Oceanogr., 41(5), 1996, 1124-l 135 0 1996, by the American Society of Limnology and Oceanography, Inc.

Limnol.

Simulated long-term temperature and dissolved oxygen characteristics of lakes in the north-central United States and associated fish habitat limits H. G. Stefan St. Anthony Falls Laboratory, Department of Civil Engineering, University of Minnesota, Minneapolis 554 14

AI. Hondzo School of Civil Engineering, Purdue University, W. Lafayette, Indiana 47907

X. Fang Department of Civil Engineering, Lamar University, Beaumont, Texas 777 10

J. G. Eaton and J. H. McCormick U.S. Environmental Protection Agency, ERL, Duluth, Minnesota 55804 Abstract Water temperatures and dissolved oxygen (DO) concentrations in lakes are related to climate. Temperature and DO in 27 lake classes (3 depth classes x 3 surface area classes x 3 trophic states) were simulated by numerical models with daily weather data input. The weather data used are from the 25-yr period 19551979. The lakes and the weather are representative of the north-central U.S. Daily profiles ofwater temperature and DO concentrations were computed and several temperature and DO characteristics extracted from this information base. Temperature and minimum oxygen requirements for good growth of cold, cool, and warm water fish were then applied to determine the length of the good-growth periods and the relative lake volumes available for good growth. All characteristics are presented in graphical form using lake surface area, maximum lake depth, and Secchi depth as independent variables. The surface area and maximum depth were combined in a lake geometry ratio which is a relative measure of the susceptibility of a lake to stratification; Secchi depth was retained as a measure of lake transparency and trophic state. To determine an effect of latitude, we investigated a southern and northern region separately. The effect of climate change due to a projected doubling of atmospheric CO, was investigated by applying the output from the GISS 2 x CO, global circulation model to the lake models.

of projecting characteristics of similar lakes for which only surface area, maximum depth, and midsummer Secchi depth are known. This procedure permits the comparative study of many lakes over long periods of time. Parameters characterizing long-term averages of water temperature, DO, and associated fish habitat in lakes are presented for climate conditions that existed from 1955 to 1979. The results are from simulations for the openwater season and include the typical summer stratification of lakes in the north-temperate region. They are presented in a form that allows interpolation and quantitative comparison among lakes with different morphometries and trophic levels. For better appreciation of the results, a brief description of the model formulations, input data requirements, and accuracy is also provided. The models have been validated with actual data from a wide range Acknowledgments of lakes with different morphometries, trophic levels, and The investigation described herein was conducted for the U.S. meteorological conditions. Environmental Protection Agency/OPPE in cooperation with The results are for two regions covering latitudes from the Environmental Research Laboratory, Duluth, as a part of a 46’10’ to 49” (northern region) and 43”30’ to 46’10’ project on climate change effects on fisheries. The Minnesota (southern region). The southern region has a mean avSupercomputer Institute, University of Minnesota, provided a erage annual temperature - 4°C warmer than the northern resource grant and access to its CRAY2 supercomputer. region. The effect of a doubling of atmospheric CO2 on Eville Gorham and Joseph Shapiro provided valuable sugthe simulated lake characteristics was also investigated. gestions that considerably improved the manuscript. 1124 The following is an analysis of recently completed simulations that project long-term average water temperature and dissolved oxygen conditions in lakes of the northcentral U.S. and their suitability for fish habitat. Minnesota was selected for this study because it has many valuable lakes, an extensive lake database, and is located in the center of the continent. It is also at a latitude where climate change may have the greatest impact on aquatic ecosystems. Baseline water-quality simulations were made with historical records of meteorological parameters known to influence the temperature and the dissolved oxygen (DO) of north-temperate lakes. We used processoriented, deterministic numerical modeling which, when sufficiently calibrated and validated, offers the possibility

Long-term temperature and oxygen The output of the Goddard Institute for Space Studies (GISS), Columbia University global circulation model, was used to specify the future climate scenario. The GISS 2 x CO2 model predicts a 3.8”C air temperature increase for the northern Minnesota region.

