environmental engineering case study short answer problems

azheng

LakeWashingtonReadingPackage.pdf

Home >Engineering homework help >environmental engineering case study short answer problems

O C T O B E R 2 0 1 0 C i v i l E n g i n e e r i n g [43]

C O

U R

T E

S Y

O F

T H

E U

.S .

A R

M Y

C O

R P

S O

F E

N G

IN E

E R

S ,

S E

A T

T L

E D

IS T

R IC

Linking the Lakes: The Lake Washington Ship Canal

C IVIL ENGINEERS usually make decisions that are quantitative and require data and mathemati- cal prowess. Other decisions, however, are not so straightforward. In some instances the civil engi-

neer must seek to gauge the effect of his or her project not just on the community but on the region, the state, and even the country as a whole. Such was the case more than a century ago when the citizens of Seattle asked the U.S. Army Corps of En- gineers to construct a canal that would link Lake Washing- ton to Puget Sound. In performing this task, such prominent engineers as Hiram M. Chittenden and James B. Cavanaugh were faced with a number of formidable challenges, among them where to construct the canal, whether to lower or raise the levels of the lakes involved, and how to link salt water with freshwater without destroying an entire ecosystem.

Seattle was settled on the eastern shore of Elliott Bay, an arm of Puget Sound. Located north of the bay are, from west to east, Shilshole Bay and Salmon Bay, both saltwater, and Lake Union and Lake Washington, both freshwater. Lake Washington is a long, ribbonlike lake that forms Seattle’s eastern border.

As early as 1854 a settler by the name of Thomas Mercer voiced the need for a canal that would link Lake Washington to the sound. Several years later John McGilvra, a local real estate investor, mentioned how a canal would help improve lakeside properties, and H.L. Pike unsuccessfully attempted to excavate a canal through the narrow ridge separating Lake Union from Lake Washington.

After the Civil War, the U.S. Army Corps of Engineers be- lieved that Lake Washington would be an appropriate place for a naval station, a station that would require a link to Puget Sound. In late 1871 Lieutenant Thomas H. Handbury sub- mitted a proposal to spend $4.7 million to excavate a channel that would connect Lake Washington to Lake Union. He also proposed constructing a channel between the south end of

Lake Union and Elliott Bay.

The government soon scrapped the project, however, in favor of a base on Sinclair Inlet at the town of Bremerton, southeast of Bainbridge Island.

In 1881 a number of Seattle investors founded the Wash- ington Improvement Company to fund the construction of a canal between Lake Washington and Lake Union. Four years later a workforce of Chinese immigrants excavated a 16 ft wide waterway between the two lakes. Named the Portage Canal, it featured two locks and was primarily used to trans- port logs to sawmills on Lake Union.

After the fi re in Seattle that destroyed most of the business district in 1889, the city’s sawmill owners relocated north to the small community of Ballard, located along the banks of Salmon Bay. The booming lumber industry, coupled with the phenomenal growth in population in Seattle in the 1890s, increased the demand for a canal extending to Puget Sound.

The demand did not fall on deaf ears. In 1891, two years af- ter Washington had become a state, the Corps formed a board to study fi ve possible routes for a canal that would connect Lake Washington to the sound. The fi rst followed Black Creek, the natural outlet of Lake Washington. The creek ran southward until it encountered the Duwamish River, which from there

[42] C i v i l E n g i n e e r i n g O C T O B E R 2 0 1 0

H i s t o r y L e s s o n fl owed northwest through a valley and into Elliott Bay. The second and third options involved direct links between Lake Union and Elliott Bay. The fourth and fi fth options, however, proposed linking Lake Union to Salmon Bay and then build- ing a second canal either west to Shilshole Bay or south to Smith Cove, located on the northwest side of Elliott Bay.

The board determined that the route along Black Creek and the Duwamish River would be impracticable. It also found the land between Lake Union and Elliott Bay to be too expen- sive, leaving the Salmon Bay options as the most feasible. The board estimated that a canal would cost $3.5 million if it ex- tended to Smith Cove and $2.9 million if it went to Shilshole Bay. Despite this signifi cant difference in cost, the Smith Cove route was preferred because it was closer to Seattle’s harbor and would, if necessary, be easier to defend militarily.

Seattle’s residents greeted the board’s report enthusias- tically, but those outside the city opposed the plan, nick- naming it Seattle’s Ditch. As a result of its cost and limited popularity, the Corps decided to omit the canal from its rec- ommended projects for 1892 and to leave its construction, as well as its funding, to local entities.

Eugene Semple, a former governor of Wash- ington Territory, believed that a privately constructed ca- nal was possible but not in the form rec- ommended by

the Corps. He was of the opinion that a canal con-

structed at the mouth of the Duwamish River could be ex-

tended through a man-made cut in a ridge referred to locally as Beacon

Hill. From there it would run through Rainier Valley and another ridge before

reaching Lake Washington. Eager to establish his credibility and attract investors,

Semple engaged the services of Captain Thomas Symons of the Corps district based in Portland, Oregon. Even though

the Corps was not offi cially involved, Symons believed it per- missible to assume the role of consultant. He drafted a report of his opinions about the practicability of constructing a ca- nal according to Semple’s design. Believing he was working in the best interests of the public, Semple presented Symons’s report as an offi cial endorsement by the Corps and used it to entice backers. By 1893 he had secured authorization from the state legislature, and the following year he succeeded in convincing eastern investors to fund the project.

Meanwhile, Ballard residents and industry owners decid- ed to ask for federal funding for their own canal. In 1894, as Semple’s project took shape, the U.S. Congress approved the formation of a committee to consider a canal linking Lake Union to the sound and assigned Symons to chair the com- mittee. The business owners called attention to the fact that the captain was also a consultant for Semple’s canal, nick- named the South Canal, and saw his involvement in the new- ly proposed “North Canal” as a confl ict of interest. The Corps transferred Symons to Buffalo, New York, in 1895.

Before being transferred, however, Symons complet-

ed a survey of the North Canal. He confi ned the survey to the Shilshole Bay location proposed in the 1891 survey. His report also suggested a dam and lock at the mouth of Salmon Bay. A second lock would be placed at Lake Washington so that the lake could retain its el- evation, which was approximately 7 ft higher than Lake Union. The estimated cost of the project, as of June 1895, was an unre- alistically low $1.4 million.

The move by Congress in 1894 also required that King County make the necessary land purchases for the canal’s right- of-way. The county offi cials, however, experienced diffi culty lo- cating land owners and completing the necessary legal steps to

HIRAM M. CHITTENDEN’S 1907 MAP OF THE PROPOSED CANAL ALIGNMENT CONNECTING PUGET SOUND WITH LAKE UNION AND LAKE WASHINGTON

Civ. Eng., 2010, 80(10): 42-45

D ow

nl oa

de d

fr om

a sc

el ib

ra ry

.o rg

b y

U ni

ve rs

it y

of W

as hi

ng to

n L

ib ra

ri es

o n

07 /1

0/ 17

. C op

yr ig

ht A

S C

E . F

or p

er so

na l

us e

on ly

; al

l ri

gh ts

r es

er ve

procure the lands, and in many cases the parcels purchased were overpriced. It was not until June 1900 that the county was able to turn the right-of-way over to the federal government.

Despite the diffi culties in securing the right-of-way, Cap- tain Harry Taylor, who was named head of the newly estab- lished Corps district headquartered in Seattle, suggested that the Smith Cove terminus be reconsidered because of a drop in land prices. As a result, the U.S. Department of War changed the canal location, and once again the county set about ac- quiring the right-of-way.

The seemingly last-minute change did not please the

owners of the Great Northern Railway Company, for they had property interests along that route. In fact, the rail company threatened to abandon Seattle altogether if the canal was constructed at Smith Cove. Faced with this ultimatum, the De- partment of War reverted to the Shilshole Bay ter- minus, probably to the frustration of the county offi cials attempting to procure the right-of-way for the Smith Cove route.

Another obstacle cropped up at the turn of the century. During the previous decade commerce to and from Seattle and the size of the vessels traveling through Puget Sound had grown, and the former specifi cations for the locks had become inadequate. Taylor proposed a new plan for locks that would be approximately 800 ft long and 100 ft wide and would ac- commodate vessels with a maximum draft of 30 ft.

In the meantime Semple had been working on excavating the South Canal through Beacon Hill. Using high-pressure hoses, he washed away the soil, and a gravity fl ume carried it to the tidal fl ats below. Although this practice was successful at fi rst, it would not be effective for very long because the an-

gle of the fl ume would be reduced as the hill itself was exca- vated. Furthermore, the high-pressure hoses and fl ume were useless against large boulders.

It seemed that Semple understood these diffi culties but counted on the fact that the North Canal’s cost would spiral out of control. Then, he hoped, the government would cease construction there and fund his project.

Semple was right on one count. The estimated cost of the North Canal had been woefully low. Major John Millis, the Corps’s new district engineer, stated in 1901 that the new locks would cost approximately four times as much as the original

ones and that the project’s overall price tag would be a prohibitive $6.3 million.

This was Semple’s chance. He of- fered to sell his canal to the government upon completion for just $2 million. Advocates for both the North Canal and the South Canal approached Con- gress, which commissioned another study by the Corps to determine the feasibility of the two canals.

The board of Corps offi cials visit- ed Seattle in August and November of 1901. To Semple’s dismay, it dismissed the South Canal almost immediately because of the height of Beacon Hill and the other ridge. The board con- cluded that fi ve times as much material would have to be removed to construct the South Canal than to build the other. It also determined that the South Canal would cut through nearby streets, water mains, and railroad tracks.

The North Canal, according to the board, was feasible from an engineer- ing standpoint. Nevertheless, the board raised another concern. It questioned

whether the commercial needs of Seattle would jus- tify the cost of the canal’s construction, operation, and maintenance. Because of the board’s concerns, the Corps’s offi cial position became that no canal of any kind could be justifi ed by economic need.

Semple either was unwilling to give up on his investment or was as unaware as his Dickensian name implied. He attempted to continue with his

canal by trying to woo potential investors. Nevertheless, the disparaging comments about the South Canal in the board’s report discouraged potential backers. In May 1904 the city forced Semple to halt operations.

It was under these bleak conditions that Major Hiram M. Chittenden replaced Millis as the district engineer in 1906. A native of western New York, Chittenden was an intelligent and enthusiastic individual who had abandoned a career in law to join the Corps. Prior to his assignment in Seattle, he had declined the superintendency of Yellowstone National Park.

Upon his arrival, Chittenden believed that the completion of the Lake Washington Ship Canal, as it came to be named, would be his highest priority. Because of a nervous condition

[44] C i v i l E n g i n e e r i n g O C T O B E R 2 0 1 0

P E

M C

O W

E B

S T

E R

& S

T E

V E

N S

C O

L L

E C

T IO

N ,

M U

S E

U M

O F

H IS

T O

R Y

& I

N D

U S

T R

that often left him incapacitated, he believed this would be his fi nal engineering project, and he wanted it to be a success.

Chittenden saw that he would have to work quickly, how- ever. A prominent Seattle investor by the name of James A. Moore was negotiating with the county to construct a wood- en lock at the eastern end of Salmon Bay. Chittenden had no faith in Moore’s engineering acumen and believed that the inadequately constructed wooden lock would eventually be turned over to the Corps, which would then have to maintain it. He later stated that the lock “could never have been built on a basis of safety and it would surely have collapsed sooner or later and precipitated Lake Wash- ington into Puget Sound.” Never- theless, because the Corps had no plans to construct any locks at that time, Chittenden could not legally halt construction of the lock.

Chittenden’s hands may have been tied legally, but he was will- ing to investigate other strategies. He befriended local leaders and took them for rides on the lakes in his own boat. It was on these trips that Chittenden would express his doubts about Moore’s project, or as he put it, he “spread a leaven of doubt as to the whole scheme and this continued to develop until it overthrew the Moore project.”

Chittenden’s tactics proved ef- fective. In 1907 the newly formed Lake Washington Canal Associa- tion endorsed new terms for fund- ing the canal’s construction by the Corps. Chittenden himself drafted the terms and convinced Congress that the previous assessment ques- tioning the commercial justifi cation for the ca- nal was ill founded. On the basis of a study of his own that he completed by the end of 1907, he concluded that the project was fi nancially feasible. Without local support, Moore knew he was beaten, and he transferred his rights to construct the lock to the Lake Washington Ca- nal Association.

Chittenden quickly formed new plans for the canal. Believing that a single lock at Salmon Bay would be in- suffi cient and uneconomical, he suggested the construction of two locks: one for small steamers, tugboats, and recreational boats and one for large merchant ships. The small lock was to be 150 by 30 by 16 ft; the larger, 825 by 80 by 36 ft. Both would be capable of raising vessels approximately 20 ft from sea level to the elevation of Lake Union. He also omitted the construction of a lock at the eastern terminus of the canal so that the level of Lake Washington could be lowered by 9 ft to the same elevation as Lake Union.

By 1908 Chittenden’s revised plans for the canal had been approved. Unfortunately, his health left him bedridden, and

he underwent painful shock treatments. In December 1909 he retired from the Corps.

Chittenden may have been gone, but his plans moved forward. In early 1910 the state legislature appropriated $250,000 for the project. The fi rst contract was to excavate a canal linking Lake Washington and Lake Union. The new canal would be located just north of the smaller Portage Ca- nal, which had been constructed by Seattle investors nearly 30 years earlier.

Major James B. Cavanaugh was appointed the district engineer for Seattle in 1911 and was placed in charge of the

project. The following year workers excavated nearly 250,000 cu yd of material, and by early 1913 the fi rst concrete was poured into forms for the lock walls.

To keep salt water carried in the locks from infi ltrating Lake Union and Lake Washington and destroy- ing the ecosystem, a saltwater basin 200 ft long and 200 ft wide was con- structed above the locks. The heavier salt water settling in the basin was to be carried through a drainage pipe with a cross section of 30 sq ft to an adjacent spillway dam, and from there it would exit approximately 4 ft below the mean high tide. The fl ow through the pipe was continu- ous at a rate of 100 to 3,200 cfs.

The locks were completed on Au- gust 3, 1916, three weeks before the new canal between Lake Union and Lake Washington was completed. By the end of 1916, more than 7,500 vessels, approximately 12,000 pas- sengers, and 201,000 tons of freight

had passed through the locks. The offi cial dedi- cation of the project took place on July 4, 1917, despite the fact that some dredging and bank re- vetment work was still needed. The project also included a 10-step fi sh ladder to facilitate the mi- gration of salmon and trout. The total cost of the undertaking was approximately $3.5 million.

The locks were later named the Hiram M. Chittenden Locks, after the engineer who

worked so tirelessly to make them possible. Now used mostly for recreational boats and commercial fi shing ves-

sels, the locks and canal are still in operation. The waterway is a monu- ment to the history of canal construc- tion as well as to the vision guiding the decisions of the civil engineer. The Lake Washington Ship Canal is listed in the National Register of Historic Plac- es and has been recognized in ASCE’s Historic Civil Engineering Landmark Program. —BRETT HANSEN

O C T O B E R 2 0 1 0 C i v i l E n g i n e e r i n g [45]

C O

U R

T E

S Y

O F

T H

E U

.S .

A R

M Y

C O

R P

S O

F E

N G

IN E

E R

S ,

S E

A T

T L

E D

IS T

R IC

Chittenden suggest- ed that two locks be constructed—one for small steamers, tug-

boats, and recreation- al boats and one for

large merchant ships.

Major Hiram M. Chitten- den was from western

New York. Intelligent and enthusiastic, he suffered from poor health and for this reason was intent on completing the Lake Washington Ship Canal.

Hansen

Civ. Eng., 2010, 80(10): 42-45

D ow

nl oa

de d

fr om

a sc

el ib

ra ry

.o rg

b y

U ni

ve rs

it y

of W

as hi

ng to

n L

ib ra

ri es

o n

07 /1

0/ 17

. C op

yr ig

ht A

S C

E . F

or p

er so

na l

us e

on ly

; al

l ri

gh ts

r es

er ve

Artificial Eutrophication of Lake Washington’

W. T. EDMONDSON, G. C. ANDERSON Department of Zoology, University of Washington, Seattle

AND DONALD R. PETERSON Washington Pollution Control Commission, Olympia

ABSTRACT

Lake Washington has been receiving increasing amounts of treated sewage, and appears to be responding by changes in kind and quantity of biota. In 1933 and 1950 the dominant phytoplankton organisms were Anabaena and various diatoms and dinoflagellates, but in 1955, apparently for the first time, there was a large population of the blue-green alga, Oscillatoria rubescens, a species which makes nuisance blooms in a number of lakes. A great increase in the hypolimnetic oxygen deficit is taken as evidence of increased productivity; the deficit was 1.18 mg/cm2/month in 1933, 2.00 in 1950, and 3.13 in 1955. There is a fairly close relation between the decrease of oxygen and increase in phosphate concentration in the hypolimnion between measurements, a much less close relation with the chlorophyll

“I

concentration in the epilimnion.

Many lakes have been made productive by enrichment with domestic sewage or other drainage rich in nutrients. Such enrichment can, over a period of years, greatly modify the character of a lake, converting an oligotrophic lake to a con- dition of eutrophy, and resulting in the annual production of large populations of algae, usually dominated by the Myxophy- ceae (blue-green algae). Such populations or “blooms” are notorious nuisances, but these situations are of great interest to limnolo- gists for the insight they permit into the productive processes of lakes. Several such cases were reviewed by Hasler (1947). One of the best studied examples is Zurichsee, Switzerland, which changed in a relatively short time from an oligotrophic lake, with trout, to a eutrophic lake which

1 Some of the data discussed were obtained with the aid of the State of Washington Research Fund in Biology and Medicine (Initiative 171). We are indebted to Dr. Richard Fleming of the Depart- ment of Oceanography for permission to use un- published data, and data in technical reports obtained by the Department of Oceanography with support by contract N8onr-520/111 with the Office of Naval Research. We acknowledge with thanks the help of Dr. Francis Drouet in identify- ing algae and providing information about distri- bution, and of Mr. Rufus Kiser in giving informa- tion about the occurrence of Bosmina in Lake Washington.

produces blooms of Oscillatoria rubescens and no longer supports trout. Late in the 19th century the summer phytoplankton populations rather abruptly assumed bloom proportions, and about a decade later the cladoceran Bosmina longirostris replaced B . coregoni. Interestingly, fossil evidence shows that Linsley Pond had the same change of Bosmina species during its de- velopment at the time it was becoming eutrophic (Deevey 1942).

0. rubescens appears to be an important nuisance in the polluted lakes of Switzer- land, and it has been reported in large populations in many lakes in the United States. It was of considerable interest, therefore, to observe that 0. rubescens oc- curred in great quantity in Lake Washing- ton during the spring and summer of 1955, probably for the first time. Lake Washing- ton has been receiving treated sewage at an accelerating rate (Fig. 1A). According to figures on the relation of human population to the phosphorus content of sewage-treat- ment-plant effluent given in Sawyer’s detailed paper (1947), and the population associated with the Lake Washington effluent recorded in Fig. lA, the annual increment of phosphorus to the lake from this source in 1955 would be 37,000 kg, enough to give an average concentration of 0.0132 mg/l (0.426 pg at/l). Nitrogen

48 EDMONDSON, ANDERSON AND PETERSON

I 1933 TIME 19ko 19155

FIG. 1. A. Daily capacity of the sewage treat- ment plants emptying effluent into Lake Washing- ton, 1932-1955. Not included is the amount of untreated sewage and drainage from septic tanks. B. Oxygen deficit below 20 meters for the period 20 June-20 August each year, with the exceptions noted in text. The deficit is given on the basis of a 30-day month.

would be 5 times this on a weight basis. Although a detailed budget of sources of nutrients has not been made, it seems very likely that the observed changes in the lake can be attributed to the increased sewage. It seems possible that if enrichment con- tinues the lake may develop serious blooms of the sort experienced in so many other lakes that have similarly been enriched by urban development.

The purpose of the present paper is to describe some of the changes that have taken place since 1933, as far as they are now known. The lake has been studied a number of times. Although the lake was sampled on 9 August, 1913 (Kemmerer, Bovard and Boorman 1923), the earliest detailed work was done by Scheffer and Robinson (1939),

Scheffer (1936), and Robinson (1938), including semi-quantitative estimates of plankton populations and analyses of oxygen, phosphorus and nitrogen. Comita (1953) and Anderson (1954) obtained quantitative data on copepods, phytoplank- ton (including chlorophyll), and some chemical features. The Pollution Control Commission of the State of Washington has presented data on pollution during 1952, and the results of a widespread sampling of surface chemical conditions throughout the year 1952-1953 (Peterson et al 1952, Peterson 1955). The University of Wash- ington Department of Oceanography, in connection with a study of salt water in- trusion into the lake, obtained data on oxygen, salinity and temperature during the years 1950-1955 (Seckel and Rattray 1953, Collias and Seckel 1954, Rattray, Seckel and Barnes 1954, and unpublished). Dur- ing the current summer the present authors took data on oxygen, phosphate, temper- ature and phytoplankton. The most de- tailed biological information exists for 1933, 1950 and 1955.

The summer standing crop of phyto- plankton has increased significantly (Table 1). Except for a strong pulse of Peridinium in late August 1950, the 1950 values are consistently much smaller than for cor-

TABLE 1. Phytoplankton population volume in Lake Washington, calculated on the basis of

cell number and cell volume Multiply the values shown by 103 to get #/ml.

Weighted means are given for the period July- August. Epilimnion only.

Date Total

Phyto- Oscillatoria Oscilla- phormi- Aphanieo-

plankton rubescens toria

agardhi diunt sp. ~o~me~~ae

1960

13 May 24 June 21 July 4 Aug 21 Aug 1 Sept 15 Sept

Mean 1966

1 July 14 July 18 Aug 22 Sept

Mean

2,140 794 211 219

3,069 567 762 935

a4 2 16 8 30 4 5 2 3 0 9 13 1 105

2,895 1,407 1;755 1,314 1,725

2,783 0 1 893 0 8 397 0 610 255 0 125

0 493 727

ARTIFICIAL EUTROPHICATION OF LAKE WASHINGTON 49

responding times in the summer of 1955. On the basis of chlorophyll content and Secchi disc transparency, it may be stated that the phytoplankton population was somewhat denser on 14 June 1955 than on 1 July, but material is not available for an actual census.

The difference in plankton is indicated further by the fact that the mean summer Secchi disc transparency in 1950 was 3.5 meters (range 3.24.0) and only 2.3 (range 1.7-2.8) in 1955. In 1955 the water looked murky and had a striking, somewhat rusty color, due to the pigment in Oscillatoria rubescens that gives the species its name.

Qualitatively the plankton was rather different from one period of investigation to another. In 1933 the major components of the summer plankton included Anabaena lemmermanni and a number of diatoms. Oscillatoria sp. and Phormidium sp. were rare at all times. In 1950 the largest popu- lations were due to diatoms in the spring, and dinoflagellates in the late summer. Species of Anabaena, other than lemmer- manni, occurred but did not become abundant. Phormidium sp. had a pulse in mid-September, and Oscillatoria agardhi formed a relatively large population in February, but 0. rubescens did not occur. In 1950 the greatest relative abundance of blue-green algae occurred on 15 September when 52 % of the plankton volume was composed of Aphanocapsa and Phormidium cells. The greatest absolute quantity of blue-green algae that year occurred on I1 February when there were 311 X IO3 pa/l, averaged for the whole lake, of Oscil- latoria agardhi, amounting to 34% of the total crop. The situation in 1955 was qualitatively very different, for of the maximum counted crop, on 1 July, 96 % was composed of Oscillatoria rubescens. In 1933 the lake contained Bosmina Zongi- spina Leydig ( = B. coregoni longispina), the earlier form in the succession observed in Ziirichsee and Linsley Pond. B. Zongiros- tris was observed in the lake as early as 1940. Thus, the change of Bosmina oc- curred before the appearance of OscilZatoria rubescens in Lake Washington, reversing the sequence in Ziirichsee.

An interesting ecological problem exists in connection with the two morphologically similar species of Oscillatoria that have occurred in Lake Washington, 0. ubardhi, and 0. rubescens. The replacement of one species by another may imply a distinct, b& perhaps subtle, difference in ecological requirements. 0. agardhi has been observed to form very dense populations in a thin layer in the upper part of the hypolimnion of Hall Lake, Washington, during the summer, and to appear at the surface in moderate quantities only during the eariy fall (Anderson 1954). In Lake Washington it was abundant only during isothermal conditions, and was about twice as abundant near the bottom of the lake as at the top on the date of the maximum observed p~pu- lations. 0. rubescens has frequently been reported in abundance during the winter, although it may occur in great quantity during the summer in the hypolimnion of some lakes (e.g., Findenegg 1943, Thomas and Msrki 1949). Nevertheless, it was abundant in the surface waters of Lake Washington during the summer at t’empera- tures up to 2O”C, although the large popu- lation on 14 June occurred at a temperature of 15”. It has been shown that in Ziirichsee, 0. rubescens adjusts its level to that- at which a low light intensity exists (Thomti 1950). Apparently in some lakes this depth is in the epilimnion, in others below it. In the former case, the population is kept dis- tributed through the epilimnion by mixing.

In the absence of direct determinations ~8 photosynthetic rate and of hypolimnetic carbon dioxide accumulation, we have used the oxygen deficit as a measure of pro- ductivity (Hutchinson 1938, Ohle 1952). Originally, the deficit was considered simply as the quantity of oxygen necessary to resaturate the hypolimnion at the end of summer stratification. Obviously, the mag- nitude of the deficit will be related to the size of the hypolimnion and the duration of stratification. The deficits have, therefore, hcen expressed on an areal basis and as rates in order to make them comparable, following Hutchinson’s example. The total amount of oxygen in the hypolimnion was

50 EDMONDSON, ANDERSON AND PETERSON

m ‘0

E E”

0.02

FIG. 2. Total content and mean concentration of oxygen and phosphate phosphorus in the hypo- limnion of Lake Washington during three sum- mers. The two scales are related by the fact that the hypolimnion contains 1.407 X 101” liters. The area of the hypolimnion (20 m) is 61.58 X lOto cm2.

calculated for two dates, two months apart, and the later amount subtracted from the earlier. This difference was divided by the area of the hypolimnion and the time dif- ference to give the deficit as mg/cm2/day. To calculate the quantity of oxygen in the hypolimnion a planimetric method was used. The concentration (mg/m”) at each depth was multiplied by the area at each depth (m2), giving a quantity with the units mg/m. These values were then plotted against depth, a line fitted to the points, and the area within the curve meas- ured with a planimeter. The area gives the quantity of oxygen in grams.

