


Page 19 of 33

364 365 366 367 368 369


16. Sen, P. K., Estimates of Regression Coefficient Based on Kendalls Tau. J Am Stat Assoc 1968, 63, (324), 1379-1389. 17. Hirsch, R. M.; Slack, J. R.; Smith, R. A., Techniques of Trend Analysis for Monthly WaterQuality Data. Water Resour Res 1982, 18, (1), 107-121. 18. Hastie, T. J.; Tibshirani, R. J. Generalized Additive Models;Chapman and Hall: London, 1993.

370

19. Cleveland, R. B.; Cleveland, W. S.; McRae, J. E.; Terpenning, I., STL: A seasonal-trend

371

decomposition procedure based on loess. Journal of Official Statistics 1990, 6, (1), 3-33.

372

20. Cleveland, W. S. Visualizing Data; Hobart Press: Summit, NJ, 1993.

373

21. Qian, S. S.; Borsuk, M. E.; Stow, C. A., Seasonal and long-term nutrient trend decomposition

374

along a spatial gradient in the Neuse River watershed. Environ Sci Technol 2000, 34, (21),

375

4474-4482.

376

22. Stow, C. A.; Dyble, J.; Kashian, D. R.; Johengen, T. H.; Winslow, K. P.; Peacor, S. D.;

377

Francoeur, S. N.; Burtner, A. M.; Palladino, D.; Morehead, N.; Gossiaux, D.; Cha, Y.; Qian,

378

S. S.; Miller, D., Phosphorus targets and eutrophication objectives in Saginaw Bay: A 35

379

year assessment. J Great Lakes Res 2014, 40, 4-10.

380 381 382 383 384 385

23. Cleveland, W. S., Robust Locally Weighted Regression and Smoothing Scatterplots. J Am Stat Assoc 1979, 74, (368), 829-836. 24. Sellinger, C. E.; Stow, C. A.; Lamon, E. C.; Qian, S. S., Recent water level declines in the Lake Michigan-Huron system. Environ Sci Technol 2008, 42, (2), 367-373. 25. Baker, D. B.; Ewing, D. E.; Johnson, L. T.; Kramer, J. W.; Merryfield, B. J.; Confesor, R. B.; Richards, R. P.; Roerdink, A. A., Lagrangian analysis of the transport and processing of



386

agricultural runoff in the lower Maumee River and Maumee Bay. J Great Lakes Res 2014,

387

40, (3), 479-495.

388 389 390 391 392 393 394

Page 20 of 33

26. Schwab, D. J.; Beletsky, D.; DePinto, J.; Dolan, D. M., A hydrodynamic approach to modeling phosphorus distribution in Lake Erie. J Great Lakes Res 2009, 35, (1), 50-60. 27. Reavie, E. D.; Barbiero, R. P.; Allinger, L. E.; Warren, G. J., Phytoplankton trends in the Great Lakes, 2001-2011. J Great Lakes Res 2014, 40, (3), 618-639. 28. Chaffin, J. D.; Bridgeman, T. B.; Bade, D. L., Nitrogen constrains the growth of late summer cyanobacterial blooms in Lake Erie. Advances in Microbiology 2013, 3, 16-26. 29. Chaffin, J. D.; Bridgeman, T. B., Organic and inorganic nitrogen utilization by nitrogen-

395

stressed cyanobacteria during bloom conditions. Journal of Applied Phycology 2014, 26, (1),

396

299-309.

397

30. Chaffin, J. D.; Bridgeman, T. B.; Bade, D. L.; Mobilian, C. N., Summer phytoplankton

398

nutrient limitation in Maumee Bay of Lake Erie during high-flow and low-flow years. J

399

Great Lakes Res 2014, 40, (3), 524-531.

400 401 402

31. Eshleman, K. N.; Sabo, R. D.; Kline, K. M., Surface Water Quality Is Improving due to Declining Atmospheric N Deposition. Environ Sci Technol 2013, 47, (21), 12193-12200. 32. Daloglu, I.; Cho, K. H.; Scavia, D., Evaluating Causes of Trends in Long-Term Dissolved

403

Reactive Phosphorus Loads to Lake Erie. Environ Sci Technol 2012, 46, (19), 10660-10666.

404

33. Horst, G. P.; Sarnelle, O.; White, J. D.; Hamilton, S. K.; Kaul, R. B.; Bressie, J. D., Nitrogen

405

availability increases the toxin quota of a harmful cyanobacterium, Microcystis aeruginosa.

406

Water Res 2014, 54, 188-198.


Page 21 of 33


407

34. Yuan, L. L.; Pollard, A. I.; Pather, S.; Oliver, J. L.; D'Anglada, L., Managing microcystin:

408

identifying national-scale thresholds for total nitrogen and chlorophyll a. Freshwater Biol

409

2014, 59, (9), 1970-1981.