--,

1125

Lake classltication by - surface area - maxlmum depth - Secchi depth

relation Area - r 0epn-d I

4il

Simulation methods for water quality and fish habitat A deterministic, process-oriented, unsteady, one-dimensional model of lake-water quality was used for these simulations. This model has been successfully applied to simulate hydrothermal processesand water quality in lakes for a variety of meteorological conditions (e.g. Ford and Stefan 1980a; Riley and Stefan 1987; Hondzo and Stefan 199 1, 1993a). The water-quality model was verified for 16 lakes of different morphometries and trophic levels, and with different meteorological conditions (Hondzo and Stefan 19933; Stefan et al. 1993; Stefan and Fang 1993, 1994). The standard error of the water temperature predictions was from 0.5 to 1. 1°C for individual lakes. The standard error of the predicted DO values was from 0.6 to 2.3 mg liter- l. With these water-quality models, simulations of water temperatures and DO were made for 27 classes of lakes in the southern and northern Minnesota regions. The 27 lake classes are derived from the product of 3 depth classes x 3 surface area classes x 3 trophic classes. The computational sequence is graphically summarized in the flow-chart of Fig. 1. The simulation model was operated on daily timesteps. Simulated vertical profiles of daily water temperatures and DO concentrations were obtained for the open-water season, the length of which was found by the model itself (Hondzo and Stefan 199 1). Southern Minnesota lakes are usually ice free from April to November and northern Minnesota lakes from May to October. The presence of fish in a lake is, in general, related to accessibility, ecological suitability, human interference, and resistance to episodic natural events (Fry 1971). Among the environmental factors limiting survival and growth of fish, temperature and DO concentrations are considered the two most significant (Coutant 1987; Christie and Regier 1988; Magnuson et al. 1990). In this study, the suitability of habitats for cold water, cool water, and warm water fish was assessedin terms of these two factors only. The concept is briefly illustrated in Fig. 2. The results indicate which lakes have temperature and DO conditions suitable for survival and good growth of fish. For this purpose, fish having similar thermal requirements were grouped as thermal guilds (cold water, cool water, and warm water). Actual fish observations in 3,002 Minnesota lakes were compared with simulated suitable fish habitat based on water temperatures and DO concentrations. Good agreement between fish observations and numerical simulations of fish habitat was found (H. G. Stefan et al. UMN-SAFL Proj. Rep. 347).

the lake surface and hydrologic inputs from the lake basin. Solar radiation and atmospheric longwave radiation heat the water column; evaporation and back radiation cool it. Convective heat transfer driven by the temperature difference between water and air can also warm or cool a lake. The differential radiative heat absorption throughout the lake depth causes thermal stratification. The stronger the stratification, the more quiescent the water body. Wind exerts a drag force on the surface of the lake which, through a variety of external and internal wave motions tends to vertically mix the stratified water column (partially or completely). The external mechanical energy input from the wind is opposed by the potential (buoyant) energy “locked” in the stratification. The stronger the stratification, the more mechanical energy is needed to mix the water column. The water temperature model simulates these processesby a set of deterministic equations which is described in detail elsewhere (H. G. Stefan et al. UMN-SAFL Proj. Rep. 347).

Water temperature and thermal stratification simulation -A lake is exposed to meteorological forcing through

Dissolved oxygen concentration simulation -The major components affecting the DO concentration in the lake

Fig. 1. Schematicillustration of the simulation procedure.

Stefan et al.