The oxygen deficit, calculated for the whole summer, has been found to be, in a series of lakes, roughly proportional to the mean quantity of seston (Hutchinson 1938), and to the mean standing crop of net plank- ton in the epilimnion (Rawson 1942). While the deficit has previously been used for comparing different lakes, it will be used here for comparing different conditions of the same lake. Anderson (1954), using the oxygen data collected in 1933 by Robinson, and in 1950 by Comita, showed that the oxygen deficit below 15 meters was dis- tinctly higher in the latter year, and that the

quantity of phosphate in the hypolimnion was higher also.

For the purposes of the present paper, oxygen deficits have been calculated for all years for which data are available, using a depth of 20 meters to delimit the hypolim- nion, in order to avoid any possible effect of sun and of the thermocline, which may ex- tend as deep as 18 meters. The quantity of oxygen in the hypolimnion decreased markedly each summer, but much more rapidly in 1955 than in any of the other years (Fig. 2A). The lowest concentration of oxygen ever measured was 3.50 mg/l at 60 m on 22 September 1955, the most recent date of sampling.

The rate of decrease, or oxygen deficit, was calculated for the period 20 June- 20 August, values on those dates being ob- tained by linear graphical interpolation. The two exceptions are 1952, calculated from 18 July, and 1954, calculated between 9 August and 13 October, there being no suitable earlier determination. The period of calculation was ended on 20 August because after that date in 1950, 1951 and 1952, significant quantities of salt water entered the bottom of the hypolimnion through the ship canal. It would be difficult to make accurate allowance for the amount of oxygen carried in with the salt. The cal- culations show that while there were ir- regularities in the oxygen deficit, there has been a definite trend toward increase, and the value for 1955 is much higher than for any previous year studied (Fig. 1B).

The quantity of dissolved phosphate tended to increase in the hypolimnion of Lake Washington during each summer, especially in 1955 (Fig. 2B). The mean summer concentration has increased pro- gressively from 1933 through 1955. The maximum concentration ever observed in the hypolimnion was 0.038 mg/l of P (1.24 pg at/l), also at 60 meters, on 22 September 1955. This value is in contrast to the previous maxima of 0.022 in 1933 and 0.020 in 1950.

Some comments on the mechanism of the relationships between oxygen deficit and epilimnetic processes may be appropriate. The hypolimnetic oxygen is consumed by organisms free in the water, and on and in

ARTIFICIAL EUTROPHICATION OF LAKE WASHINGTON 51

the bottom. In a lake of the morphology of Washington, it may be expected that a large proportion of the oxygen consumption takes place in the water, and that the relationship observed depends upon some proportionality between the standing crop in the epilimnion and the amount of decomposable material that settles into the hypolimnion. In some lakes, allowance must be made for photo- synthesis in the hypolimnion, but this is inconsequential in Lake Washington be- cause of the low transparency.

The nature of the material in the epilim- nion and the processes leading to deposition of part of it into the hypolimnion require further consideration. The material consists of phytoplankton and zooplankton, healthy, moribund, and dead, as well as feces and other organic debris. Obviously, most of the dead material, tripton, is capable of settling into the hypolimnion where it can support bacteria and other organisms which consume oxygen. But even some living phytoplank- ters can be expected to settle out and consume oxygen, at first through their own respiration, and later as substrate for bac- teria. Many of the zooplankton which spend the day in the deep water presumably migrate to the epilimnion and feed there at night. Thus, some of the respiration of healthy hypolimnetic planktonic animals represents use of material produced in the epilimnion. A very important process, leading to sedimentation of particulate materials into the hypolimnion, is the grazing activity of the zooplankton, through which organisms are removed from the water, and the partly digested remnants dropped as feces. It has been shown in a marine population that there is a fairly close proportionality between the density of the phytoplankton population and the abun- dance of copepod feces in the water, suggest- ing that the animals tended to cram ma- terials through the gut as fast as they could collect it (Harvey, Cooper, Lebour and Russell 1935). There is no reason to sup- pose that similar freshwater copepods be- have differently.

Therefore, under ordinary circumstances, the animals can be expected to collect more food and drop more feces per unit time when phytoplankton is abundant than dur-

ing periods of scarcity. The larger the crop of zooplankton, the greater the total feeding rate will be. Also, large plankton crops will produce corpses faster by reason of senes- cence and parasitism than will small ones. Therefore, lakes which develop relatively large populations of organisms of almost any kind in the epilimnion might be expected to have relatively large hypolimnetic deficits.

The low correlation actually observed by Hutchinson and Rawson between standing crop and oxygen deficit is due in part to the biological and chemical diversity of material going into the hypolimnion. The real re- lationship, however, must be with the primary productivity of the epilimnetic population, and this must, in the end, be more important than the population size or composition itself. If the phytoplankton population is reproducing slowly, relative to its rate of removal, then the standing crop will decline, and the average size will be small. On the other hand, a rapidly re- producing phytoplankton population could be kept grazed down by an active zoo- plankton population for much of the summer (see, for example, Anderson, Comita and Engstrom-Heg 1955). Naturally, a long- continued high phytoplankton production can ordinarily be expected to give rise in time either to a large phytoplankton popu- lation or to a large zooplankton population, but the fact that increased phytoplankton may result in increased transport to the hypolimnion, out of proportion to the assimilation by the zooplankton, means that the oxygen deficit will probably be more closely related to productivity than to standing crop. The looseness of the rela- tionship obtained by Rawson and by Hutchinson may measure the low degree to which mean standing crop is an indication of productivity in the particular lakes in- volved.

One might also expect to find relation- ships between epilimnetic and hypolimnetic events during different short periods in one summer, although the nature of the re- lationships would be affected by the rate at which the various materials settle out, and the rate at which they are decomposed (Kleerekoper 1953). Accordingly, short term deficits were calculated for the period be-

52 EDMONDSON, ANDERSON AND PETERSON

00,’ J 0 I

-0.5- 6 9 III1 1 II I 11 I II I

0 0.05 0.10 0.15 OXYGEN DEFICIT

mcpn./cm.*/day

FIG. 3. Relation of oxygen deficit, calculated for short periods, to chlorophyll concentration in the epilimnion at the beginning of the period, and to the rate of change of phosphate phosphorus in the hypolimnion during the same periods.

tween measurements of oxygen in the years in which the most data were taken. This is not the place for a discussion of the ecological significance of chlorophyll in plankton populations beyond, pointing out that a case can be made for the use of chlorophyll as a measure of potential primary productivity (Manning and Juday 1941, Edmondson 1955), and that it is reasonable to expect a positive relation between epilimnetic chlo- rophyll and the oxygen deficit. There was indeed a tendency for the largest decreases in hypolimnetic oxygen to take place during periods when the initial concentration of chlorophyll was large, but the relation is not strong (Fig. 3A). The fact that the cor- relation is this low calls for further in- vestigation.

Although heterotrophic bacteria are ca- pable of absorbing phosphate, as well as causing it to be released from phosphorus-

containing substrates, phosphate liberation in the hypolimnion is related in some degree to the processes that lead to removal of oxygen. In order to establish the relation- ship, the rate of change of phosphate content of the hypolimnion was calculated in the same way as the oxygen deficit (Fig. 3B). There is a distinct tendency for the larger rates of increase of phosphate to occur with high oxygen deficits, and with but one exception, decreases in phosphate are ac- companied by slow decreases of oxygen. The exceptional point is for the period 15 September-7 October 1950, at a time when there had been a relatively large intrusion of salt water into the hypolimnion, but the large oxygen deficit cannot be attributed solely to the oxygen content of this water. The slope of the upper part of the curve, associated with increases in phosphate, shows that in the hypolimnion of Lake Washington 1 atom of phosphorus is liber- ated as phosphate for every 16.4 atoms of oxygen removed. This is a very much lower O/P ratio then has been observed in the regeneration of phosphorus in the open sea (Redfield 1942), and in the converse release of oxygen by phytoplankton in photo- synthesis (Edmondson and Edmondson 1947, Edmondson 1955). It seems to indicate an effective regeneration of phosphate, relative to carbon, even allowing for probable amounts of anaerobic activity in the bottom, and may be a result of the nu- tritive conditions permitting the algae to develop unusually high phosphorus con- tents. It would be very interesting to know what the relation is in a large series of lakes with very different deficits and nutrient supplies.

In summary, Lake Washington shows definite evidence of having rather suddenly increased in productivity, the oxygen deficit in 1955 being 1.8 times that in 1952, the previous maximum, and 2.7 times the rate in 1933. The biological character of the lake has recently changed in that the former dominance of diatoms and dinoflagellates in the population has been replaced by that of the blue-green alga, Oscillatoria rubescens, a species that forms nuisance blooms in a number of European and American lakes. The most reasonable explanation of the

ARTIFICIAL EUTROPHICATIO

increase in productivity is the great increase in treated sewage added to the lake with the growth of adjoining communities. Lake Washington seems to be fitting the pattern of abrupt change, as seen in the other cases of polluted lakes which have been studied limnologically before pollution became seri- ous. It is hoped that it will be possible to study Lake Washington further as its eutrophication proceeds or, if the effluents are diverted, to see to what extent the lake regains its former more oligotrophic con- dition.

REFERENCES

ANDERSON, G. C. 1954. A limnological study of the seasonal variations of phytoplankton popula- tions. Ph.D. Thesis. Univ. Washington. 268

ANDERSON, G. C., G. W. COMITA, AND V. ENG- STROM-HEG. 1955. A note on the phytoplank- ton-zooplankton relationships in two lakes in Washington. Ecology, 36: 757-759.

COLLIAS, E. E., AND G. R. SECKEL. 1954. Lake Washington Ship Canal data. Univ. Washing- ton Dept. Oceanogr. Sp. Rep. No. 2. iv + l-27.

COMITA, G. W. 1953. A limnological study of planktonic copepod populations. Ph.D. The- sis. Univ. Washington. 195 pp.

DEEVEY, E. S., JR. 1942. Studies on Connecticut lake sediments. III. The biostratonomy of Linsley Pond. Amer. Jour. Sci., 240: 233-264, 313-338 *

EDMONDSON, W. T. 1955. Factors affecting produc- tivity in fertilized salt water. Deep Sea Re- search (in press).

EDMONDSON, W. T., AND Y. H. EDMONDSON. 1947. Measurements of production in fertilized salt water. J. Mar. Res., 6: 228-246.

FINDENEGG, I. 1943. Untersuchungen iiber die Oekologie und die Produktionsverhaltnisse des Planktons im Kiirtner Seengebiete. Int. Rev. Hydrobiol., 43: 368-429.

HARVEY, H. W., L. H. N. COOPER, M. V. LEBOUR, AND F. S. RUSSELL. 1935. Plankton production and its control. J. Mar. Biol. Assoc. U. K., 20: 407-442.

HASLER, A. D. 1947. Eutrophication of lakes by domestic drainage. Ecology, 28: 383-395.

HUTCHINSON, G. E. 1938. On the relation between the oxygen deficit and the productivity and typology of lakes. Int. Rev. Hydrobiol., 36: 336-355.

N OF LAKE WASHINGTON 53

KEMMIIZRER, G., J. F. BOVARD, AND W. R. BOOR- MAN. 1923. Northwestern lakes of the United States: biological and chemical studies with reference to possibilities in production of fish. Bull. U. S. Bur. Fish., 47: 407-437.

KLEEREKOPER, H. 1953. The mineralization of plankton. J. Fish. Res. Bd. Canada, 10: 283- 291.

MANNING, W. M., AND R. E. JUDAY. 1941. The chlorophyll content and productivity of some lakes in Northeastern Wisconsin. Trans. Wis- consin Acad. Sci . Arts & Let ., 33: 363-393.

OEILE, W. 1952. Die hypolimnische Kohlendioxyd- Akkumulation als produktionsbiologischer In- dikator. Arch. Hydrobiol., 46: 153-285.

PETERSON, D. R. 1955. An investigation of pollu- tional effects in Lake Washington. Washington Pollution Control Commission. Tech. Bull. No. 18.18 pp. + xiv.

PETERSON, D. R., K. R. JONES, AND G. T. ORLOB. 1952. An investigation of pollution in Lake Washington. Washington Pollution Control Commission. Tech. Bull. No. 14.29 pp.

RATTRAY, M. JR., G. R. SECKEL, AND C. A. BARNES. 1954. Salt budget in the Lake Washington Ship Canal system. J. Mar. Res., 13: 263-275.

RAWSON, D. S. 1942. A comparison of some large alpine lakes in western Canada. Ecology, 23; 143-161.

REDFIELD, A. C. 1942. The processes determining the concentration of oxygen, phosphate and other organic derivatives within the depths of the Atlantic Ocean. Papers in Phys. Oceanogr. and Meteorol., 9 (2) : l-22.

ROBINSON, R. J. 1938. Chemical data for Lake Washington. Typewritten report. Univ. Wash- ington Library,

SAWYER, C. N. 1947. Fertilization of lakes by agricultural and urban drainage. J. New England Waterworks Assoc., 61: 109-127.

SCIIEFFER, V. B. .1936. The plankton of Lake Washington. Ph.D. Thesis. Univ. Washington Library.

SCIIEFFER, V. B., AND R. J. ROBINSON. 1939. A limnological study of Lake Washington. Ecol. Monogr., 9: 95-143.

SECKEL, G. R., AND M. RATTRAY. 1953. Studies on Lake Washington Ship Canal. Univ. Washing- ton Dept. Oceanogr. Tech. Rep. No. 15. viii + l-101.

THOMAS, E. A. 1950. Auffiillige biologische Folgen von Sprungschichtneigungen im Zurichsee. Schweiz. Zeits. fur Hydrol., 12: l-23.

THOMAS, E. A., AND E. MXRKI. 1949. Der heutige Zustand des Ziirichsees. Verh. Internat. Ver. Limnol., 10: 476-488.

Phosphorus, Nitrogen, and Algae in Lake Washington after Diversion of Sewage Author(s): W. T. Edmondson Source: Science, New Series, Vol. 169, No. 3946 (Aug. 14, 1970), pp. 690-691 Published by: American Association for the Advancement of Science Stable URL: http://www.jstor.org/stable/1729288 Accessed: 05-07-2017 22:59 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted

digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about

JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

http://about.jstor.org/terms

American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and extend access to Science

This content downloaded from 128.95.104.66 on Wed, 05 Jul 2017 22:59:51 UTC All use subject to http://about.jstor.org/terms

Eutrophication and Recovery in Experimental Lakes: Implications for Lake Management Author(s): D. W. Schindler Source: Science, New Series, Vol. 184, No. 4139 (May 24, 1974), pp. 897-899 Published by: American Association for the Advancement of Science Stable URL: https://www.jstor.org/stable/1738185 Accessed: 20-09-2018 21:12 UTC

REFERENCES Linked references are available on JSTOR for this article: https://www.jstor.org/stable/1738185?seq=1&cid=pdf-reference#references_tab_contents You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide

range of content in a trusted digital archive. We use information technology and tools to increase productivity and

facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

https://about.jstor.org/terms

American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and extend access to Science

This content downloaded from 205.175.107.196 on Thu, 20 Sep 2018 21:12:02 UTC All use subject to https://about.jstor.org/terms

tion spectrophotometer equipped with a flame- less mercury cell. Elemental mercury was ob- served. Also, a liquid trap of cysteine was placed in the flow system immediately before the acid-permanganate trap. This trap was counted along with the acid-permanganate trap and showed no accumulated activity during the course of the experiment. Since the cysteine trap would remove mercury and methylmercuric ions but not elemental mercury from the nitrogen flow, the mercury species trapped by the oxidizing acid-permanganate trap must have been elemental mercury. Further direct measurements of evolved Hgo with the atomic absorption spectrophotometer would have required excessively large con- centrations of mercury, and the rate of release was so slow that this method of analysis was discontinued.

Although the U.S.-Canada Water Quality Agreement (1) was signed on 15 April 1972, legislation prohibiting the use of phosphorus in detergents and controlling inputs of phosphorus to the St. Lawrence Great Lakes has

not been passed by many states (2). Much of the foot-dragging on anti- eutrophication laws undoubtedly still results from the controversy and con- fusion surrounding the debate over the effectiveness of controlling phosphorus in influents to freshwater lakes (3). Among the main points debated (often on the basis of inconclusive evidence) have been:

1) Is phosphorus really responsible for eutrophication problems?

2) If sufficient phosphorus is avail- able, can carbon limit the growth of undesirable algae?

3) Is phosphorus removal alone an effective means of overcoming eutrophi- cation problems?

4) Are already culturally eutrophied lakes recoverable? Can this be done by controlling inputs of phosphorus alone?

5) What concentration of phosphorus can be considered safe?

Answers to these questions have been sought in a series of whole-lake experi- ments conducted in the Experimental Lakes Area of northwestern Ontario. Lakes in the area are set in Precam-

24 MAY 1974

1) Is phosphorus really responsible for eutrophication problems?

2) If sufficient phosphorus is avail- able, can carbon limit the growth of undesirable algae?

3) Is phosphorus removal alone an effective means of overcoming eutrophi- cation problems?

4) Are already culturally eutrophied lakes recoverable? Can this be done by controlling inputs of phosphorus alone?

5) What concentration of phosphorus can be considered safe?

Answers to these questions have been sought in a series of whole-lake experi- ments conducted in the Experimental Lakes Area of northwestern Ontario. Lakes in the area are set in Precam-

24 MAY 1974

8. M. Schnitzer and S. V. Khan, Hiumic Suib- stances in the Environment (Dekker, New York, 1972).

9. N. M. Atherton, P. A. Cranwell, A. J. Floyd, R. D. Haworth, Tetrahedron 23, 1653 (1967); R. Riffaldi and M. Schnitzer, Geoderma 8, 1 (1972); M. V. Cheshire and P. A. Cizanwell, J. Soil Sci. 23, 424 (1972); B. R. Nagar, N. P. Datta, M. R. Das, M. P. Khakhar, Indian J. Chen. 5, 587 (1967).

10. B. R. Nagar, A. Chanorasekhara Rao, N. P. Datta, Indian J. Chem. 9, 168 (1971).

11. Supported in part by funds from the Environ- mental Protection Agency under grant R- 800427. We thank P. W. Carr, E. G. Janzen, and L. R. Pomeroy for thoughtful discussions of this work.

10 December 1973; revised 21 January 1974 *

8. M. Schnitzer and S. V. Khan, Hiumic Suib- stances in the Environment (Dekker, New York, 1972).

10. B. R. Nagar, A. Chanorasekhara Rao, N. P. Datta, Indian J. Chem. 9, 168 (1971).

11. Supported in part by funds from the Environ- mental Protection Agency under grant R- 800427. We thank P. W. Carr, E. G. Janzen, and L. R. Pomeroy for thoughtful discussions of this work.

10 December 1973; revised 21 January 1974 *

brian Shield bedrock. Chemically and biologically they are similar to more than 50 percent of the waters draining to the St. Lawrence Great Lakes (4).

In an early experiment, phosphate and nitrate were added to lake 227,

brian Shield bedrock. Chemically and biologically they are similar to more than 50 percent of the waters draining to the St. Lawrence Great Lakes (4).

In an early experiment, phosphate and nitrate were added to lake 227,

Fig. 1. Lake 226, demonstrating the vital role of phosphorus in eutrophication. The far basin, fertilized with phosphorus, ni- trogen, and carbon, was covered by an algal bloom within 2 months. No increases in algae or species changes were observed in the near basin, which received similar quantities of nitrogen and carbon but no phosphorus.

which has an extremely low content of dissolved inorganic carbon, to see whether shortage of carbon would pre- vent the eutrophication of such a lake (5). The lake was transformed into a teeming, green soup within weeks after nutrient additions were begun. Algal standing crops up to two orders of magnitude greater than those in un- fertilized lakes of the area have been

observed (6, 7). No increase in phos- phate concentration was observed, and any added phosphate disappeared in minutes because of uptake by plankton (8). Gas-exchange studies revealed that some of the additional carbon required for production of this algal bloom was drawn from the atmosphere, and a comparison of dissolved inorganic car- bon concentrations and parameters af- fecting gas exchange indicated that there was no possibility that shortage of carbon could prevent the eutrophica- tion of the St. Lawrence Great Lakes or any other water body of economic importance (9).

Experiments conduLcted in smaller enclosures (2 to 3 m:;) in the same lake revealed that if phosphorus was not supplied, algal blooms did not oc- cur (10). In order to test the validity of this conclusion on a whole lake, an experiment was begun in 1973 in an- other small lake, 226. This lake, which has two similar basins separated by a shallow neck (see Fig. 1), was divided into two equal areas by using a sea cutrtain (60 by 6 m) of vinyl reinforced with nylon (Kepner Plastics, Torrance, California), which was sealed into the sediments and fastened to the bedrock in the narrow section of the lake. Be- ginning in late May 1973, additions of nitrogen and carbon were made equally to both basins, but phosphorus was added only to the northeast basin of the lake (1/).

The photograph in Fig. 1 was taken on 4 September 1973, when a bloom of the blue-green alga Anabaena spi- roides covered that basin receiving phosphorus. Throughout the year, phy- toplankton species and standing crops in the basin that received only nitrogen and carbon remained similar to those before fertilization was begun, con- sisting chiefly of Tabellaria fenestrata, Synedra acus, and other diatoms. The results indicate the efficacy to be ex- pected from controlling phosphorus

as a means of preventing eutrophica- tion.

A common belief is that phosphate, returned from anoxic sediments in

897

content of the influents to such waters

as a means of preventing eutrophica- tion.

A common belief is that phosphate, returned from anoxic sediments in

897

Eutrophication and Recovery in Experimental Lakes:

Implications for Lake Management

Abstract. Combinations of phosphorus, nitrogen, and carbon were added to several small lakes in northwestern Ontario, Canada, at rates similar to those in many culturally eutrophied lakes. Phosphate and nitrate caused rapid eutrophi- cation. A similar result was obtained with phosphate, ammonia, and sucrose, but recovery was almost immediate when phosphate additions only were discontinued. When two basins of one lake were fertilized with equal amounts of nitrate and sucrose, and phosphorus was also added to one of the basins, the phosphate- enriched basin quickly becamie highly eutrophic, while the basin receiving only nitrogen and carbon remained at prefertilization conditions. These results, and the high affinity of sediments for phosphorus indicate that rapid abatement of eutrophication may be expected to follow phosphorus control measures.

Eutrophication and Recovery in Experimental Lakes:

Implications for Lake Management

This content downloaded from 205.175.107.196 on Thu, 20 Sep 2018 21:12:02 UTC All use subject to https://about.jstor.org/terms

eutrophied lakes, would cause such lakes to retain their eutrophic charac- ter even if external sources of phos- phorus were eliminated (12). Our re- sults in 1 year from the fertilization of lake 227 showed that little or no phos- phorus was returned from the sedi- ments of that lake, even with anoxic periods of several months (13). Other studies have suggested that phosphorus return from sediments would not seri-

ously delay the recovery of a lake from cultural eutrophication once major phosphorus sources were eradicated (14). A whole-lake experiment was therefore designed to test the speed of lake recovery and the efficiency of the sediments at removing and retain- ing phosphorus.

The phytoplankton and chemistry of lake 304 in its natural state were

110

100

studied in 1969 and 1970 (4). In 1971 and 1972, phosphorus, nitrogen, and carbon were added to the lake (15). As in lakes 227 and. 226, an algal bloom occurred in response to this application of fertilizer. In 1973, we continued to add nitrogen and carbon, but discontinued phosphorus additions, simulating conditions that might exist in a culturally eutrophied lake after phosphorus control measures were taken. The recovery of the lake was nearly complete, as the chlorophyll a concentrations indicate (Fig. 2). These results can be explained by our experi- ments in lake 227, which have shown that phosphate in the hypolimnion is taken up rapidly by microplankton (probably bacteria), then sedimented to the lake bottom, where it remains, re- gardless of oxygen concentration (16).

Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec

Fig. 2. Chlorophyll a concentrations in lake 304. In 1968, 1969, and 1970 the lake was not fertilized. In 1971 and 1972, it received annual additions of 0.40 g of phos- phorus, 5.2 g of nitrogen, and 5.5 g of carbon per square meter. In 1973, additions of nitrogen and carbon were continued at the same rate but phosphorus was not added.

898

Taken together, these results provide answers to the questions listed above. They suggest that cultural eutrophica- tion problems might be solved simply and swiftly. In most lakes, reducing the phosphorus input could be expected to cause a proportional abatement in phytoplankton blooms and other symp- toms of eutrophication (17). Fully 50 percent of the phosphorus coming into the St. Lawrence Great Lakes could

be eliminated by simply banning or greatly reducing detergent phosphates, a step already taken in Canada and a few U.S. states (2). Most of the U.S. states, however, plan to remove phos- phorus at the sewage treatment plant (18). While fine in principle, this scheme will take several years to imple- ment to any effective degree, consider- ing the time lags and uncertainty in- evitable in financing, planning, and constructing such facilities. Numerous small sources, such as small communi- ties and individual homes on septic systems, will escape phosphorus re- moval for some time longer. It appears that a basin-wide ban on detergent phosphates would quickly bring about a partial recovery of Lakes Erie and Ontario, perhaps as much as a decade before full-scale phosphorus control by other means is possible. Such a re- covery would provide a savings of many millions of dollars, as well as restoring to some degree the beauty of these enormous resources.

D. W. SCHINDLER

Fisheries and Marine Service, Freshwater Institute, Winnipeg, Manitoba, Canada R3T 2N6

References and Notes

1. Great Lakes Water Quiality Agreemient with Annexes and Texts and Terms of Reference between the United States and Canada (Treat- ies and Other International Acts, Series 7312, Government Printing Office, Washington, D.C., 1972).