410

35. Davis, T. W.; Harke, M. J.; Marcoval, M. A.; Goleski, J.; Orano-Dawson, C.; Berry, D. L.;

411

Gobler, C. J., Effects of nitrogenous compounds and phosphorus on the growth of toxic and

412

non-toxic strains of Microcystis during cyanobacterial blooms. Aquatic Microbial Ecology

413

2010, 61, (2), 149-162.

414

36. Milly, P. C. D.; Betancourt, J.; Falkenmark, M.; Hirsch, R. M.; Kundzewicz, Z. W.;

415

Lettenmaier, D. P.; Stouffer, R. J., Climate change - Stationarity is dead: Whither water

416

management? Science 2008, 319, (5863), 573-574.

417

37. Stow, C.A.; Cha, Y.; Qian, SS., A Bayesian hierarchical model to guide development and

418

evaluation of substance objectives under the 2012 GLWQA. J Great Lakes Res 2014, 40

419

Supplement 3, 49-55.

420 421 422



Page 22 of 33

423

Figure Captions

424

Figure 1

425

Lake Erie Map. Upper right inset depicts all five Laurentian Great Lakes with Erie enclosed in

426

rectangle. Upper left inset shows mouth of Maumee River in the western Lake Erie Basin.

427

Orange area indicates Ohio region 1, the area represented by the precipitation data.

428

Figure 2

429

Panel a - Ohio climate region 1 monthly average precipitation data (1895-2013) with STL long-

430

term trend line (blue) and least squares line (red). Panel b – STL long-term trend centered at

431

zero (left), seasonal pattern (center), and residuals. Seasonal pattern and residuals are shown as

432

deviations from the long-term trend. Panel c – enhanced resolution depiction of STL long-term

433

trend (blue) and least squares line (red). Panel d – STL seasonal patterns depicted by month (blue

434

curves) with corresponding long-term monthly averages (black horizontal lines), shown as

435

deviations from the smoothed long-term trend.

436

Figure 3

437

Panels a and b show enhanced temporal resolution (1975-2013) depiction of Ohio climate region

438

1 monthly average precipitation STL results for comparison with Maumee River discharge STL

439

results over the same period. Panel a shows STL long-term trend line, and panel b shows STL

440

seasonal patterns by month as deviations from the smoothed long-term trend (blue curves), with

441

corresponding long-term monthly averages (black horizontal lines). Panels c and d show

442

analogous Maumee River average monthly discharge STL results for the same period of record.

443

Figure 4

444

STL long-term trend lines (blue) for monthly average Maumee River constituent concentrations

445

(left panels) and corresponding monthly average loads (right panels).


Page 23 of 33


446

Figure 5

447

STL seasonal patterns (blue) and long-term monthly averages (horizontal black lines) for

448

Maumee River constituent concentrations (left panels) and corresponding monthly average loads

449

(right panels). Seasonal patterns shown as deviations from the smoothed long-term trends.

450

Figure 6

451

STL smoothed long-term trend for the DRP:TP concentration ratio (panel a), and the STL

452

seasonal pattern for the DRP:TP concentration ratio (panel b). Seasonal patterns shown as

453

deviation from the smoothed long-term trend.

454

Figure 7

455

STL smoothed long-term trend for the molar TN:TP concentration ratio (panel a), and the STL

456

seasonal pattern for the molar TN:TP concentration ratio (panel b). Seasonal patterns shown as

457

deviation from the smoothed long-term trend.

458



Page 24 of 33

459 460

Figure 1

461

462 463 464


Page 25 of 33

465


Figure 2

466 25 ACS Paragon Plus Environment


467

Page 26 of 33

Figure 3

468 469 470


Page 27 of 33

471


Figure 4

472 473



474

Page 28 of 33

Figure 5

475

476 477


Page 29 of 33

478


Figure 6

479

480 481 482 483



484

Page 30 of 33

Figure 7

485

486 487 488 489 490


Page 31 of 33


491 492

Supporting Information Figure Caption

493

Monthly time-series of the Maumee River constituent concentrations and loads. Imputed values

494

are depicted in yellow. Red line denotes a least squares linear fit among the data, blue line

495

depicts the STL smoothed, long-term trend.

496



497

Page 32 of 33

Supporting Information Figure

498 499 500


Page 33 of 33

501


Table of Contents Art

502

503 504

MODIS satellite image of Lake Erie showing harmful algal bloom in the western basin that

505

shut down the Toledo drinking water supply. August 4, 2014. Credit: NOAA Great Lakes

506

CoastWatch


Long-Term and Seasonal Trend Decomposition ... - ACS Publications

Recommend Documents