1126

a> B 4d Mar

b)

Apr

Jun

Jul

Aug

Sep

Ott

Nov

Dee

Ott

Nov

Dee

Dissohred oxygen (mg liter“)

Mar

c>

May

Apr

May

Jun

Jul

Aug

Sep

Fishhabltat quality

Mar

Apr

May

Uninhabttable

Jun

IsI

Jul

Aug

Good

Sep

growth

Ott

Nov

0

Dee

Restricted

growth

Fig. 2. Schematic illustration of the distribution across time and depth of water temperature isotherms, dissolved oxygen (DO) isopleths, and those isotherms and DO isopleths which are considered for the survival and growth of a fish species in a seasonally stratified lake. LGGT-lower good-growth temperature limit; UGGT-upper good-growth temperature limit; LT-lethal temperature.

are plant respiration, photosynthesis, biological and sedimentary oxygen demand (largely microbial respiration and chemical oxidation), and surface-layer oxygenation. DO transfer at the water surface and photosynthesis can increase DO concentrations in the water column. The sedimentary oxygen demand (SOD), biochemical oxygen demand (BOD), and plant and microbial respiration (R), are DO sinks in the water column. The nitrification process is omitted because it makes a minor contribution to the state of oxygen in the lake as a whole (Stefan and Fang 1994). Fish habitat projection - The effects of water temperature on freshwater fish have long been recognized (e.g. Hokanson 1977; Coutant 1972; Magnuson et al. 1979; Meisner et al. 1987). An overview of the temperature effect literature and upper thermal tolerance limits for cold, cool, and warm water species was given by Eaton et al. (1995). Similarly, low DO concentrations limit the survival of fish. Water temperature and DO place severe physical constraints on fish habitat in freshwater lakes.

Other factors such as food, predation, episodic effects, and human interference are also important. Herein, water temperature and DO levels are the only two parameters used to determine fish habitat. Each parameter value is an average calculated from the simulation of 25 yr of daily water temperature and DO profiles in a lake. This procedure finds substantial justification in the extensive and well-documented studies of fish thermal biology (Eaton et al. 1995). Temperature criteria for fish habitat were developed from laboratory and field data as described by Eaton et al. (1995), and the guild (cold water, cool water, warm water) designations for various species are as suggested by Hokanson (1977). Temperature ranges for good growth of these three thermal guilds, which comprise a total of 28 species were 9.0-18.5”C for cold water, 16.3-28.2”C for cool water, and 19.7-32.3”C for warm water species. The upper survival temperature limits used for each of these guilds were 23.4”C for cold water, 30.4”C for cool water, and > 30.4”C for warm water species. Fish survival and good-growth temperature criteria were related to simulated daily water temperatures and DO concentrations as shown schematically in Fig. 2. Three isotherms were chosen for each guild; the lethal temperature (LT) threshold, the upper good-growth temperature limit (UGGT), and the lower good-growth temperature limit (LGGT). DO limits of 2.5 mg liter-l for warm water fish and 3.0 mg liter-l for cool water and cold water fish were based on the U.S. Environmental Protection Agency water quality criteria document (U.S. EPA 440/5-86-003). The isopleth that designates the critical DO survival value is also shown in Fig. 2. Between the lines in Fig. 2, three habitat qualities can be identified: uninhabitable space where the temperature is above or the DO is below the survival or threshold limit for seven consecutive days; good-growth habitat if the temperature is between the upper and lower good-growth limits and the DO is above survival limit; and restricted-growth habitat if the temperature is above the upper good-growth limit and below the upper survival limit, or if the temperature is below the lower good-growth limit (in all casesthe DO must be above the survival limit).

Databases Meteorological data- The meteorological database used as input to the long-term lake simulations consisted of that for the 25 yr from 1955 to 1979. The meteorological data file contained average measured daily values for air temperature, dew point temperature, precipitation, wind speed, and solar radiation. The period from 1955 to 1979 was chosen because it is long enough to give a representative average and variance on recent conditions in the chosen study area. Data from the Minneapolis-St. Paul International Airport (44.5 3”N, 93.13”W) and Duluth (46. 50°N, 92.1 low) were used for the southern and northern regions. Numerical values for the mean monthly meteorological parameters for these stations are summarized in Table 1.