2. Canada now prohibits the sale of detergents containing more than 2.2 percent phosphorus. This legislation became effective on 1 Janu- ary 1973. In the Great Lakes Basin of the United States, only Indiana, New York, and Michigan have passed such legislation. Minne- sota, Michigan, and Wisconsin plan to re- move phosphorus at sewage treatment plants after 1 January 1975, but only from municipal plants serving populations of more than 2000 to 2500 (18). Federal action will probably be required if an effective phosphorus control plan is to be implemented quickly.

3. For examples of the debate, see R. F. Legge and D. Dingeldein, Can. Res. Dev. 3 (No. 1), 19 (1970); J. R. Vallentyne, ibid. 3 (No. 3), 36 (1970); G. E. Likens, Ed., Nutrients and Eutrophication (Allen, Lawrence, Kan., 1972).

4. Issue No. 2 of J. Fish. Res. Board Calz. 28 (1971) was on the Experimental Lakes Area project; also see H. Kling and S. K. Holm- gren, Fish. Res. Board Can. Tech. Rep. No. 337 (1972).

5. Additions equivalent to 0.5 g of PO,-P and 6.9 g of NO.,-N per square meter of lake surface per year were made over a 21-week period. This is five to ten times the input expected from natural sources and simulates

SCIENCE, VOL. 184

This content downloaded from 205.175.107.196 on Thu, 20 Sep 2018 21:12:02 UTC All use subject to https://about.jstor.org/terms

the degree of nutrient enrichment in the Great Lakes [R. A. Vollenweider, Technical Report DAS/CSI/68.27 (Organisation for Economic Cooperation and Development, Paris, 1968)]. Before fertilization, the concentration of dis- solved inorganic carbon in the epilimnion of lake 227 was about 50 /umole/liter in mid- summer. This is only about 2 to 3 percent of that found in the lower Great Lakes.

6. Most experts consider a lake to be eutrophic when algal blooms with more than 30 aLg of chlorophyll a per liter become common. Un- fertilized lakes in the Experimental Lakes Area support a midsummer chlorophyll a concentration of only 1 to 5 A/g/liter. In lake 227, after fertilization, blooms with more than 100 Aug/liter have been colmmon, and up to 300 jug/liter has been observed for short periods (7).

7. D. W. Schindler, F. A. J. Armstrong, S. K. Holmgren, G. J. Brunskill, J. Fish. Res. Board Can. 28, 1763 (1971); D. W. Schindler, H. Kling, R. V. Schmidt, J. Prokopowich, V. E. Frost, R. A. Reid, M. Capel, ibid. 30, 1415 (1973).

8. Because of the rapid turnover of phosphorus in many freshwater lakes lF. H. Rigler, Lirnnol. Oceanogr. 9, 511 (1964)] and uncertain methods of measurment [F. H. Rigler, Verh. Int. Ver. Limnol. 16, 465 (1966)] phosphate concentration appears to be unreliable as a widespread indicator of whether a lake will develop algal blooms (7).

9. D. W. Schindler, G. J. Brunskill, S. Emerson, W. S. Broecker, T.-H. Peng, Science 177, 1192 (1972); S. Emerson, W. S. Broecker, D. W. Schindler, J. Fish. Res. Board Can. 30, 1475 (1973); W. S. Broecker, unpublished results.

10. D. W. Schindler (unpublished data) has found that if phosphorus is not supplied, no algal increases occur. If sufficient phosphorus is supplied, algal increases do occur, the magni- tude being determined by available nitrogen, which in some cases may be fixed from the atmosphere. When both phosphorus and ni- trogen are supplied in excess, algae increase until light becomes limiting.

11. Additions equivalent to 3.16 g of NO.a-N and 6.05 g of sucrose C per square meter per year were made to both basins, in 20 equal weekly increments. The northeast basin also received 0.59 g m- 'year-1 of PO,-P. The N/P and C/P ratios are greater than in sewage, while the quantity of P added is not exceptionally high for a culturally affected lake.

12. Sedimentary phosphate has been thought to dissolve under anoxic conditions by reducing ferric iron and forming Fe(OH).,-POt com- plexes. C. H. Mortimer [J. Ecol. 29, 280 (1941); ibid. 30, 147 (1942)] and G. E. Hutch- inson [A Treatise on Limnology (Wiley, New York. 1957), vol. 1, chap. 12] explain this cycle.

13. Our preliminary results suggest that dissolu- tion of iron-phosphate complexes followed by release of phosphate to the water column is prevented by the high demand of sediment bacteria for phosphate, and formation of humic acid-phosphate complexes.

14. Rapid recoveries were observed in Lake Washington in the state of Washington [W. T. Edmondson, Science 169, 690 (1970)]; Zellersee, Germany [R. Liepolt, Verhandlungen Symlposion iiber Gewassereutrophierung (Salz- burg, 1967)]; Pedersborg S0 and Lyngby S0, Denmark [H. Mathiesen, Mitt. Int. Verein. Limnol. 19, 161 (1971)]; and Little Otter Lake and Gravenhurst Bay, Ontario, Canada [M. F. P. Michalski, Ontario Ministry of Environ- ment, personal communication; M. F. P. Mi- chalski and N. Conroy, Proceedings 16th Con- ference on Great Lakes Research (1973), p. 934].

15. Additions equivalent to 0.40 g of PO,-P, 5.2 g of NH:-N, and 5.5 g of sucrose C per square meter of lake surface per year were made in 20 weekly increments.

16. D. W. Schindler, unpublished results. 17. This would hold true only for lakes that are

phosphorus-limited. Some waterways appear to have been so overwhelmed with phosphorus that it no longer limits phytoplankton produc- tion, and inputs of the element must be re-

15. Additions equivalent to 0.40 g of PO,-P, 5.2 g of NH:-N, and 5.5 g of sucrose C per square meter of lake surface per year were made in 20 weekly increments.

16. D. W. Schindler, unpublished results. 17. This would hold true only for lakes that are

phosphorus-limited. Some waterways appear to have been so overwhelmed with phosphorus that it no longer limits phytoplankton produc- tion, and inputs of the element must be re- duced enough for it to limit phytoplankton growth before abatement would be propor- tional to phosphorus removal [I. Ahlgren, Verh. Int. Ver. Limnol. 18, 355 (1972)].

18. Report to the International Joint Commission from the International Great Lakes Water Quality Board, March 1973.

21 December 1973

24 MAY 1974

duced enough for it to limit phytoplankton growth before abatement would be propor- tional to phosphorus removal [I. Ahlgren, Verh. Int. Ver. Limnol. 18, 355 (1972)].

18. Report to the International Joint Commission from the International Great Lakes Water Quality Board, March 1973.

21 December 1973

24 MAY 1974

Opaline Sediments of the Southeastern Coastal Plain

and Horizon A: Biogenic Origin

Abstract. Scanning electron microscope techniques show that Eocene opaline claystones (fuller's earth and buhrstone) of the Atlantic and Gulf Coastal Plain, deposits long considered volcanic in origin, are actually highly altered diatomites formed as transgressive facies in normal marine continental shelf environments. These findings are in agreement with a biogenic origin for time-equivalent horizon A and A" deep-sea cherts of the North Atlantic and Caribbean.

Opaline Sediments of the Southeastern Coastal Plain

and Horizon A: Biogenic Origin

Opaline (cristobalite-rich) Eocene claystone deposits of the Atlantic and Gulf Coastal Plain have recently been cited in Science (1) and elsewhere (2) as examples of altered rhyolitic ashes which accumulated in nearshore or

brackish coastal environments. Such

ashes are also thought to have been distributed by atmospheric and water currents into the North Atlantic Ocean

basin where they were presumably re- sponsible for the formation of the cristobalite-rich, horizon A Eocene cherts (1, 3). We present evidence here to show that opaline claystones of the coastal plain are altered diatomites, not ashes, and that they formed in normal marine rather than in restricted coastal

environments. Our evidence is compati- ble with a biogenic rather than a vol-

canic origin for the horizon A cherts and their Caribbean equivalents (hori- zon A").

Opaline claystones are unusually porous, lightweight siliceous rocks which possess oil clarification proper- ties (4). Accordingly, they have been referred to locally as fuller's earth (5)

brackish coastal environments. Such

ashes are also thought to have been distributed by atmospheric and water currents into the North Atlantic Ocean

environments. Our evidence is compati- ble with a biogenic rather than a vol-

canic origin for the horizon A cherts and their Caribbean equivalents (hori- zon A").

Opaline claystones are unusually porous, lightweight siliceous rocks which possess oil clarification proper- ties (4). Accordingly, they have been referred to locally as fuller's earth (5)

or buhrstone (6). Scanning electron microscopy of fracture surfaces of opaline claystones from 14 Southeast- ern Coastal Plain localities (Mississippi to South Carolina; see Table 1) reveals siliceous microfossils which occur as

molds in 90 percent of the samples examined. The fossils are most abun-

dant in samples which contain 60 to 90 percent SiO.,. The opaline material is unidimensionally disordered alpha- cristobalite (7) in the form of bladed microspherulites (8). Most of the microfossil molds are of marine di-

atoms including large and small cen- trics (Fig. 1A), pennates and forms which resemble Triceratium (Fig. 1B), and Actinoptychus (Fig. 1C). Sponge spicule (Fig. D) and radiolarian molds (9) are interspersed in the South Carolina and Alabama material.

Clearly, the opaline claystones repre- sent highly altered diatomite deposits rather than ash beds. Most microfos-

sils in the deposits, however, have been completely destroyed by dissolution.

Siliceous microfossils have not been

reported previously in South Carolina

molds in 90 percent of the samples examined. The fossils are most abun-

Clearly, the opaline claystones repre- sent highly altered diatomite deposits rather than ash beds. Most microfos-

sils in the deposits, however, have been completely destroyed by dissolution.

Siliceous microfossils have not been

reported previously in South Carolina

Table 1. Opaline claystone samples which contain siliceous microfossil molds. The Black Mingo and McBean units were collected by S. D. Heron. All other samples were collected by the authors.

Formation Samples and localities Age

Nanafalia

(Grampian Hills member) GH-l (Wilcox County, Ala.) Late Paleocene

Black Mingo 9-6-1 (Sandy Run Creek, S.C.); Early-Middle (opaline facies) 9-9-1 (Big Beaver Creek, S.C.); Eocene

6-10-4, 9-11-4 (Little Beaver Creek, S.C.); 9-18-1, 9-18-2 (Bates Mill Creek, S.C.); 9-67-1, 9-67-2, 9-67-3 (Thelma Hill property, Calhoun County, S.C.); 9-68-2 (Dicks Swamp, S.C.); A-183-1 (Williamsburg Bridge, S.C.); 43-6-1, (Tavern Creek, S.C.); 43-7-5 (Holy Cross Church, Sumter County, S.C.)

McBean A-3-1, A-3-2 (Early Branch, S.C.) Middle Eocene Tallahatta

T-3 [Choctaw County, Ala.; locality 135 of Toulmin and LaMoreaux (17)]; 1-10-12, 1-10-13, 33-1 (U.S. Highway 1-10, Meridian, Miss.)

Barnwell KL-1 (Georgia-Tennessee Clay Late Eocene (Twiggs Clay member) Corporation pit, Wrens, Ga.)

899

Table 1. Opaline claystone samples which contain siliceous microfossil molds. The Black Mingo and McBean units were collected by S. D. Heron. All other samples were collected by the authors.

Formation Samples and localities Age

Nanafalia

(Grampian Hills member) GH-l (Wilcox County, Ala.) Late Paleocene

Black Mingo 9-6-1 (Sandy Run Creek, S.C.); Early-Middle (opaline facies) 9-9-1 (Big Beaver Creek, S.C.); Eocene

McBean A-3-1, A-3-2 (Early Branch, S.C.) Middle Eocene Tallahatta

T-3 [Choctaw County, Ala.; locality 135 of Toulmin and LaMoreaux (17)]; 1-10-12, 1-10-13, 33-1 (U.S. Highway 1-10, Meridian, Miss.)

Barnwell KL-1 (Georgia-Tennessee Clay Late Eocene (Twiggs Clay member) Corporation pit, Wrens, Ga.)

899

This content downloaded from 205.175.107.196 on Thu, 20 Sep 2018 21:12:02 UTC All use subject to https://about.jstor.org/terms

LIMNOLOGY AND

OCEANOGRAPHY

January 1981

Volume 26

Number 1

The effect of changes in the nutrient income on the condition of Lake Washington’

W. T. Edmondson and John T. Lehman’ Department of Zoology NJ-15, University of Washington, Seattle 98195

Abstract

Lake Washington received increasing amounts of secondary sewage effluent from 1941 to 1963 and responded by changes in the amount of nutrients in the water and in the kind and quantity of phytoplankton. From 1963 to 1968 the amount of effluent entering the lake was progressively decreased to zero, and the lake promptly responded with decreases in the amount of nutrients, the quantity of phytoplankton, and the proportion of blue-green algae. The lake could be regarded as having recovered from eutrophication by 1975.

Phosphorus and nitrogen income from sewage, inlets, and atmosphere were calculated from direct measurements or approximated from other data and from regressions with hydrological income. Total phosphorus input varied from a maximum of 204.2 x 10” kg.yr’ in 1964 to a low of 42.9 in 1973 and 1976. The maximum fraction from sewage was 72.4% in 1962. Total nitrogen varied from 1,419 x 103 kg. yr-’ in 1964 to 734 in 1976. The total P content of the lake varied in close correlation with input from a maximum of about 200 x lo3 kg (equivalent to a mean concentration of about 69 pg.liter’) to a mean of about 50 x lo3 kg during the postrecovery era. The sewage episode affected the phosphorus regime much more than that of nitrogen.

There were seasonal differences in the deposition of phosphorus to and release from the sediments, the average net long term retention being about 57% of the income; slight changes in retention during and after diversion resulted in a greater fraction of the influent P being lost to the sediments after diversion. The amount of phosphorus lost permanently to the sediments during a year is more closely related to the annual income than to the mean concentration in the water.

The condition of Lake Washington has changed greatly in response to variations in nutrient income brought about by al-

’ Most of the financial support for this work has been provided by the National Science Foundation in a series of grants starting in 1958. The National Institutes of Health supported the work for 2 years before that, and the Environmental Protection Agency provided additional support in 1973-1976. Earlier work was aided by the State of Washington Fund for Research in Biology and Medicine (Initi- ative 171).

J Present address: Division of Biological Sci- ences, Natural Science Building, University of Michigan, Ann Arbor 48109.

terations of sewerage arrangements. An early episode of pollution with raw and treated sewage ended in the mid-1930s with the construction of a major sewage system that diverted Seattle’s effluent from the lake to Puget Sound. A later ep- isode of eutrophication with secondary effluent started as the population spread north and south of the city along the lake, as smaller towns around the lake grew, and as secondary sewage treatment plants were built (Edmondson 1972). A diversion system then reduced the amount of effluent from a maximum of

2 Edmondson and Lehman

about 75,700 m3 * d-1 in 1963 to zero in 1968. The only sewage to reach the lake since 1968 has been a relatively small amount in overflow from Seattle’s com- bined sewer system.

This situation has been treated as an experiment in lake fertilization that can improve our knowledge of the role of nu- tricnt income in controlling the biologi- cal character of lakes (Edmondson 1977a). By 1975 the lake appeared to have fin- ished responding to the diversion of sew- age and the experiment seemed to be over. In this paper we give an account of changes in the income of phosphorus and nitrogen and the response of the lake in terms of its content of those elements and the general abundance of the phyto- plankton.

Field and laboratory work has been done by too many associates to list indi- vidually. We particularly thank S. B. Abella, D. E. Allison, D. Hairston, D. J. Hall, A. II. Litt, D. Myhre, and J. Shapiro. The Tacoma office of the U.S. Geological Survey has provided data on streamflow in advance of formal publication. Mem- bers of the Municipality of Metropolitan Seattle (METRO) have provided infor- mation and unpublished data, as have in- dividuals at the II. II. Chittenden Locks.

Methods Snmpling-Frequent sampling was

necessary to determine maximum values, annual changes, and rates of change in concentrations. Sampling was therefore concentrated at a central station off Mad- ison Park which is representative of a large volume of the lake; at appropriate intervals, other widely distributed sta- tions were sampled to check horizontal variation. The sampling program was based partly on earlier studies. Scheffcr and Robinson (1939) found similar con- ditions and variations at stations at both ends of the lake and in the middle. Pe- tcrson (1955) took surface samples from 26 stations twice a month for a year. The same general results were given by all the deep-water stations away from local shore influence. For example, he found the winter mean concentration of phos-

phate-P in six deep-water stations along the axis of the lake to vary from a low of 24.6 pg. liter I at the Madison Park sta- tion to a high of 27.4 at the next station to the north. The small differences among stations were not significant and showed no consistent pattern along the lake. The annual mean varied from 7.0 to 8.9 pg. liter-’ in a graded manner from one end of the lake to the other, but the differences were not significant. IIe found differences among stations greatest when concentrations were changing rap- idly and a gradient could develop along the lake or from the middle to the ends.

During much of the period reported, our samples have been taken weekly for partial analysis; complete analyses at all depths were made only every 2 weeks. Since 1973, the frequency, number of depths, and variety of analyses has been somewhat reduced (see Edmondson 1977(i).

Hydrology-Fluvial discharge to Lake Washington was divided into four cate- gorics: Cedar River, Sammamish River, small streams cntcring the lake directly (Coal Creek, Juanita Creek, Lyon Creek, May Creek, McAleer Creek, Merccr Creek, and Thornton Creek), and direct runoff (Fig. 1). The Cedar River is gauged near its mouth (USGS station No. 1190); tabulated daily discharge values are reportedly accurate within 5 or 10% (U.S. Geol. Surv. Water Resources Data for Washington 1964-1978). Discharge values for the Sammamish River were rc- constructed from direct measurements of the flow of the river at Woodinville (USGS station No. 1252) plus the dis- charges of its three downstream tributar- ies: North Creek (USGS 1260), Bear Creek (USGS 1255), and Swamp Creek (USGS 1271). For years when not all the tributaries wet-c gauged simultaneously, we computed discharge from the avail- able data from linear regressions on ma- jor fractions of the total flow. Similarly, WC took the combined discharge of all small streams tither from the sum of flows measured simultaneously and in- dependently for all of them, or from lin- car regressions used to reconstruct total

Nutrients and Lake Washington

hydrologic discharge of all the streams from the flows of only a few when gaug- ing stations were operative on only a few (table 9: Lehman 19,78).

No measurements of direct runoff to Lake Washington are available, in part because the category includes small un- gauged streams not named in Fig. 1 and all diffuse and temporary terrestrial sources not separately identifiable. In lieu of actual measutiements, we took val- ues from a hydrological model prepared for METRO (Hydrocomp: G. Farris pers. comm.) in which weather, topography, and land characteristics of the Lake Washington basin were used to predict runoff to the lake and to calculate evap- oration at the lake surface. The confi- dence of the model-generated values was reported to be 10% o;f the monthly means (T. W. Holz pers. cornm.), but regressions of model predictions against actual gauged flows for drainages where the lat- ter are available suggest that 95% confi- dence limits are cl:oser to 20% of the means; we use the higher error values here. Because the Hpdrocomp simulation was performed only1 through September 1972, linear regressions were used to es- timate the values for later years (table 10: Lehman 1978). We also used the IIydro- camp model to estimate total discharge of the Sammamish River when no gauge measurements were available for the ma- jor portion of the flow (October 1963-Jan- uary 1965).

Estimates of hydrologic outflow from Lake Washington were based on sum- mary inflow figures :and changes in the volume of water retained in the basin from one month to the next:

outflow = inflow f precipitation - evaporation - A lake vol.

(1)

Lake volume was calculated from records of lake level at the H. H. Chittcnden Locks on the Lake Washington Ship Ca- nal (A. Ficken pers. comm.). Changes in surfiace height of th,c lake were multi- plied by the surface area to compute the changes in volume (87.615 x 10” m2, Ta- ble 1). Operation of the locks and spill-

SWAMP CREEK

L-m 0 5 KM CEDAR RIVER

Fig. 1. Tributnrics to Lake Washington. Arrows numbered l-l 1 show scwagc treatment plants (.~ee Table 4). Four-digit numbers are those of U.S. Geo- logical Survey gauging stations llscd to calculate flow (see Tables 3 and 5). Based on USGS quadran- glCS.

ways is constrained by law to regulate the height of Lake Washington within a range of 0.6 m as closely as possible, so the correction for lake volume is a minor one, although outflow always exceeds in- flow during summer and inflow exceeds outflow during winter.

Nutrients and nutrient loading-Con- centrations of nutrients were measured by standard, widely used methods and the accuracies of the chemical analysts performed in our laboratory have been assessed (Edmondson 1977a; table 11: Lehman 1978). We use the term PO,-P to describe all filterable, molybdate-reac- tivc P present; dissolved P refers to the filterable P measurable after perchlorate digestion. We have measured conccntra- tions of P and N in the Cedar River, Sam- mamish River, and Thornton Creek about

4 Edmondson and Lehman

Table 1. Above: Morphometric data for Lake Washington. Bnscd on chart 6449 by the U.S. Coast and Geodetic Survey (currently NOAA 18447). Contours were drawn at 5-m intervals and planim- etered by W.T.E. Volumes of the slices were cal- culated by a trapezoidal formula. Contours used for this table arc only slightly different from those in Fig. 2. Union Bay is not includccl.

Below: Areas of land draining into waters listed (based on maps provided by the U.S. Geological Survey, Tacoma, Washington) and percentago of land forested (based on information from METRO).

Depth (n-4

Area (10R my)

0 87,615 Total 2,885 1.000 5 80,755 O-10 805 0.279

10 72,925 10-20 673 0.233 15 67,465 20-bottom 1,407 0.488 20 61,585 25 56,360 30 50,660 35 41,645 40 34,770 45 28,020 50 20,690 55 16,025 60 4,065 65 20 65.2 0

Areas of land draining into Chester Morse Lake Cedar River below Chester

Morse Lake Lake Sammamish Sammamish River below

Lake Sammamish Lake Washington not

including Union Bay (includes 1Mercer Island in Lake Wash in&on)

km’ %

207 90

289 52 246 91

216 82

316 53

twice monthly since 1970. Series of sam- ples were collected at 3-5-min intervals on several of the sampling dates: the vari- ance encountered was considered to be representative of the uncertainty with which any single nutrient analysis esti- mates concentrations in a stream. These standard errors for single estimates were on average 12.4% for total P, 13.4% for dissolved I’, 14.7% for PO,-P, 2.4% for NO,-N, and 4.8% for total reduced N. By using linear interpolation to estimate concentrations at the first (t!,) and last (t,,) days of each month, we can calculate

mean monthly concentrations (c) by trapezoidal summations throughout the monthly intervals:

The mean monthly concentrations for any nutrient multiplied by monthly hy- drologic discharges give estimates of monthly nutrient loads discharged to Lake Washington from fluvial sources. This method may introduce some bias in estimations that include particulate P or N (e.g. total N or total P), because con- centrations of these substances are most variable between dates and tend to be positively related to discharge volumes; inputs from brief storms might thus be inadequately sampled. However, for 223 simultaneous measurements of nutrient concentrations and hydrologic discharge from the Cedar River between 1970 and 1977, there are no statistically significant relations between discharge volume and concentration for discharges ~60 m3. s-l. Only among the seven measurements made during floods in 1972 and 1975 did concentrations show a marked depen- dence on discharge volume: concentra- tions of particulate P increased in the flood waters, but dissolved P was not sig- nificantly increased. This general lack of correlation between discharge velocity and nutrient concentrations means that the calculation procedure chosen is ac- ceptable. The use of linear interpolation to estimate monthly mean concentrations is acceptable also, not because concen- trations really change linearly with time in the streams, but because measure- ments on sequential sampling dates often do not differ any more than those on the same day. The imprecision of the mea- surements will be treated in our analysis of error propagation.

Of the small streams, we have analyzed only Thornton Creek regularly since 1970. We have used analyses for the oth- er small streams made twice weekly dur- ing 19,57 by the Seattle Engineering De-

Nutrients und Luke Washington 5

partment (II. H. Phillips pers. comm.) and weekly during 1964 by METRO (G. Farris pers. comm.):. From this informa- tion we computed factors that relate mea- sured mean monthly concentrations in the more frequently sampled sources to the mean concentrations of the water dis- charged by all small streams combined (details in table 12: ,Lehman 1978).

Atmospheric sour&s-Atmospheric in- put of nutrients from dustfall and from precipitation measured at Seattle-Ta- coma International Airport (U.S. Envi- ron. Data Serv. Climatol. Data) are listed in Table 2. Sampling stations surround- ing Lake Washington were selected (Johnson ct al. 1966) for calculation of nutrient input to the lake by dustfall. Rates were not significantly different among the 3 months! studied (June, July, and August) so we pooled the data to es- timate single means and errors for load- ing to the entire surface of the lake.

We used concentrations of P and N in precipitation given by Moon (1973), who analyzed precipitation collected at Lake Sammamish (8 km from Lake Washing- ton) during 1971 and:part of 1970. No sea- sonal effects were evident in the mea- sured concentrations, so we pooled all values to estimate mean concentrations and their standard errors.

To estimate constituents not measured directly in these studies, WC assumed PO,-P to equal 0.5 (SE = 0.1) of the total P, and the total reduced N to equal 1.0 (SE = 0.5) of the NOR-N measured, based on preliminary analyses of local rainwa- ter (D. Spyridakis pers. comm.). In all casts, these amounts were only a minor fraction of the total because inputs were dominated by fluvial, discharge,

Comparisons with previous cnlculn- tions -The quantities given here were calculated quite independently from those given by Edmondson (1961, 1969, 1977h); we used the same chemical data but with additional information on flow, made different assumptions about un- measured sources, and used a somewhat different system of calculation, The pres- ent estimates consequently vary some- what from those published earlier since

Td,lc 2. Atmospheric inpllts of nutrients to Lake Washington.

TO entire

lake mg. m 2. SlllfWC

,110 ’ SE n (t.mo ‘) SE

D11stfa11” PO,-P 3.24 0.62 33 0.284 0.054 NO,,-N 11.26 1.25 33 0.987 0.110

pg.liler-’ SE 11

Concn in precipitation? Total P 22.4 2.2 17 NO:,-N 63.6 11.5 17

* Arcal fl11x computed from Johnson et al. 1966. t Mcncllrecl by Moon 1973.

they are based on more information. For example, we have included atmospheric inputs here (omitted from earlier csti- mates). On the other hand, we have not tried to estimate nutrients in combined sewer overflow, one of the least certain sources. Seattle has a combined sewer system. A program to separate sanitary sewage from storm water has not been completed, and during rainy periods some amount of dilute sewage overflows into the lake. A special study of the quan- tity in 1977 was made by METRO (J. Buf- fo and G. Farris pcrs. comm.). At that time, the best estimate of total phos- phorus in storm water and combined sewer overflow was about 7,000 kg. yr’, or aborrt 0.08 g*rn-‘*yr-‘. Most of this is accounted for within our definition of runoff.

The differences among estimates arc relatively small in most cases. Our prcs- cnt estimates of total phosphorus income in 1957 and 1964 are 93,200 and 204,200 kg; Edmondson (1977h) gave them as 99,600 and 202,308. The greatest diffcr- encc is for total P in 1962, an estimate based on the mcasurcments made in 1964. Edmondson (1977h) assumed that the fluvial contribution of total P in 1962 was the same as in 1964, 80,900 kg; how- ever those measurements were made when the Army Corps of Engineers was channclizing part of the Sammamish Riv- cr, an operation that caused considerable

6 Edmondson und Lehman

-45'

-40'

-35'

1220

I I I I I I I I 3' 15' IO'

Fig. 2. Morphomctric mq) of Lake WuslGngton. straight lines show the floating bridges, Evcrgrcen I<cdrawn from fig. 1 of Gould and Budingcr (1958) Point (ECPB) and Merccr Island (MIB). *--Main which was hascd mostly on U.S. Coast and C&o- sampling station cast of Madison Park. During detic Survey Chart 6449 (now NOAA 18447), sup- strong south winds, samples are tnkcn just north of plcmcnted by original soundings by the University EPB. Former sewage trcatmcnt plants are indicated !of’ Washington Deprtment ol’ Oceanography. Two as in Fig. 1: depth of outfall is slrown.

erosion and increased the input of partic- ulate phosphorus. Not only did this extra load of phosphorus not exist in 1962, but water flow was low relative to 1964, and we now estimate the actual flrrvial input of P in 1962 to have been only 32,400 kg.

WC know of no prospect of major de- velopmcnts that would change the pres- ent estimates in an important way, but some further refinement is certainly still possible, particularly in estimating the contribution of small streams and sewer overflow.

Outflow of nzctrients-The outlet from the lake is a canal 9 m deep with locks that lead to Puget Sound; it connects with the lake basin via Union Bay, Since the 195Os, two stations have been sampled within 300 m east and southeast of the mouth of Union Bay and the entrance to the canal (Fig. 2). They provide data on the properties of water leaving the basin and permit calculation of the rates of ef- flux of nutrients from Lake Washington. For these calculations, we have used av- erage concentrations in the 0-10-m stra- tum, because of the depth of the canal itself and because even during thermal stratification the upper mixed layer is al- most always at least that deep.

Terrigenous sources of nutrients be- fore 1970-The input of nutrients to the lake is poorly known before 1970, except for 1957 and 1964. Because magnitudes of nutrient loading vary substantially from month to month and from year to year in response to changes in hydrologic loading, no “average” loading figures can be assigned securely to intervals not rep- resented by actual measurements unless the hydrologic loading is taken into ac- count. This is because nutrient concen- trations in the streams can vary by sev- eralfold during a season or year, but discharge volumes vary by orders of mag-

Nutrients and Lake Washington 7

Table 3. Calclllation of Auvial + atmospheric loading of nlltrients to Lake Washington when nutrient concentrations in inflllcnt streams are unknown or poorly known (years before 1970, except 1957 and 1964). Mean monthly loading figures (10:’ kg.mo-‘) arc calculated from regression cyuations u,X + ~1,) computed for I.970- 1976, In some cases, data have been subjected to a fraction transform (see text) over part or all of their domain.

X (1 I (10 n r SE

Total P Hydrologic input, O-300 x 10” rn”* mo-’ 0.0341 0.809 82 0.651’” 29.7%

Total reduced N (total KJ:eldahl N) Hydrologic inpllt, O-300 x 10” m”.mo ’ 0.317 0.0904 82 0.659” 26.7%

PO,-P (solul)lc reactive P) Hydrologic input, 1) O-75 X 10” m3. mo -I 0.0116 0.475 35 0.676* 23.6% 2) 75-400 X 10” m” * mo-’ 0.0131 0.1.52 49 0.853 0.62 x LO” kgemo-1

L)issolvcd I? IEydrologic input, 1) O-75 X LO” m3. mo ’ 0.0142 0.475 26 0.624” 27.2% 2) 75-400 x IO” rn3. mo -’ 0.0146 0.141 34 0.811 0.90 x 10” kg*mo-’

Nitrate N Hydrologic input, 1) O-75 X 10” mR+ mo -I’ 0.301 1.449 33 0.641* 23.6% 2) 75-300 X 10” mzS*mo-’ 0.714 -34.16 42 0.954 14.5 x l(F kg*mo-’ Input of NQ1-N from Cedar and Sammamish Rivers, 1) O-20 x 10” kg NC&Nemo-’ 1.563 3.059 43 0.948 2.53 x IO” kg*mo-’ 2) 20-150 x 10” kg NO,-Nemo-I 1.312 10.58 35 0.984 8.60 x lo” kg* mo 1

* For transfornd data (see text). t Regressions colnptkxl for 1972-l 976.

nitude. Rates of nutrient loading to Lake Washington from external sources were consequently calcul+ed for the most part by linear regression, on hydrologic input for years before 1970, except when more direct data were available.

The regressions were calculated for the data accumulated from 1970 to 1976. Curve-fitting was plu-cly empirical, the only goal being to minimize errors of the regression estimates. A large degree of autocorrelation is expected from this pro- cedure because hydrologic discharge, the correlate, is used to calculate nutrient loadings from the streams in the first place (Scheider et dl. 1979). Deviations of single estimates IFrom the regression lines simply reflect ihc extent to which nutrient concentrations cannot be pre- dicted from dischqrge measllrements a one, 1 as mentioned earlier. The equa- tions finally selected and the associated standard errors of the estimates are listed in Table 3. In some eases the data had to

be transformed bcforc standard errors could bc computed, most commonly when the variance of the dependent vari- able (nutrient loading) increased with the magnitude of the independent variable (usually hydrologic loading). The trans- formation used in these cases put the data in terms of percentage deviations from the line of least-squares fit over the do- main of interest; since linear regressions could always be applied validly to the trans formed data, the calculated standard errors were cxpresscd in percentage units. Adopting such a proccdilre rather than using a log-log transform, for in- stance, has the advantage of keeping the standard errors of the estimates in units easily related to the original data, which Facilitates calculation of propagation of error (see helozo).

Sewage in Luke Wushington-A sub- stantial source of nutrient loading to the lake’ in the 1950s and early 1960s was ef- fluent from sewage treatment plants. Di-

8 Edmondson nnd Lehman

Table 4. Waste treatment plants discharging to Lake Washington l~eforc 1968 (see also Pigs. I mu! 2).

Population scrvcd

Source Type* Ihlk!

diverted Ref. year 1957 1964

Year diverted

I. T,ake City 2. Sand Point

Naval Air Sta. 3. Sand Point homes 4. Bryn Mawr S. Ucnton Mlmicip. 6. Kenton-Boeing 7. 13cllcvuc 8. Kirkland 9. Shorcwood

10. E. Mcrccr Island 11. Rothell

AS TF

AS AS TF TF AS TF AS AS TI’

19.5 0

0 9.1

Cedar Rivcl Cedar Rivcl

12.8 3.0

25.6 13.4

Sammamish Rivcl

1952 Mar 67 1964 26,000 73,790 81,000 1941 Feb 68 1957 900 520 520

194 1 Mar 67 1964 1953 Fell 63 1957 1943 Fcb 63 1957 1942 Fcb 63 1957 1953 Jr11 65 1964 1943 Apr 66 1964 1949 Jul 65 1964 1954 Feb 66 1964 1959 Mar 67 1964

700 4,500

14,800 6,000 4,100 4,000 2,800

500 -

700 - - -

25,980 7,330 1,630 1,170 2,600

700 5,440

12,840 6,000

30,000 7,500 1,630 1,170 2,700

* AS-xtivatcd ~111dgc; ‘I’F-trickle filtration.

version of the discharges from these plants directly to Puget Sound began in 1963 and was completed by 1968, the rc- suit of a major public works project (Ed- mondson 1972).

Detailed measurements of nutrient dis- charge from the treatment plants arc available from samples taken twice weekly in 1957 and weekly in 1964 (Ta- ble 4). For other years, magnitudes of an- nual loading were calculated in propor- tion to the recorded sizes of human populations served by the plants. When- ever possible, 1964 was used as the ref- crence year for nutrient loading in the 196Os, because the widespread use of phosphate in detergents is better repre- sented by the 1964 data than by figures from 1957. In cases where effluents were diverted from the lake before or early during 1964, only data from 1957 can be used.

scaled to the total population served by the plants, the standard errors of these unnual figures were assumed to be 10% for each plant. One plant, Rothell, oper- ated between 1959 and 1967, but was not included in the 1964 study. Estimated discharge of this plant was scaled to pop- ulation as was the similar treatment plant in Kirkland; the resultant annual figures were assumed to have a standard error of 20%.

Sediment traps-To quantify the downward flux of particulate matter, we have used collecting devices in the lake since November 1965. The traps were suspended 50 m down and 10 m above the bottom, from the Evergreen Point Floating Bridge. Contents of the traps h ave been retrieved at intervals of 2 weeks for analyses of the accumulated seston; N and I? have been determined in the collcctcd material since 1971.

After diversion started, some plants The traps are of two types. For one were discharging for only a fraction of a type, used during the entire study, plastic year; annual input figures were scaled cylinders 15 cm across, 30 cm deep and accordingly. During the rcferencc years, tapering further 10 cm to a narrow neck nutrient concentrations were measurccl were made by cutting the bottom off a 4- discontinuously but frequently, and dis- liter polyethylene bottle. Scdirnent col- charge volumes were monitored contin- lects in the necks of the inverted bottles, uously: the standard errors of mean or in IO-cm-long plastic extensions added monthly loading figures computed for to the necks. In the other type of sedi- each plant were thus estimated to be no ment trap, used since May 1974, a IO-cm more than 5% of the estimates. For other fln-mcl leads through a stopper into a 50- years, when annual loading figures were ml plastic centrifuge tube. Four such

Nutrients and Lake Wushington

400

300

T 2z

5 E

% 200

“0 -

100

0 ):;I

I I I I I I I 1 I

I970 1971 1972 1973 1974 1975 1976 1977 1978

Fig. 3. Monthly mean fluvial discharge entering Lake Washington, corrected for precipitation and evaporation at the lake surface. Error bars denote +2 SE, which approximate 95% confidence interval.

units are mounted on a piece of acrylate (Birch 1976). All traps were raised to the surface slowly to minimize loss of mate- rial. Because several different designs and collection schedules were experi- mented with over the years of this work, data are available in different years from traps suspended at 30 or 50 m, and col- lected at intervals of 1 or 2 weeks. When- ever a new design or configuration was introduced, results were continued in parallel with previous procedures long enough for substantial comparisons.

Regardless of the collection scheme, monthly yields of accumulated nutrients were calculated for each trap by the in- terpolation and summation procedure al- ready described. Subsamples of the ma- terial collected in the settling traps were analyzed for total P and total Kjeldahl N; reported mass flux was scaled to the area of the orifice of the collecting device (mg-cm-“* mo-l). Analyses of the vari- ance among the monthly yields of all traps and collection schedules in use si- multaneously were used to test the con- sistency and uniformity of different methods.

Error Drowaention-Our analvsis of

error propagation follows the technique described by Meyer (1975) to deal with random errors that may be either corre- lated or uncorrelated. A practical intro- duction to the method, called “first order uncertainty analysis,” as it applies to cal- culations of nutrient loading is given by Lettenmaier and Richey (1979), Reckhow (1979), and Lehman (1978). In general, unless the function Y =f( Xi) contains correlated errors among the Xi, the vari- ance of Y is approximated by the usual error propagation formula:

i= ft (Ty* = C (aflaXi)*UzY,*.

i=l

For instances when the errors are corre- lated, as when some of the Xi are deter- mined by difference, or from linear regression on other variables, then co- variance terms must be included in the description:

Cl. * = ig [ ((jj-,(jxi)*qx,:! i=l

+ jz 2(afiaxi>(afJaxj>

10 Edmondson und Ilehman

50- TOTAL P

-c 40-

E 30-

,* 20-

“2 IO-

U I I I 1970 1 1971 1 1972 1 1 1973 1974 1975

I 1976 1977 1978

.4 “E

E .3

c E

.2 a cr,

G 400 5

E 0 4 E 300

G E

z 3 2

2200 75 2 ‘;;

z too

I cI,

0 0

-L -.08 2 z 6 5

-.06 E a

P 4 E

-.04 r

“a E

a 2

-.02 cI,

0 ) 1 I I I r I I I 0 1970 1971 1972 1973 1974 1975 1976 1977 1978

250

1 NO,-N I

0 0 1970 1971 1972 1973 1974 1975 1976 1977 1978

Pig. 4. Monthly mean inprlts of N and 2’ to I,ake Washington from tcrrestriul and atmospheric SOIIICCS. Error bars denote +_2 SE.

crx, is the standard error of Xi, and p(Xi Xj) is the correlation between Xi and X,i. Most of the hydrologic inputs to Lake Washington were determined indepcn- dently, but there are several years during which the combined discharge of all small streams was computed from mea- sured discharge of the Sammamish River, and other years in which runoff was com-

puted from small stream How. Where ap- plicable, the covariance terms, using cor- relation coefficients listed by Lehman (1978: table 9) have been included in the analysis of error propagation.

The standard errors and confidence in- tervals associated with the results that follow have all been computed by these propagation formulae. Nutrient budgets

Nutrients und Luke Wushirzgton 11

20-

-i TOTAL P c -6

,” IO-

a o,w.& ’ ’ ’ I( _.,;‘I4 1 1970 1971 1972 1973 1974 1975 1976 1977 1978

0 0

1970 1971 1972 1973 1974 1975 1976 1977 1978

150

1 NO,-N

I I

1976 1977 ’ 1978

0 0

1970 1971 1972 1973 1974 1975 1976 1977 1978

Fig. 5. Monthly net inputs of N and P to Lake Washington (input minll\ olltflow). Error lxu-s denote +2 SE.

and errors were calculated numerically from primary data, their measurement errors, mass balance equations, and co- variance matrices in the cases of corre- lated errors.

Results Fluvial inputs-Because the fluvial

discharge of the major inlets to Lake

Washington is known with great accura- cy, the total hydrologic input to the lake is associated with very little error (Fig. 3). The means and errors of the hydrolog- ic inputs, including monthly precipita- tion measured at the Seattle-Tacoma In- ternational Airport, can be combined with the measured or calculated values (and standard errors) of nutrient concen-

12 Edmondson and Lehman

Table 5. Hydrologic loading. Volume of water supplies to Lake Washington from fluvial and aerial sources, corrected for evaporation at the lake sur- face.

Year

1950 1957 1962 1963 1964 1965 1966 1967 1968 1969 1970 1071 1972 1973 1974 1975 1976 1977 1978

Ill” mR.yr ’ FiWbllpf i1

Total SE renewed

1,646 10 0.570 987 6 0.342 918 5 0.318

1,010 9 0.350 1,497 16 0.519 1,101 10 0.382 1,118 6 0.387 1,105 6 0.383 1,448 6 0.502 1,129 5 0.391 1,127 7 0.391 1,506 8 0.522 1,687 11 0.585 1,054 16 0.365 1,477 16 0.512 1,706 16 0.591 1,056 16 0.366 1,030 16 0.357 1,017 15 0.352

trations in these sources to estimate the total input of nutrients to the lake.

The results of these computations are shown in Fig. 4. The vertical axes iden- tify both the total mass (in metric tons, t)

of each nutrient species calculated to en- ter the lake each month (t* mop’) and the areal loading, that is total input divided by the surface area of the lake. All the graphs show a pronounced seasonality, with maxima each year in winter. The effects of large floods in the Cedar River valley are evident in 1972 and 1975, es- pecially for properties that include par- ticulate matter (e.g. total P and total re- duced N). The loading figures are corrected for outflow in Fig. 5, which gives the difference between total input and the nutrients lost from the basin with the outflow.

Because for most years before 1970 mean monthly inputs of N and P are cal- culated from linear regressions using measured hydrologic discharges, the monthly figures are sometimes associated with substantial estimation errors, and plots would show less confidence than Figs. 4 and 5. Nutrient loading during most of the 1950s and 1960s was affected by discharges from treatment plants, and that source is known on a month-by- month basis for only 1957 and 1964.

Annual summaries and lake totnls- Total and net inputs of nutrients to Lake

Table 6. Total P. Annual loading figures for I,akc Washington. Units are 10” kg.yr ‘, with SE in parentheses.

Yeill Waste Fluvinl Atmospheric Total Outflow Net inpIt

1950 1957” 1962 1963 1964-t 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 J 977 1978

19.0(0.6) 42.4(0.4)

107.5(6.7) 98.4(6.8)

104.9(1.1) 95.6(7.1) 77.0(7.2) 14.0(1.2) 0.1

56.2(6.4) 9.6(0.4) 84.8(6.4) 42.3 8.5(0.4) 93.2 32.4(3.9) 8.6(0.4) 148.5(7.8) 49.4(8.6) 8.7(0.4) 156.5(11.0) 90.4( 1.8) 8.9(0.4) 204.2(2.1) 38.7(4.8) 8.5(0.4) 142.8(8.6) 39.1(4.7) 8.7(0.4) 124.8(8.6) 31.7(3.8) 8.6(0.4) 54.3(4.0) 49.7(5.6) 9.3(0.4) 59.1(5.6) 39.7(4.6) 8.5(0.4) 48.2(4.6) 50.3(1.7) 8.7(0.4) 59.0(1.7) 44.8(1.3) 9.0(0.4) 53.8(1.4) 94.2(2.4) Y.Z(O.4) 103.4(2.4) 34.3(1.5) 8.6(0.4) 42.9( 1.6) 49.8(1.8) 8.7(0.4) 58.5(1.8) 90.3(2.2) 9.0(0.4) 99.3(2.2) 34.8(1.1) 8.1(0.4) 42.9(1.2) 51.0(1.5) 8.4(0.4) 60.3(1.6) 40.0(1.4) 8.6(0.4) 48.6(1.5)

25.2(0.7) 66.3 51.4(0.4) 97.1(7.8) 68.8(0.7) 87.7(11.0) 92.7( 1.1) 111.5(2.4) 66.7(0.7) 76.1(8.6) 65.8(0.4) 59.0(8.6) 5O.cqO.4) 3.4(0.4) 46.2(0.4) 12.9(5.7) 26.5(0.3) 21.7(4.6) 25.3(0.3) 33.7(1.7) 29.0(0.4) 24.8(1.4) 30.0(0.4) 73.3(2.4) 18.8(0.4) 24.1(1.5) 29.6(0.4) 29.0( 1.8) 29.9(0.5) 69.4(2.1) 15.7(0.3) 27.2(1.1) 16.8(0.4) 43.5(1.6) 18.9(0.3) 29.7( 1.4)

* Data for inputs arc from II. M. Phillips; standard errors were not documcntrd. t I);tta for inpntq are from METRO.

1:3

250

I f TOTAL t

1 WA:TE 1 1

62 I

63 64 65 66 67 68 69 70 71 72 73’74 75’76 77 78

YEAR

2000

7 I

- i

: 1500 z-7

“0 ~looo- +

z 0

2 500- J

0 Z

+ TOTAL

f t t

WASTE

o 0 0

0 0

62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78

YEAR

Fig. h. Mass of‘ P and N added annually to Lake Washington ti-om \va\te treatnlc~nt l)lcmt\, ,md total inputs \rlmmed from all sources except in \itu N fixation. The error bar\ denott, 22 SE. \\‘lrvn not \ho\\ II, confidence interval does not exceed dimensions of‘ symbol.

\Vashington have been expressed as an- nual summaries in Tables 5-10 for the years during which detailed limnological studies were done. The loading figures show the substantial impact of effluent from treatment plants on the rate of sup- ply of phosphate and other dissolved P to the lake. By comparison with atmospher- ic ant1 fluvial sources, sewage effluents affected the P economy of the lake much more than that of N (Fig. 6). No measure- ments of nitrogen fixation were made when heterocystous blue-green algae were abundant. Measurements finally made in the early 1970s showed negli- gible fixation in the open water (J. Richey pers. comm.).

The total mass of substance in the lake on any given day was calculated from

vertical profiles by the formltla: /=1/-l 1 c-1 3 c,A, + C’, +,A, , i=l

+ p,A,c,.,A,+,) (zv+, ~ G> 65)

where c’, is concentration atlcl A, is area at depth z, , with clepth measured posi- tive downward. Mean monthly values fo1 total P and particulate P in the lake arc’ plotted in Fig. 7. The amount of partic- ulate P is closely related to the abnn- dance of plankton, partic~lllarly phyto- plankton, as shown by the strong11 marked seasonal changes. The effect of sewage diversion on the phytoplankton is shown more clearly 1,). changes in chlo- rophyll (Fig. 7). The proportion of fila- mentous blue-green algae was very high

14 Edmondson and Lehmun

Table 7. As Table 6, but for total N.

Inpnts YWW Waste FIllvial Atmospheric Total* outflow Net inprlt

1950 74( 1) 1,321(63) 39(3) 1,434(63) 1957-t 171(2) 1,712 34(3) 1,8 17 515(8) 1,392 1962 350( 18) 627(45) 34(3) 1,O 1 l(48) 605(4) 406(48) 1963 307( 18) 736(47) 35(3) 1,078(50) 646( 5) 432( 50) 1964$ 328(3) 1,057 35(3) 1,419( 13) 967( 9) 452( 16) 1965 295( 19) 820( 52) 33(2) 1,148(55) 808( 7) 340( 55) 1966 224( 18) 8 15(51) W2) 1,073(54) 766(4) 307( 54) 1967 45(3) 749(42) 34(3) 828(42) 665(4) 163(42) I968 1 1,111(58) 38(3) 1,150(58) 853( 5) 297( 59) 1969 8 15(49) W2) 848(49) 607( 4) 242(49) 1970 940(24) 34(3) 974( 24) 602( 4) 371(23) 197 1 1,067(2 1) 36(3) 1,103(2 I) 743( 5) 360(2 1) 1972 1,474(29) 37(3) 1,5 I l(29) 8 18(6) 693(28) 1973 864(24) W2) 898(24) 478(6) 420(22) 1974 1,131(27) 34(3) I, 165(27) 686( 7) 479(25) 1975 1,400( 30) 36(3) 1,436( 30) 756( 7) 680( 28) 1976 703( 17) 31(2) 734( 17) 451(6) 283( 14) 1977 875( 18) W2) 908( 18) 393( 5) 515( 17) 1978 885( 18) 34(3) 919( 18) 414(4) 505( 18)

* In situ N fixation not included. t Data inputs arc1 from II. M. Phillips; standard errors were not documented. f Data from inpIlls are from METRO.

during eutrophication, exceeding 90% by magnitudes of errors are not adequately volume much of the slimmer, and re- known. The lake totals are computed mained high until 1972 after which it de- from vertical series of samples collected creased markedly. at standard sampling sites in the central

No confidence limits are included with region of maximum depth of the lake (sta- the values plotted in Fig. 7 because the t-ion MP or EPB). We assume in making

Table 8. As Table 6, but for dissolved P.

Inputs

Waste Fuvinl Total Olltflow Net input

1950 1957” 1962 1963 1964t 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978

lg.O(O.6) 42.4(0.4)

107.5(6.7) 98.4(6.8)

104.9( 1.1) 95.6(7.1) 77.0(7.2) 14.0( 1.2) 0.1

22.2(2.6) 4.8(0.2) 46.0(2.7) 35.9 4.3(0.2) 82.6 12.8(2.3) 4.3(0.2) 124.6(7.1) 14.0(2.3) 4.4(0.2) 116.8(7.2) 20.4(2.6) 4.4(0.2) 129.7(2.8) 15.2(2.5) 4.2(0.2) 115.0(7.5) 15.3(2.5) 4.4(0.2) 96.7(7.6) 15.8(2.2) 4.3(0.2) 34.1(2.5) 19.4(2.6) 4.7(0.2) 24.2(2.6) 15.9(2.4) 4.2(0.2) 20.1(2.4) 15.5(2.5) 4.3(0.2) 19.8(2.5) 20.5(2.6) 4.5(0.2) 25.0(2.6) 28.3( 1.0) 4.6(0.2) 32.9( 1.0) 18.2(0.7) 4.3(0.2) 22.5(0.7) 21.4(0.8) 4.4(0.2) 25.8(0.8) 15.2(0.7) 4.5(0.2) 19.7(0.7) 13.8(0.5) 4.1(0.2) 17.9(0.5) 22.9(0.7) 4.2(0.2) 27.1(0.7) 18.4(0.6) 4.3(0.2) 22.7(0.6)

8.4(0.7) 73.3 35.4(0.3) 89.2(7.1) 50.3(0.5) 66.5(7.2) 63.4(0.9) 66.4(2.9) 47.0(0.6) 68.0(7.5) 38.0(0.3) 58.7(7.6) 29.7(0.3) 4.4(2.5) 19.6(0.3) 4.6(2.6) 14.9(0.3) 5.2(2.4) 13.5(0.3) 6.3(2.5) 16.9(0.3) 8.1(2.6) 13.9(0.4) 19.0( 1.0) 10.7(0.3) 11.7(0.7) 15.9(0.4) 10.0(0.8) 18.2(0.4) 1.6(0.7)

8.7(0.3) 9.2(0.5) 10.4(0.3) 16.7(0.7) 12.0(0.3) 10.7(0.6)

* Data for inputs are from H. M. Phillips; standard errors were not documented. t Data for inpntq are from METRO.