1127

Long-term temperature and oxygen Table 1. Mean monthly meteorological data for Minneapolis-St. Paul and Duluth, Minnesota, 25-yr averages ( 1955-l 979). Air temperature (“C)- AT; dew point temperature (“C)DT; solar radiation (cal cm-2)- SR; wind speed (m s- l)- WS. Duluth Mar Apr May Jun July Aw Sep Ott Nov Mean SD

AT -4.7 2.9 9.9 15.0 18.4 17.4 12.4 7.1 -2.2 8.5 7.9

DT -9.8 -3.4 2.6 9.1 12.8 12.5 7.9 2.0 -6.3 3.0 7.7

Minneapolis-St. Paul SR

WS

344.8 417.2 471.5 504.4 517.0 441.3 317.3 211.5 125.9 372.3 127.1

5.1 5.6 5.3 4.6 4.3 4.2 4.6 4.9 5.2 4.9 0.4

Lake data-The Minnesota Lakes Fisheries Database (D. Schupp pers. comm.; B. Goodno pers. comm.) contains lake survey data for 3,002 lakes. The database includes 22 physical variables and all common fish species. Nine primary variables explain 80% of the variability among lakes. These nine variables include surface area, volume, maximum depth, alkalinity, Secchi depth, lake shape, shoreline complexity, percent littoral area, and length of growing season. Geographic subdivision of the lakes was approached in a variety of ways. First ecoregions (Omernick 1987) were considered, but found to give too detailed a picture. An ecoregion is defined as a region with homogeneous trophic, geologic, vegetative, and landuse features. Then the entire set was considered as a regional entity but that was rejected as too large because of the diversity of climate. Finally, division of the lakes into a northern and southern set was considered appropriate because there is a significant difference in geological, hydrological, climatological, and ecological parameters between the northern and southern half of the state (Baker et al. 1985; Heiskary et al. 1987; D. Schupp MNDNR Invest. Rep. 417; Hondzo and Stefan 1993b). To develop a classification, we selected lake surface area, maximum depth, and mean Secchi depth as the

AT -1.3 7.7 14.6 19.9 22.8 21.5 15.8 9.8 0.5 12.4 8.3

DT -6.9 0.2 6.6 12.8 15.9 15.0 10.0 3.9 -3.5 6.0 7.7

SR

WS

333.6 396.4 477.2 528.3 546.6 469.9 358.4 250.8 139.3 388.9 126.5

4.9 5.4 5.0 4.5 4.1 4.0 4.2 4.5 4.8 4.6 0.4

main independent lake parameters. These three parameters were chosen because the first two have a direct relationship to stratification dynamics of lakes. Secchi depth was chosen because it is correlated with lake transparency, trophic state, and biochemical and biological oxygen demand. Lake surface area (surrogate for wind fetch) and maximum lake depth are used as indicators to differentiate between seasonally stratified and seasonally polymictic lakes (Lathrop and Lillie 1980; Niirnberg 1988; Gorham and Boyce 1989; Demers and Kalff 1993). The criterion for lake stratification given by Gorham and Boyce (1989), mainly for the mid-North American continent (for lakes of surface area 2.9) are expected to be seasonally polymictic according to the seasonal stratification criteria given above. The rest of the lakes are expected to be dimictic. Solar radiation and wind also play an important role

Table 2. Physical parameters used to define 27 Minnesota lake classes.

Key parameter Max depth, H,,,,, (m) Surface area, A, (km2) Secchi depth, 2, (m)

Descriptive term Shallow Medium Deep Small Medium Large Eutrophic Mesotrophic Oligotrophic

Lake class Representative value used Range 4.0 13.0 24.0 0.2 1.7 10.0 1.2 2.5 4.5

14.0 4.1-20.0 20.1-45.0 10.4 0.5-5.0 5.1-40.0 51.8 1.9-4.5 4.6-7.0

Cumulative frequency Lower 30% Central 60% Upper 10% Lower 30% Central 60% Upper 10% Lower 20-50%

Central 20-50% Upper 0- 10%

Lake geometry AS 0.2 0.2 0.2 1.7 1.7 1.7 10.0 10.0 10.0

H max

4.0 13.0 24.0 4.0 13.0 24.0 4.0 13.0 24.0

4 : Kmx

5.3 1.6 0.9 9.0 2.8 1.5 14.1 4.3 2.3

1128

Stefan et al.