Nutrients und Luke Wushi)lgto~~ 115

Table 9. As Table 6, but for PO,-P.

18.1(0.6) 38.4(0.4) 96.8(6.2) 88.6(6.3) 94.4( 1.7) 86.0(6.5) 69.3(6.6) 12.6(1.1) 0.1

19.4( 1.8) 4.8(0.2) 17.0 4.3(0.2) 11.2(1.6) 4.3(0.2) 12.3( 1.6) 4.4(0.2) 18.0( 1.8) 4.4(0.2) 13.5( 1.7) 4.2(0.2) 13.4( 1.7) 4.4(0.2) 13.9( 1.5) 4.3(0.2) 17.1( 1.8) 4.7(0.2) 13.9( 1.6) 4.2(0.2) 15.7(0.5) 4.3(0.2) 15.8(0.5) 4.5(0.2) 23.4(0.8) 4.6(0.2) 16.6(0.7) 4.3(0.2) 16.6(0.6) 4.4(0.2) 20.1(0.7) 4.5(0.2) 11.8(0.3) 4.1(0.2) 15.4(0.5) 4.2(0.2) 14.5(0.5) 4.3(0.2)

42.3( 1.9) l&7(0.6) 2X6( 1.9) 59.7 3.3(0.4) s5.Yj

112.3(6.4) :3:3.3(0.2) 79.0(6.4) 105.3(6.5) 44.8(0.4) 60.5(6.Fi) 116.8(2.5) 57.9(0.8) 58.9(2.6) 103.7(6.7) 32.3(0.6) 61.4(6.7) 87.1(6.8) 27.8(0.2) 59.3(6.8) 30.8( 1.9) 24.3(0.2) 6.ri( 1.9) 21.9( 1.8) L3..5(0.1) 8.4( 1.8) 18.1( 1.6) 9.5(0.1) 8.6( 1.6) ZO.O(O.S) 9.3(0.1) 10.7(0.5) 20.3(0.5) 1 l.Y(O. 1) 8.4(O.rS) ZS.O(O.8) 9.6(0.1) 18.4(0.8) 20.9(0.7) 9.3(0.1) ll.fi(O.7) 20.9(0.6) 15.0(0.2) 5.9(0.6) 24.6(0.7) 15.4(0.2) 9.2(0.7) 15.9(0.4) 6.6(0.1) 9.:3(0.4) lU.S(O.S) 9.1(0.1) lO.S(O.rj) l&8(0.5) 7.6(0.1) ll.l(O.5)

* Data for input\ are from H. M. Phillip\, standsd error\ were not docllmented. t Data for Input\ are from METRO.

the calculation for lake totals that mea- (error propagation) suggests that changes sured concentrations are representative in the masses of nutrients stored in Lake at all points xi across the basin. Data from Washington may be subject to smaller lake surveys show that the assumption is errors than are the magnitudes of the lake inexact, but for calculating year-to-year totals themselves. trends, we think the point relatively un- Nutrient ~&LX to the se&ments---Mean important. The reasoning outlined above monthly accumulations of total P and to-

Table 10. As Table 6, hut for NO,,-N.

Input\

1 t“,r W<l\k Fluvial Atmo\pherlc TOLlI OlltfioM bet InpIt

1 YFjO 1957* 1962 1963 1964t 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978

10 20(l) 49(3) 46(3) 48(l) 48(3) 33( 2) 12(l)

817(42) 20(2) 847(42) 296 17(2) :333 3Fj2( 36) 17(2) 418(36) 432(36) 17(2) 495(36) 659( 9) 1W 725( 9) 486(39) 17(2) Fj51(39) 477(39) 17(2) 527( 39) 507( 34) 17(2) .536(34) 670( 42) 1W) 690(42) 472(37) 17(2) 489(337) Fj56(21) 17(2) 573(21) 661( 19) 18(2) 679( 19) 820(25) 19(Z) 839(25) 526(21) 17(2) .543(21) 693( 24) 17(2) 710(24) 784( 24) lfG9 802( 24) 416( 15) 16(2) 432( 105) 491( 14) 16(2) *507( 14) 510( 16) 17(2) Fj27( 16)

170(.5) 1rjH 248(2) 170(36) :316(3) 180( 36) 462( 7) 263( 12) :3x9( Fj) 162(38) 27:3(:3) 2e54(:39) ‘322( 3) 2 14( 34) 346(3) 345( 42) 282(2) 207( 37) 288(:3) 285(2 1) :371(3) :308( 19) 3 K3(4) 427(24) 247(4) 29Fj( 20) :365(4) 346( 24) :366(4) 436( 24) 219(3) 212( 14) 182(.3) 32*5( 14) l&4(2) :343( 16)

16 Edmondson und Lehman

0, m; 150

y 100 a -I

‘E

4o m E

’ 162163164165166167168169170171 ‘72’73’74’75’76 I 1 77 78

60-

- 0 50-

‘E

F 40-

lJ 30-

5 g 20-

E -I

5 IO-

I YEAR

Ol Y ’ ’ ’ - ’ I

ifi 62 63

’ 64

’ 65 r66167’ 68 69 70 772173 74 75

g I.0 (3 s 0.8

d 0.6

z 0.4 0

E 0.2

B 1-7 -7. - 0 O 69 70 71 72 fY a YEAR

Fig. 7. Total mass of p ;md of pwticuI:&e I’ in Lake Washington plotted by months, concentration of chlorophyll (I in surface samples, and pwportion of total phytol)lankton biomass rcprescntcd by filarncn- tous blue-green algnc in s1irfwc samples.

tal N in sediment collecting traps are to yield a single estimate of sedimenta- plotted in Figs. 8 and 9. Accumulation tion rate whenever ANOVA could detect rates were computed for each trap indi- no significant heterogeneities among re- vidually, and the standard error bars rep- sults from the several trap designs used resent the degree of agreement among simultaneously. Only one sampling de- values from replicate samplers. Results vice showed a consistent tendency to de- from all sampling devices were merged viatc from the group mean: the fLlnnel-

Nutrients and Luke Wushillgtorl 17

0.5

0.4

T -f 0.3 5 E

‘E 0.2 a

0.1

-0.1

NET FLUX OF P TO THE SEDIMENTS

BUDGET CALCULATIONS

0.3 SEDIMENT TRAPS

Fig. 8. Net flux of P to sediments of Lake Washington computed from net input\ of I’ and changes in content\ of water column expressed per unit surface area of the lake (above), and f&n analysis of materiul captured in sediment collecting traps (below). Error bars denote +1 SE; n ranges from 3 to 5. n -Sedi- mentation rates computed for funnel sampler of Birch (1976) which hs a\ . 1 leen ll\ed since 1974 but which usually records higher fluxes of P than do other traps suspended in the lake (ice text). Total volumes of diatom\ in surf:,lce samples are plotted in upper panel (broken line) to show temporal correlation of material flux with spring bloom.

shaped collecting device of Birch (1976), which has been used in Lake Washington since mid-1974, records significantly greater accumulation of P than all other traps used simultaneously (P < 0.0003) by about 50% (see Fig. 8). This anomaly does not exist with regard to total N re- corded by the Birch sampler. Funnel- shaped sediment collectors have been criticized for their relatively poor collec-

tion efficiencies (Pennington 1974), prob- ably due to resuspension and loss of par- ticles when the traps are exposed to advective flows (Gardner 1980). Our re- sults, on the contrary, show elevated cap- ture rates for P by these traps. The ex- periments and development outlined by Gardner (1980), h owever, suggest that funnels may be very effective traps and retainers for particles least subject to re-

18 Edmondson und Lehmun

NET FLUX OF N TO THE SEDIMENTS

1 BUDGET CALCULATIONS,

1970 1971 1972 1973 1974 I975 1976 1977 1978

i- 2- AZ E SEDIMENT TRAPS

E Cu

‘E I-

0 1 I I I I I I I I I

1970 I971 1972 1973 I974 I975 I97 6 1977 1978

Fig. 9. Net flux of N to sediments of Lake Washington, computed as in Fig. 8 from net inputs of N and changes in contents of the water column (above) and from analysis of material captured in sediment collecting traps (below).

suspension. These would be the particles of greatest density, especially mineral particles, which arc more likely to be as- sociated with P than with N. We have plotted values from the Birch sampler separately in Fig. 8.

These results from the sediment traps provide a good way to check on the ac- curacy of computed mass storage of nu- trients in Lake Washington. Intercompar- ison of budget calculations and sediment trap collections is particularly straightfor- ward for P, because changes in lake total P represent a balance between net rates of P input from allochthonous sources an d exchange with the sediments. If the mass of nutrient in the lake is a continu- ous function A4 of which monthly means

Mi are known, the change in nutrient content of the lake during a month is ap- proximated by

mfi-1 + MiPl - wfi + w+m = (Ml-1 - M/+*)/2. (f-3

If, in addition, net input is represented by a step function 1; where each value is mean monthly input minus outflow, then flux to the sediments can be specified as

Si = Ii + [(Al-j - M&,)/2] (7)

where Si is net flux of nutrients to the sediments during month i.

The results of these calculations are shown in Figs. B(P) and 9 (N). Estimates of nutrient flux to the sediments based on

Nutrients and Lake Washington 19

budget calculations (Figs. 8 and 9, upper panels) can be compared with fluxes re- corded by sediment traps (lower panels). The two independent estimates of nu- trient flux agree in showing that flux of P and N to the sediments is maximal during spring. The influence of floods in 1972 and 1975-1976 (Figs. 3-5) is also appar- ent in the figures.

The budget calculations show that there are times of net loss of P and N from the sediments as well as times of net gain. Net loss of sedimentary nutrients appears during late autumn or early win- ter, just when thermal stratification has been thoroughly abolished and mixing begins. Net flux of nutrients to the sedi- ments is usually at a maximum, on the other hand, right at the peak of the spring diatom bloom. Species common at this time are Fragilaria crotonensis, Asterio- nella formosa, Melosira italica, and Ste- phanodiscus niagarae. The total abun- dance, by volume, of all diatoms is plotted in Figs. 8 and 9 (upper panels) for comparison with nutrient flux. Apart from the two episodes of flooding, when much silt containing P and N entered the lake, the correlation between diatom abun- dance and nutrient flux to the sediments is unmistakable; this suggests that the diatom cells, or perhaps fecal pellets con- taining the cells, are the principal vectors of P and N to the sediments.

Budget calculations and loading models-The annual loading figures in Tables 5-10 provide a good way to test some simple mass-balance models that have been applied to lakes (e.g. Vollen- weider 1969, 1975) as well as some of the common assumptions used to predict lake trophic state from phosphorus load- ing (e.g. Chapra and Tarapchak 1976; Dillon and Rigler 1974; Vollenweider 1976). Lake Washington has gone through a wide range of loading rates within a sin- gle basin, so that there are no bathymet- ric or geographic effects to confound comparisons among them. The basin pro- vides a good opportunity for checking the goodness-of-fit of different assumptions for predicting dynamic responses on an annual basis.

One of the assumptions common to all the models cited above is that the annual flux of P to sedimentary sinks in a lake is a simple function of the mean annual P content of the water. The assumption is based on biological intuition and the knowledge that in many lakes nutrient flux to the sediments is mediated princi- pally by biotic events in the water col- umn (cf. Figs. 8 and 9). If biological pro- duction is fueled by ambient stores of nutrients, then that fraction of production which is lost to the sediments must be a function of them, too.

For Lake Washington, annual flux of P to the sediments can be calculated for different years by the relation:

flux to sediments = input - outflow - A lake-P. (8)

ALake-P-the change in P content of the lake from one year to the next-is taken as the difference between values on I January of successive years, computed by linear interpolation between sampling dates. Calculations show that the expect- ed relationship is upheld, for a very sim- ple reason (Fig. 10, upper panel). There are basically two clusters of points on the graph. One represents the situation in the early and mid-1960s when the burden of P in the lake water was high and so was the annual flux of P to the sediments; the other set of points represents the I97Os, when both lake totals and flux to the sed- iments were lower. The clusters differ significantly in mean values along both axes, and so a significant positive regres- sion results, even though the relation ex- plains only half of the overall variance of the data (r2 = 0.505) and virtually none of the variance within each cluster of points. In fact, flux to the sediments can be predicted much more accurately just from total annual rates of P loading (Fig. 10, lower panel). The significant positive slope of the regression line indicates that on average, before, during, and after its recovery from sewage enrichment, Lake Washington lost 49% (SE = 6.0%, n = 17) of its annual P income to its sedi- ments. This relationship explains 82% of

20 Edmondson and Lehman

A713

I 1 I I t 0 50 100 150 200

ANNUAL MEAN LAKE P (IO3 kg)

I I I , 50 100 150 200

P-LOADING (IO3 kg year-‘)

Fig. 10. Annual flux of P to sediments of Lake Washington calculated from budgets from 1962 to 1978, plotted against mean annual content of total P in the lake (above), and against total annual in- come of P to the lake year-by-year from 1962 to 1978 (below). Regression line in upper panel has slope 0.35 (SE = 0.07) and intercept 17.9 (SE = 8.9); lint in lower panel has slope 0.49 (SE = 0.06) and in- tercept 11.4 (SE = 6.0).

the variance in the data ( r2 = 0.818), and it is not significantly improved in multi- ple linear regressions by taking into ac- count the P content of the lake.

As a result of the empirical relationship between rates of P loading and the ac- cumulation of sedimentary P in Lake Washington, we can make a simple mod- el that predicts P content of the lake water from one year to the next. We begin with the mixed reactor model that Ahl- gren (1977) proposed for Lake Norrviken:

c = (C, - C)Q/v (9)

where C is the concentration of P in the lake water, C, is concentration of P in the inflowing water, Q is hydrologic loading (volume of water per unit time), and V

is lake volume. This treats P as a con- servative element, but Ahlgren thinks that influx and efflux of P from the sedi- ments approximately balance in Lake Norrvikcn, making the treatment appro- priate. In our terms, C, = I/Q, and C = P/V, where 1 is the mass of P added to the lake per unit time, and P is the mass of phosphorus in the lake at any moment. The differential Eq. 9 can be re-expressed a.5

i) = I - P(Q/V),

or as its discrete analog, w-v

P i+l = Ii + Pi( 1 - Qi/V). (11)

For phosphorus, empirical evidence (Fig. 10) shows that a fraction f of the input is lost to the sediments each year. If we replace Ii by Ii(l - f), we have

P it1 = 1i(l -f) + Pi(l - Qi/V) (12)

where Pi is the phosphorus content of the lake on 1 January of successive years (10” kg), Ii is annual input (10” kg. yr-I), and Qi is annual hydrologic loading (10” m3.yr1) corrected for evaporation in the basin (i.e. it is the annual discharge vol- ume). Effective P loading to the water column is reduced by the factor J An analogous treatment was developed by Piontclli and Tonolli (1964) but rejected on intuitive grounds by Vollenweider (1969).

Pi (Q,iV) is the term used to describe outflow of phosphorus from the lake; its validity is established in Fig. 11 where the predictions are plotted against annual outflow of P listed in Table 6. The best- fit regression line is not significantly dif- ferent from a line of slope 1.0 and inter- cept 0, and it explains 95% of the vari- ance. Autocorrelation of hydrologic discharge affects the comparison but merely underscores the point that most of the hydrologic loading to Lake Wash- ington is during winter when the lake is mixed completely from top to bottom, a point that makes our theoretical devel- opment all the more secure. Values pre- dicted by Eq. 12 show good agreement with observed lake P content (Fig. 12). When model predictions are plotted

100

1 g 80

2 k 60 0’

/ 64

MODEL PREDICTION

Fig. 11. Agreement between ohserved ann11a1 outflo\v of P from Lake Washington and prediction\ made 1)~ a simple mixed-reactor model. Line is the- oretical line of 100% agreement between prediction and ol)\ervation. Best-fit regression line has slope 0.929 (SE = 0.057) and intercept -0.3 (SE = 2.8); I-‘) = 0.947.

against the real values, the line of best fit is not significantly different from a line of slope 1.0 and intercept 0 and explains 98% of the variance. We tried several oth- er formulations for model equations (cf. Schindler et al. 1978), but none of them improved upon the relation shown in Fig. 12.

The fact that the P content of the water proves to be a poor predictor for flux of P into the sediments on an annual basis (Fig. 10, upper panel) might at first seem incongruous with the results shown in Fig. 8, because the flux is clearly tied to in situ biological production of plankton cells. What appears to be happening, howeLTer, is that the high rates of nutrient input during winter (Figs. 3 and 4) pro- vide the resources for biological produc- tion during the subsequent spring. Mean annual lake totals have very little to do with it; nutrients that enter the lake are swiftly channeled into productivity, and a fraction of the resulting biomass (and nutrients) is lost to the sediments. The two exceptions were during the floods of 1972 and 1975-1976 when most of the phosphorus entering the lake was partic- ulate and went to the bottom without passing through the biological compo- nent. In all these cases, nonetheless, nu-

2 200 lo 0

- 1

- 2 150

/i//

/ w > A6 r

IOO- a

LLI

:: 1

50- : > a W U-J

E 0 1 I 1

0 50 100 150 200

1(1-f) + LAKE-P(I O/V)

Fig. 12. ‘4grcYmc~nt Iwt\l’wll ol)4er\‘t’d al~llllal mean m;ls\ of‘P in Lake \\‘c1411illgtoll and prcdiction4 made 1,) a simplca loading modc~l clc~\c*ril)ed in te\t. Line is theoretical lint, ot‘ 100% ,iqrc~c~tiiriit I)rt\vcc~ir prediction and ol)\(~r\ ation.

trients which enter the lake are either lost to the sediments or entchr into biotic cycling on a time sc~tle milc~li shorter than a year.

We can ustl this simple model to mimic the actllal dynamic response of Lake Washington to the diversion of sewage’ and also to investigate se\,eral hypothet- ical alternative loading scl~c~~nes. The measured and interpolated mass of total P in the lake on 1 Janual-y of sllccessivc‘ years is plotted in Fig. 13, together with several model predictions. The moclt~l values were generated by recursive so- lutions of Eq. 12, using measured values for annual fluvial discharge, (1 (Table Fj), and for lake volume, V (Taljle 1). For the period 1962-1978, .f‘ = 0.*57 minimiztls the residuals in thcx simulation of’ actllal conditions. The vallle differs from 0.49, the slope fit in Fig. 10, low<>r panel, be- cause the model in E(4. 12 re(lllires that the intercept for the relation be zc’ro (the intercept of the regression compiitcd fol Fig. 10 is in fact not significantly differ- ent from zero, nor is the slope signifi- cantly different from O.Fi7 at the Fi% cbon- fidence level). Some impro\~~~rnent reslllts if f = O.i3Fi from 1962 to 1966, and .f = 0.61 thereafter (Fig. 13, 111,1~‘r panel), l)llt the improvement hralps explain only 22% of the residllal \~ari:~nce, ;i s111;Jl amolllrt

22 Edmondson and Lehman

250 fi

r7 1 I

;2 II 63 1 II 1

64 65 66 II 11 1

67 68 69 11 II l

70 71 72 73 74 75 76 77 7-E

250

1 3 200 Y VI 0

150 LL

e 100 3

-A --7.-*,-y no diversion .-----

-. A /

l -.

--.* \

‘L./“-l-*,-

*-*_ --.

\ ‘**.* diversion only to I966 level ___ L----c.- ““L. \

-. #C -- *. _*----.__ .- .-- -- .___---- _*--

\ A \ complete diversion

L-5‘ A

diversion in I963 IA A

. 1.0

= -0.8 : .-

-5 Q)

-0.6 jjj =

Of 62'63

I I I 1 I I I I I I I I 64 65 66 67'68 69 70 71 72 73 74 75'76'77 78'

Fig. 13. Total P in Lake Washington on 1 January of successive years (A). Above-values calculated by rccllrsive solution of Eq. 12 (see text), with S = 0.57 (solid 1’ lne , or with f = 0.55 before 1967 and f = ) 0.61 afterward (dotted line). Individual annual estimates of ji, the fraction of annual P loading retained in Lake Washington sediments (dashed line) arc compared with retention at steady state. Below-com- parison of actual measurements with simulations of four hypothetical loading schemes for total P.

considering that a constant f explains The second term on the right of Ey. 13 about 99% of the year-to-year variability. has been called “retention” (e.g. Vollen-

Individual annual values for f can bc weider 1976); it is the fraction of P that computed from the data: enters a lake but does not leave with the

outflow. At steady state, Pi - Pi,, = 0,

(13) and f = retention. Annual values for re- tcntion (I - Poul~low/Pi,lTlow) comyutcd from

Nutrients and Lake Washington 23

data in Tables 5 and 6 are plotted in Fig. 13, upper panel, together with annual predictions off from Eq. 13. Three points can he made. First, there is no trend to fi; the fraction of input lost to the sedi- ments was not markedly different before, after, or during diversion, given the year- to-year evident variability. Second, the lake was approximately in equilibrium with its hydrologic and P loading before 1966 and after 1969, but during the pe- riod 1966-1969 the lake was in a tran- sient washout phase. Note that apparent retention is lowered during this transient phase. That may seem to mean that sed- iments are releasing P or that sedimen- tation efficiency has decreased. Neither is the case. The equivalent of more than 50% of the P input during that time con- tinued to be lost to the sediments; the effluent was composed of P already in the water that was flushed from the basin. Under these nonsteady state conditions, changes in the mass of P in the water pre- vent computed retention from accurately documenting the net flux of P to the sed- iments (Eq. 13).

The third point evident from Fig. 13 is that the deviation and return to near- equilibrium represented a relatively short episode in Lake Washington. The time required for the lake to reach a new quasi-equilibrium after a systematic change in the loading regime depends on the variation in fluvial discharge from year to year. For the 19 years listed in Table 5, the mean fraction of lake volume replaced each year (Q/V) is 0.43. Thus the average halving-time to a new equi- librium after systematic perturbation as predicted by Eq. 10 is (In 2)/(Q/V) = 1.6 yr. Because of this relatively rapid re- sponse time, changes in the water column of the lake track changes in the loading regime of the basin very closely.

The rapid response of Lake Washing- ton to changes in nutrient income is fur- ther amplified in the lower panel of Fig. 13 where we contrast four hypothetical alternative loading schemes. The only differences in calculations are the values of I, total P loading. In one case, diver- sion was assumed to have proceeded

only to the level it had reached by 1966. In two other cases, all diversion was as- sumed to have been completed in either 1963 or 1967, rather than progressively between those years as actually hap- pened. Finally, we consider the case of no diversion. For these hypothetical pre- dictions, sewage treatment plants were assumed to continue discharging at the same annual rates that they did when they were actually diverted. This means that no increases in loading rates have been projected and calculated values are therefore conservative, especially in the case of no diversion. The results imply that the P content of the lake was effec- tively in equilibrium with fluvial and at- mospheric inputs by 1970, and that much the same picture would have emerged even if diversion had been accomplished completely during 1967.

Discussion Efforts to learn how the nutrient in-

come of lakes expresses itself in the pro- duction and maintenance of populations, especially phytoplankton, have involved a variety of approaches, from theoretical models to empirical correlation analysis. The empirical approach usually is based on data from many lakes with a variety of nutrient incomes. Evaluation of such re- lations is complicated by the fact that the lakes vary in many additional features that affect the way they use their nu- trients, such as relative depth, proportion of littoral zone, water income, climate and the proportions of the major ions. In the theoretical approach, a set of equa- tions or other computational device is es- tablished to predict a dependent variable when the value of an independent vari- able is altered, and it is intended that the values correspond to events in a real lake. In this respect the theoretical approach offers a more dynamic description of pro- cesses within a lake than does the em- pirical approach, but it is complicated by the fact that quantitative relations must be established empirically among certain formally defined processes. Some of these relations can be obtained from data

24 Li&nondson nncl Lehman

Table 11. Propcrtics of Lake Washington. A-Phosphate-l’, Jumlary-March; I%---total P, anmlal mean; C-total P, 1 January; IS-nitrate-N, Janllary-March; E-total rcdl~ced N, annual mean; F-dissolved inorganic carbon, January-March; C-seston, July-hugllst, top 10 m; II-chlorophyll u, July-August, top 10 m; I-Secchi disk tnmsparcncy, July-August; J-maximlun transparency, July-August; K-minimum transparency, July-August; I,--maximlm~ transparency, whole year. Concentrations are /-mg. liter-’ except seston (mg. liter -‘) ;lnd carbon (PM). Transparency in mctcrs. Concentrations are for the whole lake based on snmplcs at depth weighted for amount of water involved, except those in summer which arc for top 10 m.