OLIGOTROPHIC 20

25

30

MESOTROP”, 40

35

EUTROPHIC 504

45

55

HYPEREUTR0Pf-K 65

60

70

75

60

TROPHIC STATE INDEX 10

15

87

6

514

3

21

1.5

0.5

1

0.3

TRANSPARENCY (rn)

21

0.5

CHLOROPHYLL hwb)

3

4

5

71

10

15

20

30

40

60

60 100

150

a

7

ld

15

40

50

60

80

100

150

TOTAL P (ppb)

Fig. 3. Carlson’s trophic state index related to several other Minnesota lake parameters (aftei Heiskary and Wilson 1988).

in lake stratification but are less variable from lake to lake in a given region than is lake morphometry (Ford and Stefan 1980a). Secchi depth was chosen as a lake trophic state indicator (Heiskary and Wilson 1988) because it is a commonly available parameter and can be related to direct indicators of primary productivity through Carlson’s trophic state index (Carlson 1977) (Fig. 3). Secchi depth or transparency also affects solar radiation attenuation and oxygen balance. Combinations of the three values for the three key parameters defined 27 (3 x 3 x 3) lake classes for which simulations were made (Table 2). Representative area-depth relationships were obtained from 3 5 lakes in a set of 122 lakes. Separate empirical equations were fit to the data for the three lake sizes and used in the simulations as representative area-depth relationships.

Results and discussion Water temperature and thermal stratijkation characteristics-parameters which can be used to characterize water temperatures and the thermal stratification in a lake include surface temperature, bottom temperature, volume-weighted (avg) temperature, surface-to-bottom temperature difference, thermocline depth, dynamic stratification stability, linear temperature gradient, seasonal stratification ratio, and heat content. Values of these parameters were estimated from the daily water temperature profiles simulated over a 25-yr span. Each parameter is correlated with lake geometry ratio (Aso.25 : H,,,) and Secchi depth. The results were plotted in a series of graphs with lake geometry ratio and Secchi depth as axes (H. G. Stefan et al. UM-SAFL Proj. Rep. 352). Samples of lake water-quality characteristics with a direct relation to lake biology are presented herein. The reader is reminded that all values are 25-yr average values obtained by model simulations. They cannot be compared to instantaneous measurements in a lake unless the measurements span an equally long period and are averaged.

The surface- water temperature isotherms plotted in Fig. 4 represent maximum daily surface-water temperatures and are obtained by interpolation among 27 simulated values more or less uniformly distributed over the graph (Table 2 gives the coordinates). As can be seen, the surface temperatures are about the same over a wide range of lake morphometries and Secchi depths, but there is a noticeable difference between north and south (i.e. regions with different meteorological conditions). One can conclude that maximum surface-water temperatures are influenced primarily by the meteorological forcing and much less significantly by lake geometry and trophic state. Daily average meteorological variables that correlate most significantly with maximum surface-water temperatures and which are different in the north and the south are air temperature, dew-point temperature, and solar radiation. There is a difference of - 1°C between lakes with a permanent seasonal stratification (dimictic lakes) and intermittently stratified lakes (polymictic lakes). The transition from one type of lake to another occurs at As0*25 : -2.5-5.0 m-o.5 as indicated by Gorham and Boyce (79%). Surface temperature is given here because it is important to growth kinetics and survival of plants, fish, and other organisms in lakes. The water temperature at the bottom of a lake, unlike surface-water temperature, is not directly related to the meteorological forcing, especially in deeper lakes. The bottom temperature reported is the maximum daily water temperature 1 m or less above the lake sediments. It is the temperature that can affect sediment oxygen demand or fish survival. The simulated daily maximum bottom temperatures given in Fig. 5 increase strongly when As0.25: H maxincreases. This increase is related to the gradual transition from dimictic (seasonally stratified) lakes to polymictic lakes. The transition occurs when the geometry ratio exceeds a value of -4 (Table 2). The transition is gradual as the strength of the stratification diminishes gradually with higher geometry ratio, going from seasonally stratified (dimictic) lakes to polymictic lakes with shorter and shorter stratification periods. Constant water