YfXr A R c 1) E F C: H I J K I,

1933 1950 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 197 1 1972 1973 1974 1975 1976 1977 1978 1979 1980

7.3 14.0

18.0 - 123 148 -

- - - - 3.1 25.6

15.9 26.8 34.5 34.6 - -

46.1 55.3 56.6 56.2 41.0 37.7 15.7 18.6 14.9 16.6 8.7

11.2 19.2 15.6 10.9 13.5 13.2 19.1 12.6

58.8 58.7 70.3 68.0 66.7 60.7 44.2 29.5 21.4 21.6 17.6 16.1 18.8 19.4 15.8 14.4 17.1 19.6 17.8

- - - - - - - -

61.3 61.5 64.3 62.1 65.5 61.3 39.2 29.8 22.9 23.2 16.7 19.6 20.0 16.7 21.1 17.3 19.0 24.6

- - - - - -

231 295 281 288 - - - -

448 444 464 512 378 423 366 404 351 375 309 265 349 302 292 240 282 300 302

- 354 348 289 300 351 375 279 307 256 269 240 234 224 196 216 204 188 211 199

658 - - -

578 - - - -

5;7 622 63s 786 708 697 682 729 691 680 693 698 683 684 775 760 766 799

- - - -

3.17 3.06

- 2.7 3.0

11.9 12.9 12.4

- -

- 12.2

- - 6.80 31.8 6.2 1 34.8 6.44 41.0 5.29 24.8 4.64 23.9 5.06 14.9 3.27 12.7 2.87 6.8 2.74 8.8 2.20 6.1 2.44 7.2 2.76 4.7 2.37 4.3 2.25 3.9 1.80 4.0 0.64 2.8 1.32 3.0 1.21 1.9

- - - - 3.7 4.0 3.2 - 2.1 2.4 1.6 - 1.6 2.0 1.2 - 2.2 3.2 1.7 4.0 1.9 2.3 l.5 6.9 2.0 2.0 2.0 4.4 2.2 2.4 2.1 - 1.3 1.8 1.1 - 1.1 1.3 0.9 5.0 1.0 1.0 0.9 5.7 1.0 1.2 0.9 4.1 1.0 1.2 0.9 4.4 1.1 1.4 0.8 3.2 1.3 1.5 1.0 3.9 2.2 2.7 1.9 5.2 2.5 3.6 1.9 5.2 2.3 2.6 2.0 5.5 3.5 4.5 2.6 6.3 2.7 3.7 1.8 5.2 3.4 4.5 2.7 6.5 3.5 4.7 2.5 6.4 4.0 5.6 3.2 7.5 5.7 8.2 4.6 10.2 7.6 10.7 4.5 12.9 6.4 7.2 5.3 10.0 7.8 9.3 S.0 13.5

on cultures or other simple systems, but some involve complexities that cannot easily be duplicated in laboratory sys- tems. Also, some of the information necd- ed by the modeler is of littlc general in- terest beyond its specific need in the model and is not likely to bc available without special effort, Therefore, there is good reason to adopt experimental tech- niques in which the nutrient supply to a single lake is varied over a wide range, and the response of that lake is studied (Table 11). Of course there will be year- to-year variations in weather and hydrol- ogy, but that will bc less than among the large number of lakes needed for com- parative study, and the basin morphology is essentially invariant.

Until recently, systematic fertilization studies like those in the Canadian Ex- perimental Lakes Arca have been rare (Schindler 1974), and most have been confined to small lakes (Nelson and Ed- mondson 1955). Knowledge of the rc- sponse of large lakes has come mostly from lakes where the nutrient income was increased by disposal of sewage, raw or treated. An outstanding exception is the fertilization of large lakes on Vancou- ver Island (LeBrasseur et al. 1978; Stock- ner et al. 1980). In a few cases, of which Lake Washington was the first, the re- sponse to a large decrease in nutrient in- come has been observed in detail.

The present study began in 1955, stim- ulated by the first sighting of Oscillutorin

ruhescens, but had been preceded by two studies that described the condition of the lake before major enrichment (Schef- fer and Robinson 1939; Comita and An- derson 1959). Edmondson (1972) dem- onstrated a strong correlation between phosphorus and the abundance of plank- ton that develops during summer, as measllred by its chlorophyll content. In particular, the correlation is high when the concentration of chlorophyll in the top 10 m in summer (July-August) is re- lated to the mean concentration of phos- phate during the previous winter (Janu- ary-March) or to the mean annual concentration of total phosphorus. Simi- lar correlations exist when particulate phosphorus or total cell volume are used instead of chlorophyll and when summer is defined as June-September. We have shown here a close correlation between the phosphorus income and the phos- phorus content of the lake (Fig. 11). It follows that the summer chlorophyll will be closely correlated with phosphorus in- put (Fig. 7).

A number of preliminary reports were published when Lake Washington was still changing in response to the diver- sion of sewage (e.g. Edmondson 1972). In addition, some of the Lake Washing- ton data have been used by other workers in connection with empirical compari- sons of lakes, from either published data or unpublished information in personal communications (e.g. Imboden and G:,ichter 1978; Vollenweider 1976; Chap- ra and Tarapchak 1976; Chen and Orlob 1975; Lorenzen et al. 1976; Rast and Lee 1978). Several different criteria have been used in these papers to describe the condition of the lake and its populations. The two measures of phosphorus most used were concentration of phosphate in the surface water or in the top 10 m dur- ing January-April or better January- March. Phosphate reaches its maximum concentration during those months (Fig. 7) and is a major share of the supply avail- able to the phytoplankton the following summer. In the preliminary reports, the abundance of phytoplankton was judged

mostly l)y the cl~loroph~ 11 ~ol~cc~lltl-~~tioll in the top 10 m during s11111111cbr (Jolly-AII- gust or June-Septt~rnl)~~~); c~hlorophyll has been llsed rathtlr than total cell \,ol- ume or calculated bionnas\ l~~aiise chlo- rophyll data becarnt~ :~vailal,le much sooner than the l~hytol>l~~nktol1 collnts (Edmondson 1979). These different mea- sures of the same properties (e.g. “siiii1- mer” chlorophyll) are strongly correlated but details differ from year to ytlar. ilit> have used the top 10 m l)ecause that is the thickness of the epilimnion dllring much of the s~~mmt’r, where the phyto- plankton is most abundant. When the> lake is mixed, concentrations are nearl> uniform from top to 1)ottoin most of the time and those in the top 10 m thlls rt’l,- resent the whole water coliin~n. Sliilinic~i was defined ;is J11ly-.~llg11st in earlier l,a- pers bt~aiise the ctinpli;tsis was on tlick effect of elltrol,llicatior1 on phytol,lank- ton, and those were> the months of maxi- mum development of blurl-green algar~. Through the entire period of elltrophi- cation, the lake ret;~inetl ;I pronlint’nt spring l~loom of‘ di;itoms \~~hich, clllrirlg the years of enrichment, \V;IS ti~llo~ved 1)~ prominent growth of l)liic~-green ;~lg,‘;1(‘ (Fig. 7).

In general the v:ilues for L:tke \Vash- ington in its various statc>s of’ enriclliiic~ilt fit in well with data from other lakes (\‘()I- lenweider 1976; Lermall 1974; Oglc~sl)y and Schaffner 1978; Bachmann and Jont~s 1974). Iii some relations, thcb points foi Lake Washington lit> some\\7hat above 01 below the regression line for the wholc~ group, for example ii] tlich relatioil l)e- tween chlorophyll and pliosl)horiis (fig. 1: Bachmann and Jones 1974). The tlat:l from Lake W’ashington also agree reason- ably well with thosth from tile fch~~ other lakes that have been stlldic~tl in a similar way after a rediiction in stl\5’;1,ge sricli ;I\ Norvikkcn (Ahlgrc>Il 1977) allcl Aljiisa (Holtan 1979).

In comparing laktbs most in\,estigators have presented the relations 011 ;I tloul)lc~ logarithnlic grid u.hich rninimizcs the subjective appearailce of‘ sc;itter, l)iit tlic variation usually is c~onsitlcr;ll)lt. For cl\-

26 Etlmondson and Lehman

ample, fig. 22 of Rast and Lee (1978) shows chlorophyll as a function of phos- phorus loading. In the middle of the load- ing range, chlorophyll values vary through an order of magnitude. This is rmder- standable, for the lakes vary greatly in their morphology, general chemistry, and surrounding circumstances. Only within single lake districts do the differences among basins seem small enough to ad- mit easy graphical comparisons on arith- metic scales (e.g. Schindler ct al. 1978).

It is worth noting some other sources of scatter in the empirical correlation graphs. Most calculations of mean annual conditions have been made for the cal- endar year, which means that for some comparisons a value of a dependent vari- able will be related to conditions that de- pended partly on events that occurred after that value was generated. This is the case in some of the figures of Edmondson (1972), where, for example, summer chlo- rophyll is related to mean annual P con- centrations. This problem of course does not arise in correlations between summer conditions and those of the previous win- ter. The total phosphorus content of the lake has a pattern of variation slightly dif- ferent from that of winter phosphate (fig. 1: Edmondson 1977~2). In some climates it may be more useful to use the water year (October-September) rather than the calendar year for compiitations. Lake Washington, which does not freeze, re- ceives about 71% of its water income in the months October to March (average for 1970 to 1978). The two major floods in recent years are therefore each divided between two calendar years, 1971-1972 and 19751976 (Fig. 3), influencing the nutrient budgets of 4 years, cvcn though there were only two events.

Variability in correlation graphs can also result from the fact that in most stud- ies, the income of total phosphorus is used but part of that is not biologically available. Much of the phosphorus brought in attached to silt or clay during winter floods will be deposited without becoming involved in the 1, iological cycle. This is shown clearly in Lake

Washington by the two floods, which in- creased the total P income greatly but not that of dissolved phosphorus (Fig. 4). Phytoplankton in the following summers did not increase in proportion to total P. Attempts to define the biologically avail- able phosphorus more exactly have been made by Peters (1977, 1978) and Oglesby and Schaffner ( 1978).

The empirical relations between abun- dance of phytoplankton and nutrient sup- ply have considerable interest in terms of the mechanism of control of phytoplank- ton production and abundance of phyto- plankton. They are also of practical inter- est for programs of lake management. For these reasons, a more detailed analysis of the mechanism is desirable The most obvious thing to consider is the water in- come, since dilution and flushing will af- fect the disposition of the income of phosphorus. The water budgct can be in- corporated into empirical correlation graphs (figs. 7 and 9: Vollenweider 1976) but functional analyses for individual systems are more promising. The primi- tive graphical models of Edmondson (1961) have been replaced by progres- sively more elaborate formulations taking account of the water budget, morphom- etry, and the deposition of phosphorus and its retention by sediments (Vollen- wcider 1975, 1976; Imboden and Gachtcr 1978; Lerman 1974). The mathematical models have focused mainly on predic- tion of the phosphorus regime of the lake (e.g. Lorcnzen 1974), but some of them also try to predict the consequent pro- duction or abundance of phytoplankton.

We have shown that the quantity of I’ in the water of Lake Washington varies on an annual scale as a very simple func- tion of hydrologic discharge and P load- ing. Efforts to simulate exchangeable and unexchangeable pools of sedimentary P (Lorcnzen et al. 1976) are not necessary to explain the grand course of events in the basin from year to year. The com- plexities of nritrient and biotic cycling in the lake emerge, however, when we deal with levels of resolution shorter than the year. Fluxes vary with the character and

Nutrients und Luke Washington 27

abundances of the plankton (Figs. 8 and 9) and probably also with short term changes in the physical or chemical re- gime. The loading models discussed so far are inadequate for dealing with the details and specificities that concern many biological limnologists and popu- lation ecologists. Nor are there other models able to fill the void at present. More detailed models to simulate season- al patterns of plankton dynamics are pro- liferating rapidly (Bierman et al. 1974; Park et al. 1974; Lehman et al. 1975; Ca- nale et al. 1976; Steele and Frost 1977; Scavia 1979) but in general the more de- tailed the predictions attempted, the eas- ier it is to show them in error. Deficien- cies are particularly glaring, for instance, in the treatment of zooplankton (Lehman 1979), which we know to be extremely important for cycling rates of nutrients in Lake Washington (Lehman 1978, 1980; Devol 1979).

The mean values that we have used here to characterize conditions in Lake Washington are static measures of a very dynamic system. To go any further with the analysis will require detailed study of the seasonal changes and the way they have varied over the years, which will be done in connection with an analysis of the data on phytoplankton.

We have not tried to relate the magni- tudes of areal fluxes measured in the sed- iment traps to budget estimates for sev- eral reasons. Sediment deposition is demonstrably not uniform across the ba- sin, and the traps are concentrated at a single site. More important, however, sediment traps provide inexact estimates of particulate fluxes to the sediment sur- face, despite their widespread use (Gard- ner 1980). Particles falling in water are greatly influenced by turbulence because the falling velocities of most particles in water are between one and six orders of magnitude less than velocities resulting from advection and turbulent eddy dif- fusion. Furthermore the retention of par- ticles at the sediment interface depends on cohesive properties of the surface sed- iment which may not be mimicked pre-

cisely in the sediment traps. Particles of different specific gravity or chemical composition may be captured in traps with different efficiencies.

Something of this sort may have been happening in Lake Washington. The ra- tio of N:P (by weight) found in sediment traps from April 1971 to December 1977 was 6.52 (40.60 g N * rn-“:6.23 g P* mpS), whereas the budget estimate for N:P lost inside the lake during the same period was 12.74 (3,532 x 10” kg N:277.2 x 10” kg P). Neither of these estimates agrees with the observed ratio of about 2.0 for N:P accumulating in the surficial sedi- ments of the lake (Shapiro et al. 1971). Part of the reason for this may also be that denitrification reduces the amount of N found in sediment traps and in the sedi- ments. The sink for N includes not just losses to the sediments, but fluxes to the atmosphere as well.

A related problem is that budget esti- mates for nutrient flux sometimes show a net flux out of the sediments (Fig. 5), but sediment traps always show a net flux into the sediments. When we compare areal rates of sedimentation of P and N from budget calculations with areal flux- es to the sediment traps on a month-by- month basis, using data plotted in Figs. 8 and 9, the slopes of the regression lines are 1.095 (SE = 0.202, n = 81) for P and 0.917 (SE = 0.304, n = 81) for N. Neither is significantly different from a slope of 1.0. There is, however, a significant non- zero intercept in the relation for P (-0.042 gem “*mo I, SE = O.OlS), but not for N (0.060 g.rn’*mo*, SE = 0.170). This menns that the sediments of Lake Washington release about 0.04 g P* rn-‘)* mo ’ even at times when there is little or no measurable influx of particu- late debris. We cannot tell whether the efflux changes substantially at other times. Despite our hesitance to relate the seasonal fluxes of nutrients measured in sediment traps to budget estimates by a simple multiplicative factor, the two measures are congruous, as shown by the similar temporal patterns of fluxes plot- ted in Figs. 8 and 9. They illustrate most

28 Edmondson und Lehman

graphically the pronounced seasonality of nutrient flux to the sediments of the lake and show their general concordance with the computed nutrient budgets for N and P.

References

AIILCREN, I. 1977. Role of sediments in the pro- cess of recovery of a eutrophicnted lake, p. 372- 377. Zn II. I,. Golterman [ed.], Interactions be- tween sediments and fresh water. Junk.

BACHMANN, R. W., ANI) J, R. JONES. 1974. Phos- phorus inputs and algal blooms in lakes. Iowa State J. Rcs. 49: 155-160.

BIERMAN, V. J,, Jn., F. II. VEHEIOFF, T. L. POULSON, ANI) M. W. TENNEY. 1974. Multi-nutrient dy- namic models of algal growth and species com- petition in eutrophic lakes, p. 89-109. Zn II. J. Middlebrooks et al. ]eds.], Modeling the eutro- phication process. Ann Arbor Sci.

BIRCII, P. B. 1976. The relationship of sedimen- tation and nutrient cycling to the trophic status of four lakes in the Lake Washington drainage hsin. Ph.D. thesis, Univ. Washington.

CANALE, R. I’., L. M. DI#AI,MA, AND A. II. VOGET,. 1976. A plankton-based food well model for Lake Michigan, p. 33-74. Zn R. I’. Canale [cd.], Modeling biochemical processes in aquatic ecosystems. Ann Arbor Sci.

CIIAP~A, S. C., AND S. J. TAIMPACIIAK. 1976. A chlorophyll (1 model and its relationship to phosphonls loading plots for lakes. Water Re- sour. Res. 12: 1260-1264.

CIIEN, C. W., ANI) G. T. ORI,OK 1975. Ecologic simidation for aqrlatic environments, p. 475- 588. Zn B. C. Patten [cd.], Systems analysis ancl simulation in ecology, v. 3. Academic.

COMITA, G. W., ANL) G. C. ANUEMON. 1959. The seasonal development of a population of Diclp- tornus nshlnndi Marsh, and related phytoplank- ton cycles in Lake Washington. Limnol. Ocean- ogr. 4: 37-52.

DEVOL, A. H. 1979. Respiratory enzyme activity of net zooplankton from two lakes of contrasting trophic state. Limnol. Oceanogr. 24: 893-905.

DILLON, p. J., AND F. I+. RICLER. 1974. A test Oi a simple nutrient budget model predicting the pl~ospl~orus concentration in lakewater. J. Fish Res. Rd. Can. 31: 1771-1778.

~CI)MONI)SON, W. T. 1961. Changes in Lake Wash- ington following an increase in the nutrient in- come. Proc. Int. Assoc. Theor. Appl. Limnol. 16: 153-158.

-. 1969. Eutrophication in North America, p. 124-149. Zn Elltrophication: Causes, conse- qilcnces, correctives. Natl. Acad. Sci. Publ. 1700.

-. 1972. Nutrients and phytoplankton in Lake Washington. Am. Sot. 1,imnol. Oceanogr. Spec. Symp. 1: 172-193.

-. 1977n. Trophic cquilil~rimn in I,nke Wash-

ington. U.S. EPA, Corvallis, Oregon. Rep. EPA- 600/3-77-087.

-. 197717. 1,ake Washington, p. 288-300. Zn North American Project-A stlldy of U.S. water bodies: a report for the organization for eco- nomic cooperation and development. U.S. EPA, Corvallis, Oregon. Rep. EPA-600/3-77- 086.

-. 1979. Lake Washington and the predict- ability of limnological events. Arch. Hydrobiol. Beih. ErgelI. Limnol. 13: 234-241.

GARDNER, W. D. 1980. Sediment trap dynamics and calibration: A laboratory evaluation. J. Mar. Res. 38: 17-39.

GOULD, II. R., AND T. F. BUIXNGER. 1958. Control of sedimentation and 1)ottom configuration by convection currents, Lake Washington, Wash- ington. J. Mar. Res. 17: 183-198.

IIOLTAN, H. 1979. The Lake Mjiisa story. Arch. IIydrobiol. Beih. Ergeh. Limnol. 13: 242-258.

IMISODEN, D. M., AND R. G&IITEI~. 1978. A dy- namic lake model for trophic state prediction. Ecol. Model. 4: 77-98.

JOHNSON, R. l3., A. T. ROSSANO, JR., AND R. 0. SYLVESTER. 1966. Dustfall as a source of water quality impairment. Proc. Am. Sot. Civ. Eng. J. Sanit. Eng. Div. 92: 245-267.

LEBMSSEUR, R. J., AND OTHERS. 1978. Enhance- ment of sockeye salmon (Onchorhynchus ner- k(l) by lake fertilization in Great Central Lake: Summary report. J. Fish. Res. Bd. Can. 35: 1580-1596.

LEIIMAN, J. T. 1978. Aspects of nutrient dynamics in freshwater communities. Ph.D. thesis, Univ. Washington. 180 p.

-. 1979. Grazing, nlltrient release, and theil formulation in plankton models, p. 59-96. Zn Phytoplankton-environmental interactions in reservoirs. U.S. Arrny Corps of Engineers, Waterways Exp. Sta. Rep. TC-3265, DACW39- 78-C-0088.

-. 1980. Nutrient recycling as an interface between algae and grazers in freshwater com- munities. Am. Sot. Limnol. Oceanogr. Spec. Symp. 3: 251-263. UPNE.

-, D. B. BOTKIN, AND G. E. LIKENS. 1975. The assumptions and rationales of a computer mod- el of phytoplankton population dynamics. Lim- nol. Oceanogr. 20: 343-364.

LERMAN, A. 1974. Eutrophication and water qual- ity of lakes: Control by water residence time and transport to sediments. Bull. Sci. IIydrol. 19: 25-33.

TXTTENMAIER, D. P., AND J. E. RICTIEY. 1979. Use of first order analysis in estimating mass lxil- ante errors and planning sampling activities, p. 80-104. Zn E. Ilalfon Ted.], Theory and systems analysis in ecology. Academic.

LORENZEN, M. W. 1974. Predicting the effects of nlltrient diversion on lake recovery, p. 205-210. Zn E. J. Middlebrooks et al. [cds.], Modeling the eutrophication process. Ann Arbor Sci.

, 1). J. SWII-II, AND L. V. KIMMEL. 1976. A long-term phosphorus model for lakes: Appli-

Shifting Regimes and Changing Interactions in the Lake Washington, U.S.A., Plankton Community from 1962–1994 Tessa B. Francis1*, Elizabeth M. Wolkovich2,3¤a, Mark D. Scheuerell4, Stephen L. Katz5¤b,

Elizabeth E. Holmes6, Stephanie E. Hampton2¤c

1 University of Washington Tacoma, Puget Sound Institute, Tacoma, Washington, United States of America, 2 National Center for Ecological Analysis and Synthesis,

University of California Santa Barbara, Santa Barbara, California, United States of America, 3 The Biodiversity Research Centre, University of British Columbia, Vancouver,

British Columbia, Canada, 4 Fish Ecology Division, Northwest Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmospheric

Administration, Seattle, Washington, United States of America, 5 Channel Islands National Marine Sanctuary, National Ocean Service, National Oceanic and Atmospheric

Administration, Santa Barbara, California, United States of America, 6 Conservation Biology Division, Northwest Fisheries Science Center, National Marine Fisheries Service,

National Oceanic and Atmospheric Administration, Seattle, Washington, United States of America

Abstract

Understanding how changing climate, nutrient regimes, and invasive species shift food web structure is critically important in ecology. Most analytical approaches, however, assume static species interactions and environmental effects across time. Therefore, we applied multivariate autoregressive (MAR) models in a moving window context to test for shifting plankton community interactions and effects of environmental variables on plankton abundance in Lake Washington, U.S.A. from 1962–1994, following reduced nutrient loading in the 1960s and the rise of Daphnia in the 1970s. The moving-window MAR (mwMAR) approach showed shifts in the strengths of interactions between Daphnia, a dominant grazer, and other plankton taxa between a high nutrient, Oscillatoria-dominated regime and a low nutrient, Daphnia-dominated regime. The approach also highlighted the inhibiting influence of the cyanobacterium Oscillatoria on other plankton taxa in the community. Overall community stability was lowest during the period of elevated nutrient loading and Oscillatoria dominance. Despite recent warming of the lake, we found no evidence that anomalous temperatures impacted plankton abundance. Our results suggest mwMAR modeling is a useful approach that can be applied across diverse ecosystems, when questions involve shifting relationships within food webs, and among species and abiotic drivers.

Citation: Francis TB, Wolkovich EM, Scheuerell MD, Katz SL, Holmes EE, et al. (2014) Shifting Regimes and Changing Interactions in the Lake Washington, U.S.A., Plankton Community from 1962–1994. PLoS ONE 9(10): e110363. doi:10.1371/journal.pone.0110363

Editor: Elliott Lee Hazen, UC Santa Cruz Department of Ecology and Evolutionary Biology, United States of America

Received March 25, 2014; Accepted September 18, 2014; Published October 22, 2014

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Data Availability: The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are within the Supporting Information files.

Funding: This research was initiated while TBF held a National Research Council Research Associateship Award at NOAA’s Northwest Fisheries Science Center, and conducted in part while TBF was employed by the Puget Sound Institute, funded by the Environmental Protection Agency (Grant #PC-00J303-01). This work was conducted in part while EMW was a postdoctoral associate at the National Center for Ecological Analysis and Synthesis, a Center funded by the National Science Foundation (Grant #EF-0553768), the University of California, Santa Barbara, and the State of California, and in part while she was a National Science Foundation Postdoctoral Research Fellow in Biology (DBI-0905806), and also while she was supported by the NSERC CREATE training program in biodiversity research. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing Interests: The authors have declared that no competing interests exist.

* Email: [email protected]

¤a Current address: Organismic and Evolutionary Biology, Harvard University, Cambridge, Massachusetts, United States of America ¤b Current address: School of the Environment, Washington State University, Pullman, Washington, United States of America ¤c Current address: Center for Environmental Research, Education and Outreach, Washington State University, Pullman, Washington, United States of America

Introduction

One of the most important challenges facing ecologists is

specifying how global change will affect community stability and

the production of associated critical ecosystem services. Commu-

nity stability is mediated by species interactions, which are

sensitive to changing environmental conditions [1,2], and there-

fore estimating the effects of environmental drivers on food web

dynamics is critical for understanding how anthropogenic forces

have altered ecosystems and for anticipating further change [3,4].

Analyzing food web dynamics is complicated in part because the

communities we observe are likely not in ‘‘equilibrium’’ as we

might have once expected [5]. There is increasing evidence that

the structure of communities and the nature of species’ responses

to each other and to their environments are not static, but rather

shift over time. In particular, anthropogenic pressures may be

pushing communities further from equilibrium [6], with commu-

nities exhibiting a variety of non-equilibrium dynamics from

smooth trends to abrupt step changes [7]. Changes in abiotic

conditions of ecosystems can directly and indirectly affect food web

structure [8]. Thus, food web models must account for diverse

temporal changes in community dynamics. In some systems, while

we may have a good understanding of average species interactions

or effects of the environment on food web dynamics over key time

periods, we may still lack important information about whether

PLOS ONE | www.plosone.org 1 October 2014 | Volume 9 | Issue 10 | e110363

http://crossmark.crossref.org/dialog/?doi=10.1371/journal.pone.0110363&domain=pdf

and how such dynamics changed over time in response to large

shifts in the ecosystem.

Lake Washington, U.S.A., is an example of an aquatic

ecosystem that experienced a series of well-described dramatic

changes in its environmental conditions and plankton community

in the mid-20 th

Century. This time period included a regime shift

from one of high nutrient loading from sewage inputs to one of

increased water clarity, as well as temperature and species

abundance changes [9–11]. The lake also experienced shifting

regimes in terms of plankton community dominance. During the

era of high sewage inputs, the lake experienced extensive nuisance

algal blooms, especially of the cyanobacterium, Oscillatoria rubescens. Following sewage diversion, water clarity increased substantially [12]; subsequently, the influential grazer Daphnia established in the lake [11] and Oscillatoria effectively disappeared from the record. In more recent years, warming temperatures

have caused phenological changes in phytoplankton and zoo-

plankton [10,13,14]. What is unclear is how these changes in the

plankton community and abiotic conditions affected interactions

within the food web concomitant with the changing environment.

Such shifts in plankton community interactions – such as

weakening of grazer effects on phytoplankton, or increased

competition among grazing zooplankton guilds – would have

consequences for higher trophic levels in lakes, as plankton

provides an important component of the energetic support for

some lacustrine fish [15], including in Lake Washington [16].

Moreover, plankton community structure and indirect effects of

herbivore-plant interactions can influence fundamental lake

characteristics such as light, temperature and water clarity

[17,18]. In this paper, we introduce an extension of a well-used

static food web model – a multivariate autoregressive (MAR)

model [19–21] – to study Lake Washington’s dynamically

changing food web and ecosystem.

Over the last several decades, multivariate time-series methods

have been used to estimate the strength and pattern of species

interactions and the effect of abiotic drivers on communities

[20,22]. MAR models provide a locally linear approximation of

non-linear stochastic multispecies processes. They have been

particularly useful in aquatic ecosystems and for understanding

plankton dynamics in part because of the tight coupling between

plankton and their environment. MAR models have also become

useful in broader aquatic food web analyses [23,24], as they can

incorporate multiple trophic levels and environmental drivers.

Prior implementations of MAR models have assumed that the

interactions in the study system were unchanging over the time

period encompassed by the data. This approach maximized the

performance of parameter estimation given the properties of

monitoring data, but only estimated the average interaction

strengths over a time series. In contrast, if food web dynamics shift

in response to changing drivers [25], then a better analytical

approach would accommodate and capture this non-stationarity in

modeling the food web. A suite of statistical methods can be

applied to ecological time series to examine non-stationarity – such

as shifts in abiotic conditions or periodicities – through time.

Methods such as wavelets [26,27], single-spectrum [28] and

breakpoint analyses have been used in climatology and paleocli-

matology, and have also recently been applied to ecological data

[29,30]. Such methods allow ecologists to see how abundances

may be shifting [30] or how interactions among species may

change over time in simple lab systems [29], but they do not

provide a cohesive ecosystem approach to examining how

integrated abiotic and biotic forces may change through time. In

particular, food web responses to changes in the strength or nature

of abiotic drivers would be predicted to cause cascading shifts in

the interactions among many members of a food web, and may

also feed back to how community members respond to other

environmental drivers. Examining such a suite of interactions and

drivers, however, would require a model that analyzes all the

variables at once, and that allows estimation of such shifts through

time.

A running or moving window approach is another tool that has

long been used in other disciplines, such as finance, to examine

non-stationarity in time series. In this approach, consecutive and

overlapping subsets of time series – or windows – are analyzed

individually to detect changes through time in a historical record

[31,32]. This approach has recently been used with univariate

autoregressive models to develop leading indicators of regime

change [33–35]. Here we offer an extension of the MAR model,

which we term ‘‘moving-window MAR’’ (mwMAR), and we use it

to examine a case of shifting species interactions and environ-

mental effects on species through time. Our approach blends the

community focus of the MAR model with the moving window

approach of detecting historical changes in time-series data. We

describe the mwMAR model and then apply the model to long-

term monitoring data from Lake Washington, U.S.A., to show

how interactions among dominant taxa of the plankton commu-

nity shifted following sewage diversion. Because food webs show

sensitivity to changes in their abiotic environment [6–8], we

hypothesize that changes in the nutrient status, clarity, and

dominant plankton taxa of the lake would cascade throughout the

plankton food web, resulting in shifts in the direction and strength

of community interactions, which would in turn affect community

stability.

Materials and Methods

Model configuration We estimated interaction strengths among phytoplankton and

zooplankton guilds, environmental effects on phytoplankton and

zooplankton abundance, plankton intrinsic growth rates, and

plankton community stability in Lake Washington from 1962–

1994 using multivariate autoregressive (MAR) models. MAR

models are stochastic models describing changes in species

abundance through time as a function of species interactions

and environmental influences, while accounting for temporal

autocorrelation in species abundances [20,36,37]. MAR models

can also be used to estimate various metrics of community

stability, such as return time to a stationary state following an

ecosystem perturbation, or the distance away from a stationary

state that an ecosystem can be pushed by a perturbation. Previous

work has used MAR models to describe environmental effects on,

and interactions among, lake phytoplankton and zooplankton

[20,22,38,39], effects of climate regime shifts on interactions

among marine plankton [40], causes of estuarine fish declines [24],

and effects of fishing on marine food webs [23]. Extended

descriptions of MAR approaches to time-series data have been

given previously [19,20,37], so we provide only a brief review of

the model structure here.

MAR models are written in matrix form as:

Xt~AzBXt{1zCUtzEt ðEq:1Þ

where, for p interacting species and q environmental covariates, Xt is a p61 vector of species abundances (here, natural log- transformed) at time t; A is a p61 vector of constants, representing intrinsic per-capita growth rates; B is a p6p species interaction matrix, with off-diagonal elements describing inter-specific interac-

tions, and diagonal elements describing intra-specific interactions

Shifting Interactions in the Lake Washington Plankton Community

PLOS ONE | www.plosone.org 2 October 2014 | Volume 9 | Issue 10 | e110363

(i.e., density-dependence); C is a p6q matrix with elements describing environmental effects on species abundance; Ut is a q61 vector of environmental covariates at time t; and Et is a p61 vector of process errors at time t, representing environmental variation not otherwise accounted for in the model. Et is distributed as a multivariate Normal with mean 0 and a diagonal variance matrix g. Elements of B and C typically range from -1 to 1, with distance from 0 representing increasing negative or positive

interaction strength. The diagonal elements of B typically range from 0 to 1, with values closer to 0 representing higher density

dependence.

We also used MAR models to estimate community stability.

Specifically, we estimated the rate at which the system returns to

its stationary distribution following a disturbance by the maximum

eigenvalue of the B matrix (that maximum eigenvalue is henceforth referred to as lambda, l). Systems with values of l closer to 0 are considered to be more stable because they tend to

return to equilibrium conditions faster than systems with values of

l farther from 0 [21]. MAR models estimate mean intrinsic growth rates (captured by

the A vector), community interactions (captured by the B matrix), environmental effects (captured by the C matrix), and community stability (captured by l) across a given time series [20]. Here we use MAR models to quantify changes in interactions through time,

by modeling community interactions for overlapping subsets of a

time series, or moving ‘‘windows’’ of time, thereby estimating

trends in MAR parameters. For a p6n matrix X of time series observations consisting of successive p61 vectors X1, X2,…, Xn, and a moving window of size W,n, we estimated MAR parameters within n-W-1 successive windows. These windows contained data from X2:XW+1, X3:XW+2,…, Xn-W+1:Xn. Note that the time series starts at t = 2 to allow for the lag-1 effect in Eq. 1. The output of the mwMAR analysis is a new time series of

estimated MAR parameters.

Lake Washington data and analysis To investigate changes in interactions among zooplankton and

phytoplankton guilds and the effects of environmental covariates

in Lake Washington through time, we implemented the mwMAR

approach using monthly plankton and environmental data from

Lake Washington (Washington, U.S.A.) spanning 1962 to 1994

(396 timesteps; see Figure S1 for plankton time series). Our 33-

year time series begins in the year of maximum sewage input

(1962) when the lake experienced extensive nuisance algal blooms,

especially of the cyanobacterium, Oscillatoria rubescens. Sewage diversion began the following year (1963), and was completed in

1968. Water clarity increased substantially by 1971 [12] and

continued to improve through 1976, when the influential grazer

Daphnia established in the lake [11] and Oscillatoria abundance decreased dramatically. Despite low abundances at times, and

periods when they were not observe in samples, neither Daphnia nor Oscillatoria ever technically went extinct in Lake Washington. Before they begun to be observed at high abundances in 1973,

Daphnia were observed every year but one (1971). Likewise, after their period of dominance ended in 1980, Oscillatoria continued (and continue) to appear in plankton samples, appearing in all but

3 years between 1980–1994.

The lake has additionally undergone significant warming

throughout the historical record [10], which has altered the

timing of zooplankton abundance cycles [14,41]. Recent work,

however, suggests species and nutrient (phosphorus) shifts related

to the sewage effluent have had a stronger influence on the lake

than shifts associated with warming [42]. These well-documented

shifts in environmental drivers and plankton dynamics make Lake

Washington an ideal ecosystem for evaluating the mwMAR

model’s sensitivity to non-stationary process. Indeed, the dominant

environmental drivers and species interactions in Lake Washing-

ton are well-studied via observational [9,12], experimental [43,44]

and traditional MAR approaches [39,45], offering the necessary

background to build informed community and environmental

interaction matrices (B and C matrices, respectively).

For our analyses we aggregated physical, chemical and plankton

community data, which were collected at various intervals, into

monthly means. Previous analyses of the Lake Washington

plankton community interactions identified a simplified food web

containing species that demonstrated strong roles in structuring

the community [39,45]. We targeted the most strongly-interacting

taxa of this simplified food web with the present analysis, to

determine how the dominant interactions changed through time.

While weak species interactions can be important in structuring

food webs, we chose to focus on the dominant taxa and

interactions as a first test of this new method. These taxa were

pooled into four taxonomic groups: diatoms and green algae –

‘‘DG,’’ both palatable food for grazing zooplankton; Oscillatoria – known to suppress Daphnia [44]; Daphnia; and non-daphnid and non-cladoceran crustaceans – ‘‘NDC,’’ comprised of non-daphnid

cladocerans, Cyclops, and Diaptomus. Group abundance data were log-transformed to better capture non-linearities [20]. A more

complete description of the data is available in Hampton et al.

[39], and the raw data are available in Appendix S1.

We included as covariates in the mwMAR model surface

temperature and total phosphorus, because they were previously

identified as the strongest environmental drivers of plankton

abundance in the lake [39,45]. However, rather than simply use

temperature as a covariate by itself, we instead used the data to

estimate (1) a mean monthly signal indicative of long-term seasonal

forcing, and (2) monthly deviations from the mean to capture

short-term anomalies (e.g., a particularly warm July) or long-term

trends (e.g., an overall increase). To ease comparison of effect sizes

across all environmental covariates, we standardized all covariate

data to a mean of 0 and a standard deviation of 1.

For our environmental covariate matrix (C) we included a priori only biologically meaningful interactions based on established

environmental relationships: we assumed total phosphorus could

not directly affect Daphnia or other zooplankton taxa. We expected shifts in mwMAR coefficients to lag behind known dates

of change in the biotic community, sewage diversion and water

clarity because our moving window size (7 years) is much larger

than the timescale of most known changes. We graphically present

all data at the end year of the moving window; thus, in our figures,

results based on data from 1963–1970 would appear on the x-axis

at year 1970.

Sensitivity analysis The accuracy and precision of parameter estimates by the

mwMAR model, as with other statistical methods, are sensitive to

and affected by multiple factors, including food web configuration

(i.e., the number of interacting species and covariates), window

size, the variance structure of the process errors, and outliers in the

data (see Appendix S2 for discussion and additional model

validation). We conducted several sensitivity analyses to ensure

such factors were not influencing the mwMAR model estimates.

For example, the Lake Washington dataset is of high quality, and

our outlier inspection showed no influence of outliers on the final

results. In addition, because there is a tradeoff between precision

of parameter estimates and accuracy of those estimates that is

defined by window size, we conducted tests using simulated time

series based on the Lake Washington food web configuration, to

Shifting Interactions in the Lake Washington Plankton Community

PLOS ONE | www.plosone.org 3 October 2014 | Volume 9 | Issue 10 | e110363

determine the appropriate window size for analysis of the Lake

Washington dataset. In those simulations, the parameter estima-

tion accuracy decreased sharply at window sizes smaller than 75

time steps (see Figure S1), and therefore we use a window size of

84 (the next factor of 12 larger than 75, given the monthly time

step in the Lake Washington data). We also conducted simulations

to determine what bias, if any, exist in parameter estimates during

periods when a system is undergoing transition between states, for

example between a eutrophic and clear-water state as was the case

with Lake Washington. Last, to ensure that the mwMAR model

was capable of capturing shifts in species interactions and

environmental conditions outside of the Lake Washington case

study, we fit mwMAR models to simulated time series with known

interactions (see Appendix S2).

Statistical programming MAR and mwMAR modeling was done in MATLAB (2007,

The MathWorks), using the open-source program LAMBDA

([46]; freely available from http://conserver.iugo-cafe.org/user/

e2holmes/LAMBDA) with additional programming by the

authors. The coefficients of the A, B and C matrices were estimated using conditional least squares (CLS), and confidence

intervals around each coefficient were established using 2,000

bootstrapped data sets. Each bootstrapped data set was generated

by creating random E matrices and fitting the rest of the parameters using CLS (see [21] for details).

Results

The mwMAR approach revealed changes in interaction

strengths in the Lake Washington plankton community between

1962 and 1994 (Figures 1–5; Figures S2–S3). For example, there

were changes in the effects of Oscillatoria on Daphnia and diatoms and green algae (DG) coincident with the community composition

shift during which Oscillatoria abundance decreased and Daphnia appeared (Figure 1, Table 1). In the period following the first

appearance of Daphnia in Lake Washington, the effect of Oscillatoria on Daphnia became increasingly negative and was strongest in 1976 (Figure 1A). Following the decrease in

Oscillatoria abundance, the negative effect of Oscillatoria on Daphnia weakened, and there was no significant effect of Oscillatoria on Daphnia from late in 1982 until the end of the time series. There was no effect of Daphnia on Oscillatoria (Figure 1B) until after the decline in Oscillatoria and increase in Daphnia. By 1980, the interaction coefficient became negative, weakened in the late 1980s, and returned to neutral after 1990.

Oscillatoria also had a negative effect on DG in the beginning of the time series, and this effect disappeared by the mid-1970s

(Figure 1C).

The effects of Daphnia on other plankton groups in Lake Washington also varied through time (Figure 2). Daphnia had a negative effect on its main food source, DG, starting in the early

1980s, and the effect strengthened until the mid-1980s (Fig-

ure 2A). The effect remained negative, though slightly weaker,

until the end of the time series. The effect of Daphnia on other zooplankton (NDC) also varied through time (Figure 2B). Similar

to the Daphnia-DG interaction, after Daphnia established in Lake Washington, the effect of Daphnia on NDC became increasingly negative, reached its peak in the mid-1980s, then remained

negative but weakened to the end of the time series.

Density dependence also varied through time for all plankton

groups. Density dependence in DG decreased (i.e., the diagonal B matrix coefficient associated with DG increased) until after

Daphnia established in the lake, after which density dependence

Figure 1. Shifting impacts of Oscillatoria on the Lake Washing- ton plankton community. Effects of Oscillatoria on Daphnia (A); Daphnia on Oscillatoria (B); and Oscillatoria on diatoms/green algae, DG, (C) estimated by a mwMAR model using an 84-timestep window (indicated by solid red horizontal line shown in A). The mwMAR- estimated effect of Oscillatoria on NDC was non-significant. Estimates are shown with 95% upper and lower CIs. Grey dotted lines indicate a neutral interaction; solid black lines indicate the average interaction across the full time series, as estimated by a traditional MAR model. The raw time-series data are given in (D), with years of significant known changes shown in shaded vertical bars. All results are presented at the end year of the moving window. doi:10.1371/journal.pone.0110363.g001

Shifting Interactions in the Lake Washington Plankton Community

PLOS ONE | www.plosone.org 4 October 2014 | Volume 9 | Issue 10 | e110363

http://conserver.iugo-cafe.org/user/e2holmes/LAMBDA

increased (Figure 3A). Density dependence in the grazer group

NDC increased steadily from the early 1970s until the late 1980s,

after which it weakened until the end of the time series (Figure 3B).

Daphnia density dependence increased from the time it established in Lake Washington until the end of the time series (Figure 3C).

Oscillatoria density dependence weakened until its decline in abundance in the late 1970s, at which point it increased and held

more or less steady from the early 1980s until the end of the time

series (Figure 3D).

The MAR model also estimates density-independent intrinsic

population growth (the A vector), and while many of the confidence intervals surrounding the A estimates overlapped zero for a portion of the time series, there were consistent trends in the

estimates among different plankton groups (Figure 4). For all four

plankton groups, there were three distinct periods of intrinsic

growth rate estimates: (1) before regular appearances of Daphnia

Figure 2. Shifting effects of Daphnia on the Lake Washington plankton community. Effects of Daphnia on diatoms and green algae, DG, (A) and non-daphnid cladocerans and non-cladoceran crustaceans, NDC, (B) through time as estimated by a mwMAR model with an 84-timestep window (indicated by solid red horizontal line in A). Estimates are shown with 95% upper and lower CIs. Grey dotted lines indicate coefficient values of 0; solid black lines indicate the average interaction across the full time series, as estimated by a traditional MAR model. The raw time-series data are given in (C), with years of influential known changes shown in shaded vertical bars. All results are presented at the end year of the moving window. doi:10.1371/journal.pone.0110363.g002

Figure 3. Shifting density-dependent effects of all plankton groups. Coefficients are estimated by a mwMAR model with an 84- timestep window (indicated by solid red horizontal line in A). Estimates are shown with 95% upper and lower CIs. DG = diatoms and green algae; NDC = non-daphnid cladocerans and non-cladoceran crusta- ceans. Grey dotted lines indicate coefficient values of 0; solid black lines indicate the average effect across the full time series, as estimated by a traditional MAR model. All results are presented at the end year of the moving window. doi:10.1371/journal.pone.0110363.g003

Shifting Interactions in the Lake Washington Plankton Community

PLOS ONE | www.plosone.org 5 October 2014 | Volume 9 | Issue 10 | e110363

in Lake Washington (pre-summer 1973); (2) between when

Daphnia began to make regular appearances, and when Daphnia established in the lake and Oscillatoria declined dramatically (summer 1973 – spring 1976); and (3) after the rise of Daphnia and decline of Oscillatoria (summer 1976 onward). During the first period, there was high variability and negative trends in all A estimates. During the second period, the DG growth rate was

mostly constant (Figure 4A), both grazer groups growth rates

increased (though some CIs overlapped 0; Figure 4B, C), and the

Oscillatoria growth rate decreased (Figure 4D). During the final period, from the mid-1970s to the end of the time series, the

growth rates of most groups were constant, except for an increase

in the DG growth rate. Both Oscillatoria and NDC had growth rates equal to zero during this period.

Stability (l) decreased sharply from the beginning of the time series, and the system was least stable (i.e., l was at its maximum value) in the early 1970s (Figure 5). Following this nadir,

community stability increased and reached maximum stability

(i.e., the lowest l value) at the end of the time series. Following bootstrapping, mean temperature had significant effects on all

plankton groups. In contrast, very few effects of temperature

anomalies or total phosphorus on plankton groups in Lake

Washington were retained in the final mwMAR model (Table 1;

Figure S3).

We assessed the fit of the best mwMAR model to the Lake

Washington data, and found that fewer than 1% of correlations

between model residuals and data were significant. We also tested

the model assumption of normally-distributed errors by applying

the Shapiro-Wilk test [47] to the residuals of the MAR fit to each

data window (E, from Equation 1), with a Bonferroni-corrected alpha [48] to account for multiple null hypotheses. We rejected the

null hypothesis of normally distributed errors in 65/312 windows

for Daphnia, and in 217/312 windows for Oscillatoria (and in 0 windows for DG and NDC; Figure S4). These data windows for

which the null hypothesis was rejected corresponded to periods in

the time series when the abundance of each species was zero, i.e.,

the long one-sided tails in the data.

Figure 4. Intrinsic growth rates of Lake Washington plankton groups. Growth rates are estimated by a mwMAR model with an 84- timestep window (indicated by solid red horizontal line in A). Estimates are shown with 95% upper and lower CIs. Grey dotted lines indicate coefficient values of 0; solid black lines indicate the average rate across the full time series, as estimated by a traditional MAR model. DG = diatoms and green algae; NDC = non-daphnid cladocerans and non- cladoceran crustaceans. All results are presented at the end year of the moving window. doi:10.1371/journal.pone.0110363.g004

Figure 5. Shifting community stability. Stability is given by l, the maximum eigenvalue of the community interaction matrix, as estimated by a mwMAR model using an 84-timestep window (indicated by solid red horizontal line). Estimates are shown with 95% upper and lower CIs. The grey dotted line indicates coefficient value of 0; the solid black line indicates the average stability across the full time series, as estimated by a traditional MAR model. Results are presented at the end year of the moving window. doi:10.1371/journal.pone.0110363.g005

Shifting Interactions in the Lake Washington Plankton Community

PLOS ONE | www.plosone.org 6 October 2014 | Volume 9 | Issue 10 | e110363

Discussion

Shifting plankton dynamics in Lake Washington We hypothesized that the mwMAR would show shifts in the

interactions among the major taxa corresponding roughly with

known periods of change in Lake Washington (e.g., years of and

around 1968–1971, and 1976). For example, it has long been

hypothesized that the highly-abundant Oscillatoria, owing to its low palatability, inhibited Daphnia before Daphnia’s increase in Lake Washington in 1976 [11], during the period of time when the

two species overlapped but Oscillatoria abundance was decreasing. These dynamics have been demonstrated experimentally [44], but

our results are the first to corroborate this hypothesis using

historical data. During the period of time between the peak in

water quality (1971) and the dramatic increase in Daphnia abundance (1976) – the period of overlap between Oscillatoria and Daphnia and hypothesized inhibition of Daphnia by Oscillatoria – we found an increasingly negative effect of Oscillatoria on Daphnia. Once the mwMAR window included only dates following the large increase in Daphnia (i.e., 1976 and later), there was no detectable effect of Oscillatoria on Daphnia. The long, filamentous shape of Oscillatoria generally makes it inedible for Daphnia, which is one likely source of the negative per-capita effect estimated here during their period of overlap.

Oscillatoria also had a negative effect on diatoms and edible green algae, the main food source for Daphnia and other grazers in the lake. High intrinsic growth rates in edible phytoplankton

estimated at the start of the time series decreased during the period

when Oscillatoria was dominant. At the same time, density dependence in diatoms and green algae also decreased, suggesting

inhibition in growth, possibly resulting from competition for

limiting nutrients, or physical shading or toxic effects of excretions

by Oscillatoria. Such inhibition of algae by Oscillatoria has also been demonstrated experimentally [44]. This apparent inhibition

of phytoplankton by Oscillatoria rapidly decreased following an abrupt transition in the mid-1970s when the negative effect of

Oscillatoria on DG decreased and disappeared (Figure 1). Coincident with these dynamics, the effect of Oscillatoria on Daphnia also weakened and the intrinsic growth rate of Daphnia increased from its minimum in 1972 to its peak in 1976

(Figure 4C). After 1976, Daphnia’s intrinsic growth rate decreased and density dependence increased (Figure 3C) as the Daphnia population increased in abundance. In addition, while the result

was not significant (95% CIs overlapped zero), DG may have had

a bottom-up positive effect on Daphnia after being freed from inhibition by Oscillatoria, in the latter half of the time series (Figure S2). Taken together, these results corroborate the

hypothesis that the establishment of Daphnia following the improvement of water quality in Lake Washington was impeded

directly and indirectly by the cyanobacterium Oscillatoria. Grazers are known to inhibit cyanobacteria under some

environmental conditions [49], and our analysis found a negative

effect of Daphnia on Oscillatoria coincident with Oscillatoria’s decrease in abundance. In general, the frequency of cyanobacteria

blooms is associated with the relationships between grazers and

edible phytoplankton, such that when grazers and edible phyto-

plankton dynamics are stable (i.e., abundances do not undergo

large, intrinsic oscillations), cyanobacteria are controlled by grazers

[49]. These dynamics are often associated with phosphorus inputs to

a lake. We observed a similar pattern in Lake Washington. As

phosphorus inputs decreased, the grazing effect of Daphnia on edible phytoplankton increased concomitant with the inhibiting

effect of Daphnia on Oscillatoria (Figures 1 and 2).T a

b le

1 .

In te

ra c

ti o

n ra

n g

e s

e st

im a

te d

b y

th e

m w

M A

R m

o d

e l

fo r

L a

k e

W a

sh in

g to

n a

c ro

ss a

ll 8

4 -t

im e

st e

p w

in d

o w

s o

f d

a ta

in th

e h

is to

ri c

a l

ti m

e se

ri e

C o

m m

u n

it y

In te

ra c

ti o

n s

C o

v a

ri a

te E

ff e

c ts

D G

N D

C D

a p

h n

ia O

sc il

la to

ri a

S e

a s

o n

T e

m p

e ra

tu re

P h

o s

p h

o ru

D G

0 .1

5 –

0 .7

0 2

0 .2

8 –

0 .0

4 2

0 .5

2 –

0 .0

5 2

0 .9

2 –

1 .2

1 2

1 .2

5 –

0 .1

0 2

2 .0

9 –

0 .6

(2 0

.0 9

, 0

.8 9

) (2

0 .4

0 ,

0 .1

8 )

(2 0

.8 9

, 0

.2 0

) (2

2 .7

5 ,

2 .7

4 )

(2 2

.3 3

, 1

.2 3

) (2

3 .9

9 ,

1 .4

9 )

N D

C 0

.0 0

– 0

.1 8

0 .3

9 –

1 .0

5 2

0 .2

4 –

0 .0

0 2

0 .0

8 –

0 .0

9 2

1 .5

3 –

1 .7

0 2

0 .5

9 –

0 .7

3 n

(2 0

.1 2

, 0

.2 8

) (0

.2 2

, 1

.1 9

) (2

0 .3

0 ,

0 .0

5 )

(2 0

.1 7

, 0

.1 8

) (2

0 .7

7 ,

0 .8

0 )

(2 1

.2 3

, 1

.2 5

)

D a

p h

n ia

2 0

.3 2

– 0

.4 6

0 .0

8 –

1 .4

6 2

0 .0

1 –

0 .6

6 2

0 .8

2 –

0 .2

9 )

2 1

.3 8

– 2

.7 2

2 1

.8 8

– 2

.1 4

n /a

(2 0

.6 7

, 0

.9 1

) (2

0 .6

5 ,

2 .1

0 )

(2 0

.2 1

, 0

.8 0

) (2

1 .1

8 ,

0 .6

6 )

(2 4

.4 0

, 5

.3 9

) (2

3 .6

1 ,

4 .5

3 )

O sc

il la

to ri

a 2

0 .4

4 –

0 .2

1 2

0 .6

4 –

0 .6

9 2

0 .4

3 –

0 .1

5 0

.2 0

– 0

.8 4

2 0

.6 3

– 3

.6 0

2 2

.0 2

– 1

.4 5

2 2

.1 2

– 1

.6 6

(2 0

.9 1

, 0

.4 8

) (2

1 .1

6 ,

1 .4

3 )

(2 0

.6 8

, 0

.2 7

) (0

.0 2

, 0

.9 7

) (2

4 .1

8 ,

6 .7

4 )

(2 4

.3 7

, 4

.2 2

) (2

4 .9

6 ,

3 .9

6 )

E m

p ty

c e

ll s

re fl

e c

t in

te ra

c ti

o n

s n

o t

re ta

in e

d b

y th

e m

o d

e l.

P a

re n

th e

ti c

a l

v a

lu e

s re

p re

se n

t ra

n g

e s

o f

9 5

% c

o n

fi d

e n

c e

in te

rv a

ls e

st im

a te

d fo

r 2

,0 0

0 b

o o

ts tr

a p

p e

d d

a ta

se ts

. C

o e

ff ic

ie n

ts re

p re

se n

t e

ff e

c ts

o f

v a

ri a

b le

s in

c o

lu m

n s

o n

v a

ri a

b le

s in

ro w

s. D

G :

D ia

to m

s/ G

re e

n a

lg a

e ;

N D

C :

N o

n -d

a p

h n

id c

la d

o c

e ra

n s

a n

d n

o n

-c la

d o

c e

ra n

c ru

st a

c e

a n

s. n

/a in

d ic

a te

s a

c o

e ff

ic ie

n t

th a

t w

a s

a p

ri o

ri e

x c

lu d

e d

fr o

m th

e m

o d

e l.

d o

i:1 0

.1 3

7 1

/j o

u rn

a l.p

o n

e .0

1 1

0 3

6 3

.t 0

0 1

Shifting Interactions in the Lake Washington Plankton Community

PLOS ONE | www.plosone.org 7 October 2014 | Volume 9 | Issue 10 | e110363

MAR coefficients have been shown previously to reflect changes

in community dominance, when an increase in the abundance of

one species or group coincides with a decrease in another [40],

and therefore the negative effect of Daphnia on Oscillatoria may represent shifting dominance between the two taxa. The transition

from Oscillatoria to Daphnia dominance was reflected in interactions among other plankton groups in the community. As

the negative effect of Oscillatoria on Daphnia declined in the mid- 1970s through to the early 1980s, and as Daphnia increased in abundance, Daphnia had stronger impacts on their main food source (DG) and competitors (NDC; Figure 2). At the same time,

the strength of density-dependence (Figure 3) and density-

independent growth rates increased for the grazing zooplankton

groups (Figure 4), suggesting the release of the grazer community

from inhibition by Oscillatoria. No previous work has shown an effect of Oscillatoria on other grazer groups beyond Daphnia, but the increase in population growth rates (A matrix elements) of the NDC group following Oscillatoria’s decline suggests a possible negative interaction.

The negative effects of Daphnia on their food and competitors weakened towards the end of the time series, apart from an

intensified grazing effect of Daphnia on DG at the very end. One potential explanation for the weakened grazing effect at the end of

the time series relates indirectly to the warming of the lake during

this time. Between 1962 and 2002, the lake surface temperature

increased by 1.4uC during the stratified months, and associated with this warming was an advance in the spring phytoplankton

bloom by 19 days [10]. Most of the warming and spring bloom

advance occurred in the period 1962-1994. The weakening of the

effect of total Daphnia on the phytoplankton group during that period, in the present analysis, could be a reflection of shifts in

species-specific phenology and grazing characteristics [14,41].

The results presented here highlight opportunities to learn more

from time series data about how species interactions shift with

changes in the environment across ecosystem types, and how those

changing food web dynamics are liable to affect community

stability and resilience to further disturbance. Ecosystem-based

approaches to management often include a focus on food web

dynamics, but quantifying changes in species interactions, and

how those changes map onto the environmental template, proves

difficult. Linking shifts in species interactions to specific environ-

mental drivers opens opportunities to focus efforts aimed at

retaining resilience as ecosystems undergo rapid change.

Community stability and environmental covariates The Lake Washington system has undergone major shifts in

chemistry and ecology that are reflected in community stability.

The peak of instability occurred in April 1973, (a window that

included data from May 1966 – April 1973; Figure 5). Values of l greater than 1 indicate an unstable system [21], and here l exceeded 1 for windows ending in November 1970 – November

1974, representing the period of time between December 1963 –

November 1974, inclusive. This is the time period that included

major ecosystem shifts: high nutrient levels, sewage diversion, and

nutrient reduction; high Oscillatoria abundance followed by its decline; and the first rare appearances of Daphnia. By the time of Oscillatoria’s disappearance, maximum water clarity, and estab- lishment of Daphnia in 1976, the community stability was increasing, and continued to increase until the end of the time

series. Thus, the period of time the lake was undergoing the most

substantial and dramatic shifts throughout the ecosystem, and

before Daphnia gained a foothold, was the least stable period in the community as well.

We observed effects of monthly mean temperature on the

abundance of all plankton guilds (Figure S3), which agrees with

previous MAR analyses [39,45], and with the MAR model

estimated here from the whole Lake Washington time series (Table

S1). Previous work has suggested that Lake Washington plankton

phenology also responds to lake warming [10,50], and that the

relationships between temperature and plankton taxa are evidence

of the potential influence of a warming lake on food web dynamics

[39]. However, we found no significant effects of deviations from

the long-term seasonal temperature patterns, suggesting that lake-

warming effects are not detectable in the abundance of these

plankton guilds.

Caveats and considerations Our results suggest moving-window MAR models may be useful

in systems with sufficient time-series data for understanding

shifting abiotic and biotic dynamics. As with all statistical methods,

however, practitioners must consider possible caveats and issues in

advance of and throughout analyses. The data and ecosystem

considerations applicable to prior MAR model applications also

extend to our moving-window approach. Users must have

sufficient time-series data for valid parameter estimation, which

varies depending on the time scale of interactions in the system

and frequency of observations. The moving-window MAR model

imposes the further consideration of having sufficient time-series

data for multiple windows and surrounding the event(s) of interest.

Importantly, bias in model estimates shrinks as the ratio between

window size and system transition period increases, and users are

cautioned to interpret model estimates during system transitions

with consideration of such bias. However, the window could be

configured for different purposes: made smaller to detect changes

before they occur, or sized to optimize detection of a change in a

particular state variable.

Applications of this method will benefit from a priori knowledge of ecological interactions and drivers in the modeled system to

build a robust MAR model. In our analysis of the Lake

Washington plankton community, we simplified the plankton

community based on previous work that highlighted the strongest

food web interactions and key environmental covariates [39].

However, Hampton et al. [39] also pointed out the importance of

other plankton taxa in driving the dynamics of the dominant

species in Lake Washington, such as Cryptomonas, picoplankton and non-colonial rotifers. Therefore, it is possible that additional

food web dynamics contribute to the interaction coefficients

observed here, which could be highlighted by future analyses.

Furthermore, if the model failed to include an influential

environmental driver of Lake Washington plankton dynamics,

the model results might be erroneously interpreted: if one plankton

guild responds negatively to an unmeasured environmental

variable, and another guild responds positively, this might

incorrectly be interpreted as a negative interaction between the

two guilds. In the Lake Washington case, years of experimental

work and field observations have identified environmental

variables that are robust driving signals. In addition, preliminary,

exploratory MAR model runs were performed to screen a broad

suite of potential drivers on plankton time series data. The analyses

here rely heavily on those two approaches to validation, and

potential users are advised to similarly behave as ecological

detectives.

Additionally, as with prior MAR approaches, users must invest

time in simulation modeling that allows them to test how the

approach is likely to work with data similar to theirs. Simulation of

data from a model with similar parameters to the study ecosystem

helps identify the appropriate moving window size and, thus,

Shifting Interactions in the Lake Washington Plankton Community

PLOS ONE | www.plosone.org 8 October 2014 | Volume 9 | Issue 10 | e110363

estimate the precision associated with future predictions of system

change. Because much of the MAR approach is based on iterative

fitting approaches, creating and testing simulation data sets from

known parameter values with similar lengths, covariate and taxa

numbers, and variance, is critical to interpreting knowledge gained

from MAR models. For the moving-window approach, users

should carefully examine the effect of window size on their

simulation datasets (see Appendix S2 for an example analysis using

simulated datasets). A priori knowledge or hypotheses related to the resolution of data and interactions as well as the strength and

timing of the predicted shift should be considered during the

process of simulation modeling. Comparison of the mwMAR

output with whole time-series MAR estimates is useful in assessing

when the broad confidence intervals estimated with the mwMAR

model are potentially masking significant interactions.

Conclusions

Ecologists have recently gained an appreciation for the need to

develop methods based on the underlying hypothesis that many

systems are rarely, if ever, stationary. Here we present a method that

allows researchers and managers alike to examine long-term

monitoring data and develop a dynamic record of shifting

interactions and drivers. By calculating indirect and direct effects

over time, and their changes, mwMAR allows researchers to

understand how species invasions and extinctions, shifts in temper-

ature and nutrient loadings, and other anthropogenic perturbations

may cascade and feedback through food webs and ecosystems.

Supporting Information

Figure S1 Lake Washington plankton densities from 1962–1994. Monthly means of densities for aggregated plankton groups used in mwMAR analyses. NDC = non-daphnid

cladocerans; DG = diatoms and green algae.

(DOCX)

Figure S2 Time series of all community interactions. Interaction coefficients estimated for the Lake Washington time

series with a mwMAR model, using an 84-month window. Figures

show per-capita effects of plankton guilds in columns on plankton

guilds in rows. Diagonal figures represent self-effects, or density-

dependent effects on abundance.

(DOCX)

Figure S3 Time series of environmental covariate effects. Interaction coefficients estimated for the Lake Washing- ton time series with a mwMAR model, using an 84-month

window. Figures show the effects of covariates in columns on

plankton guilds in rows.

(DOCX)

Figure S4 Quantile-quantile plots of residuals for the Daphnia and Oscillatoria time series. Shown are theoretical versus observed distributions of mwMAR model residuals for all

windows where the Shapiro-Wilk test statistic was below the alpha

value required to reject the null hypothesis of normally-distributed

errors (61/1248 for Daphnia, 175/1248 for Oscillatoria, 0 for DG and 0 for NDC).

(DOCX)

Table S1 Community and covariate matrix coefficients estimated by a MAR model for the full Lake Washington time series.

(DOCX)

Appendix S1 Lake Washington plankton and covariate data, 1962–1994.

(CSV)

Appendix S2 Moving-window MAR Model Testing. Validation of the moving window MAR model approach,

including accuracy of parameter estimation and estimation of

bias during system transition.

(DOC)

Acknowledgments

Many thanks to J. Regetz for assistance with coding. We thank D.E.

Schindler for generous access to the Lake Washington data, the numerous

people who have contributed to the data set, including A. Litt and S.

Abella, and the organizations that have financially supported it, including

the Mellon Foundation. The manuscript was improved by comments from

D.E. Schindler, E.J. Ward and two anonymous reviewers.

Author Contributions

Analyzed the data: TF EW MS SK. Contributed reagents/materials/

analysis tools: TF EW MS SK EH. Wrote the paper: TF EW MS SK EH

SH.

References

1. Burkle LA, Marlin JC, Knight TM (2013) Plant-Pollinator Interactions over 120

Years: Loss of Species, Co-Occurrence, and Function. Science 339: 1611–1615.

2. Tylianakis JM, Didham RK, Bascompte J, Wardle DA (2008) Global change

and species interactions in terrestrial ecosystems. Ecol Lett 11: 1351–1363.

3. Holyoak M (2000) Habitat subdivision causes changes in food web structure.

Ecol Lett 3: 509–515.

4. Schlesinger MD, Manley PN, Holyoak M (2008) Distinguishing stressors acting

on land bird communities in an urbanizing environment. Ecology 89: 2302–

2314.

5. Hastings A (2004) Transients: the key to long-term ecological understanding?

Trends Ecol Evol 19: 39–45.

6. Harley CDG, Hughes AR, Hultgren KM, Miner BG, Sorte CJB, et al. (2006)

The impacts of climate change in coastal marine systems. Ecol Lett 9: 228–241.

7. Walther GR (2010) Community and ecosystem responses to recent climate

change. Philos Trans R Soc B-Biol Sci 365: 2019–2024.

8. Tunney TD, McCann KS, Lester NG, Shuter BJ (2014) Effects of differential

habitat warming on complex communities. Proceedings of the National

Academies of Science 111: 8077–8082.

9. Edmondson WT (1970) Phosphorus, nitrogen, and algae in Lake Washington

after diversion of sewage. Science 169: 690–691.

10. Winder M, Schindler DE (2004) Climatic effects on the phenology of lake

processes. Glob Change Biol 10: 1844–1856.

11. Edmondson WT, Litt AH (1982) Daphnia in Lake Washington. Limnol

Oceanogr 27: 272–293.

12. Edmondson WT (1991) The Uses of Ecology. Lake Washington and Beyond:

University of Washington Press. 329 p.

13. Winder M, Schindler DE, Essington TE, Litt AH (2009) Disrupted seasonal

clockwork in the population dynamics of a freshwater copepod by climate

warming. Limnol Oceanogr 54: 2493–2505.

14. Hampton SE, Romare P, Seiler DE (2006) Environmentally controlled Daphnia

spring increase with implications for sockeye salmon fry in Lake Washingon,

USA. J Plankton Res 28: 399–406.

15. Francis TB, Schindler DE (2009) Shoreline urbanization reduces terrestrial

insect subsidies to fishes in North American lakes. Oikos 118: 1872–1882.

16. Beauchamp DA, Sergeant CJ, Mazur MM, Scheuerell MD, Scheuerell JM, et al.

(2004) Spatial-temporal dynamics of early feeding demand and food supply for

sockeye fry in Lake Washington. Transactions of the American Fisheries Society

133: 1014–1032.

17. Mazumder A, Taylor WD, McQueen DJ, Lean DRS (1990) Effects of fish and

plankton on lake temperature and mixing depth. Science 247: 312–315.

18. Sarnelle O (1993) Herbivore effects on phytoplankton succession in a eutrophic

lake. Ecol Monogr 63: 129–149.

19. Ives AR (1995) Measuring resilience in stochastic systems. Ecol Monogr 65: 217–

233.

20. Ives AR, Dennis B, Cottingham KL, Carpenter SR (2003) Estimating

community stability and ecological interactions from time-series data. Ecol

Monogr 73: 301–330.

Shifting Interactions in the Lake Washington Plankton Community

PLOS ONE | www.plosone.org 9 October 2014 | Volume 9 | Issue 10 | e110363

21. Ives AR, Gross K, Klug JL (1999) Stability and variability in competitive

communities. Science 286: 542–544. 22. Hampton SE, Holmes EE, Scheef LP, Scheuerell MD, Katz SL, et al. (2013)

Quantifying effects of abiotic and biotic drivers on community dynamics with

multivariate autoregressive (MAR) models. Ecology 94: 2663–2669. 23. Lindegren M, Mollmann C, Nielsen A, Stenseth NC (2009) Preventing the

collapse of the Baltic cod stock through an ecosystem-based management approach. Proc Natl Acad Sci U S A 106: 14722–14727.

24. Mac Nally R, Thomson JR, Kimmerer WJ, Feyrer F, Newman KB, et al. (2010)

Analysis of pelagic species decline in the upper San Francisco Estuary using multivariate autoregressive modeling (MAR). Ecol Appl 20: 1417–1430.

25. Kordas RL, Dudgeon S (2011) Dynamics of species interaction strength in space, time and with developmental stage. Proc R Soc B-Biol Sci 278: 1804–1813.

26. Torrence C, Webster PJ (1999) Interdecadal changes in the ENSO-monsoon system. Journal of Climate 12: 2679–2690.

27. Menard F, Marsac F, Bellier E, Cazelles B (2007) Climatic oscillations and tuna

catch rates in the Indian Ocean: a wavelet approach to time series analysis. Fisheries Oceanography 16: 95–104.

28. Jevrejeva S, Moore JC, Grinsted A (2004) Oceanic and atmospheric transport of multiyear El Nino-Southern Oscillation (ENSO) signatures to the polar regions.

Geophys Res Lett 31.

29. Beninca E, Johnk KD, Heerkloss R, Huisman J (2009) Coupled predator-prey oscillations in a chaotic food web. Ecol Lett 12: 1367–1378.

30. Winder M, Cloern JE (2010) The annual cycles of phytoplankton biomass. Philosophical Transactions of the Royal Society B: Biological Sciences 365:

3215–3226. 31. Hsieh CH, Chen CS, Chiu TS, Lee KT, Shieh FJ, et al. (2009) Time series

analyses reveal transient relationships between abundance of larval anchovy and

environmental variables in the coastal waters southwest of Taiwan. Fisheries Oceanography 18: 102–117.

32. Breaker LC (2006) Nonlinear aspects of sea surface temperature in Monterey Bay. Prog Oceanogr 69: 61–89.

33. Carpenter SR, Cole JJ, Pace ML, Batt R, Brock WA, et al. (2011) Early

warnings of regime shifts: A whole-ecosystem experiment. Science 332: 1079– 1082.

34. Dakos V, Carpenter SR, Brock WA, Ellison AM, Guttal V, et al. (2012) Methods for detecting early warnings of critical transitions in time series illustrated using

simulated ecological data. PLoS One 7: e41010. 35. Seekell DA, Carpenter SR, Pace ML (2011) Conditional heteroscedasticity as a

leading indicator of ecological regime shifts. American Naturalist 178: 442–451.

36. Elkinton JS, Healy WM, Buonaccorsi JP, Boettner GH, Hazzard AM, et al.

(1996) Interactions among gypsy moths, white-footed mice, and acorns. Ecology

77: 2332–2342.

37. Ives AR (1995) Predicting the response of populations to environmental change.

Ecology 76: 926–941.

38. Hampton SE, Izmest’eva LR, Moore MV, Katz SL, Dennis B, et al. (2008) Sixty

years of environmental change in the world’s largest freshwater lake - Lake

Baikal, Siberia. Glob Change Biol 14: 1947–1958.

39. Hampton SE, Scheuerell MD, Schindler DE (2006) Coalescence in the Lake

Washington story: interaction strengths in a planktonic food web. Limnol

Oceanogr 51: 2042–2051.

40. Francis TB, Scheuerell MD, Brodeur RD, Levin PS, Ruzicka JJ, et al. (2012)

Climate shifts the interaction web of a marine plankton community. Glob

Change Biol 18: 2498-2508.

41. Winder M, Schindler DE (2004) Climate change uncouples trophic interactions

in an aquatic ecosystem. Ecology 85: 2100–2106.

42. Law T, Zhang WT, Zhao JY, Arhonditsis GB (2009) Structural changes in lake

functioning induced from nutrient loading and climate variability. Ecological

Modelling 220: 979–997.

43. Edmondson WT (1991) Sedimentary record of changes in the condition of Lake

Washington. Limnol Oceanogr 36: 1031–1044.

44. Infante A, Abella SEB (1985) Inhibition of Daphnia by Oscillatoria in Lake Washington. Limnol Oceanogr 30: 1046–1052.

45. Hampton SE, Schindler DE (2006) Empirical evaluation of observation scale

effects in community time series. Oikos 113: 424–439.

46. Viscido SV, Holmes EE (2010) Statistical modelling of communities and

ecosystems using the LAMBDA software tool. Environmental Modelling and

Software 25: 1905–1908.

47. Shapiro SS, Wilk MB (1965) An analysis of variance test for normality (complete

samples). Biometrika 52: 591–611.

48. Hochberg Y (1988) A sharper Bonferroni procedure for multiple tests of

significance. Biometrika 75: 800–802.

49. Carpenter SR (1992) Destabilization of planktonic ecosystems and blooms of

blue-green algae. In: Kitchell JF, editor.Food Web Management, A Case Study

of Lake Mendota.New York: Springer New York. pp. 461–481.

50. Hampton SE (2005) Increased niche differentiation between two Conochilus

species over 33 years of climate change and food web alteration. Limnol

Oceanogr 50: 421–426.

Shifting Interactions in the Lake Washington Plankton Community

PLOS ONE | www.plosone.org 10 October 2014 | Volume 9 | Issue 10 | e110363

Edmondson P_N in LW after diversion.pdf

Contents

Issue Table of Contents

Science, New Series, Vol. 169, No. 3946 (Aug. 14, 1970), pp. 625-708

Front Matter [pp. 625-634]
Letters

Fire Ant: Whose Pest? [p. 630]
Computers as Chess Partners [pp. 630-631]
Revising the Publication Process [p. 631]

Science: Attack and Defense [p. 633]
The Structure of Ordinary Water [pp. 635-641]
A Multiple Origin for Plastids and Mitochondria [pp. 641-646]
Aboriginal Drained-Field Cultivation in the Americas [pp. 647-654]
News and Comment

Behaviorial Sciences: The View at the Center for Advanced Study [pp. 654-658]
Academic Finance: British System Smoothly Functions in 50th Year [pp. 658-660]
News in Brief [p. 659]
The Global Environment: M.I.T. Study Looks for Danger Signs [pp. 660-662]
Appointments [p. 662]
Recent Deaths [p. 662]

Book Reviews

Review: Private Thoughts of Lyell on Progression and Evolution [pp. 663-664]
Review: Almost There [pp. 664-666]
Review: Regulating Man [p. 666]
Review: Nociperception and Analgesia [pp. 666-667]
Review: Properties of Ores [pp. 667-668]
Books Received [pp. 668-669]

Reports

Giant Radioactive Halos: Indicators of Unknown Radioactivity? [pp. 670-673]
Monosodium Glutamate: Lack of Effects on Brain and Reproductive Function in Rats [pp. 673-674]
Sterols in Recent Marine Sediments [pp. 674-676]
Oil Spills: Method for Measuring Their Extent on the Sea Surface [pp. 676-678]
Microwave Detection of Thioformaldehyde [pp. 679-680]
A Search for the 1$_{10}\leftarrow $ 1$_{11}$ Transition of Interstellar Thioformaldehyde [pp. 680-681]
Formaldehyde Absorption Coefficients in the Vacuum Ultraviolet (650 to 1850 Angstroms) [pp. 681-683]
Amnesia Produced by Electroconvulsive Shock or Cycloheximide: Conditions for Recovery [pp. 683-686]
Fitness of an Escherichia coli Mutator Gene [pp. 686-688]
Lesch-Nyhan Syndrome: Preventive Control by Prenatal Diagnosis [pp. 688-689]
Phosphorus, Nitrogen, and Algae in Lake Washington after Diversion of Sewage [pp. 690-691]
Bone Marrow Colonies: Stimulation in vitro by Supernatant from Incubated Human Blood Cells [pp. 691-692]
RNA and DNA Puffs in Polytene Chromosomes of Rhynchosciara: Inhibition by Extirpation of Prothorax [pp. 692-694]
Circadian Rhythms in Human Heart Homograft [pp. 694-696]
Cell-Mediated Immunity Shown by Lymphocytes from the Respiratory Tract [pp. 696-697]
Crayfish Swimming: Alternating Motor Output and Giant Fiber Activity [pp. 698-700]
Epileptic Focus Location: Spectral Analysis Method [pp. 701-702]
Drinking and Eating Elicited by Cortical Spreading Depression [pp. 702-704]
Hyperbaric Oxygen [pp. 704-705]
Polywater Discovered 30 Years Ago? [p. 705]

Back Matter [pp. 706-708]

Schindler ELA 1974.pdf

Contents

image 1
image 2
image 3

Issue Table of Contents

Science, Vol. 184, No. 4139, May 24, 1974, pp. 829-928

Front Matter [pp. 829-920]
Letters

Laser Isotope Separation [p. 849]
Disruptions at AAAS Meetings [p. 849]
Japanese Conception of Nature [pp. 849-853]
Oncogene Theory [p. 853]

Maintaining a Pluralistic Society [p. 855]
Natural Marine Oil Seepage [pp. 857-865]
Chromatin Structure: Oligomers of the Histones [pp. 865-868]
Chromatin Structure: A Repeating Unit of Histones and DNA [pp. 868-871]
Budget and the National Cancer Program [pp. 871-875]
News and Comment

Britain: A Touch of Austerity for Research and Universities [pp. 876-878]
Low Marks for AEC's Breeder Reactor Study [p. 877]
Science and Crime: Engineers Claim a Rosy Outlook, but Police Aren't Sure [pp. 878-881]
Briefing [p. 880]
Airlines: Half-Empty Planes Keep Profits Low, Waste Fuel [pp. 881-884]

Research News

Energy Storage (II): Developing Advanced Technologies [pp. 884-887]
The Finite Element Method: A Mathematical Revival [pp. 887-889]

Book Reviews

Geological Changes [pp. 890-891]
Protozoology as Cell Biology [p. 891]
Protein Anomalies [pp. 891-892]
Books Received [pp. 892-921]

Reports

Continental Pleistocene Climatic Variations from Speleothem Age and Isotopic Data [pp. 893-895]
Elemental Mercury Evolution Mediated by Humic Acid [pp. 895-897]
Eutrophication and Recovery in Experimental Lakes: Implications for Lake Management [pp. 897-899]
Opaline Sediments of the Southeastern Coastal Plain and Horizon A: Biogenic Origin [pp. 899-901]
Jovian Atmosphere: Structure and Composition between the Turbopause and the Mesopause [pp. 901-903]
Purgatorius, an Early Paromomyid Primate (Mammalia) [pp. 903-905]
Holocene Stratigraphy and Archeology in the Middle Missouri River Trench, South Dakota [pp. 905-908]
Elephant Seals: Genetic Variation and Near Extinction [pp. 908-909]
Carnosine in the Primary Olfactory Pathway [pp. 909-911]
Illusory Correlation of Brightness Enhancement and Transients in the Nervous System [pp. 911-913]
Maternal Lymphocytes: Suppression by Human Chorionic Gonadotropin [pp. 913-914]
Pollution in Coastal Waters [pp. 914-915]
Conditioning or Control? [pp. 915-916]

Back Matter [pp. 921-928]