Long- term temperature and oxygen 5.0

I

I ,

I

I

0.8 1.0

2.0

4.0

I

I

nor h

’

I ‘III

o 0.6 5.0

I

,

I

(

south

6.0

’

1129

8.010.0

0.6

20.0 5

1’1’1

'i

111, 0.8 1.0 I,',

- south

I 2.0

I 4.0

1

I

I

8

I

I

1'1'1 6.0 8.010.0

20.0

1’1’1

2-

l-

0 0.0

(

0.6

5 , I ,

0.8 1.0 A

I

I

2.0

4.0

S

0.25

’ H max

0

0.6

1’1’1

6.0

8.010.0

20.0

(m-09

Fig. 4. Maximum daily surface-water temperature (“C) isotherms (simulated 2%yr averages),

temperatures at A,o.25: Hmax > 8 indicate independence from lake geometry (i.e. daily mixed lakes). There is probably also an asymptotic value of 4°C at As0.25: Hmax< 0.4. Oligotrophic lakes have higher bottom-water temperatures than eutrophic lakes because of the difference in solar radiation attenuation. Similar water temperature trends are evident for northern and southern lakes. In seasonally stratified lakes, hypolimnetic water temperatures are mainly determined by temperatures at spring turnover. In summary, the maximum bottom-water temperature is highly correlated with the lake geometry ratio. The d@erence between surface and bottom temperature of a lake is an indicator of the strength of stratification in a lake. Values of maximum daily water temperature differences between the surface and the bottom are given elsewhere (H. G. Stefan et al. UMN-SAFL Proj. Rep. 352). This temperature difference increases as the lakes become more strongly stratified (i.e. as A,o.25: Hmax decreases). The largest difference is estimated for lakes with the greatest depth and the smallest surface area. These lakes typically have more wind sheltering. The strength of the temperature stratification also depends somewhat on the trophic state of a lake. Eutrophic lakes (with high solar radiation attenuation with depth) have higher temperature differences than oligotrophic lakes.

,,I, 0.8 1.0

I 2.0 A

0.25 . H max s ’

4.0

1’1’1

6.0

8.010.0

20.0

(m-09

Fig. 5. As Fig. 4, but of daily bottom-water temperature.

An average water temperature (weighted by volume) in the entire lake was also determined from the simulated water temperature profiles. The maximum of the average daily water temperatures is lower for stratified (dimictic) lakes. The volume-averaged lake temperatures are higher, by &4.0°C, in southern polymictic lakes compared to northern polymictic lakes (geometry ratio >5). The difference drops to 3°C and 2°C when geometry ratios fall below 4 and 1, respectively (i.e. stratification diminishes differences in mean temperatures between lakes in different regions). Deep oligotrophic lakes have a higher average temperature (i.e. heat content per unit volume) than do eutrophic lakes of similar depth. This is because they warm to greater depth than do their less transparent counterparts. The thermocline depth is another measure of stratification. It is estimated from the water density gradient profile (Patterson et al. 1984). In most cases,thermocline depth corresponds to the position of the maximum density gradient. Values for the maximum daily thermocline depth relative to maximum lake depth are of interest. A value of 1.O designates a fully mixed lake. Thermocline depth decreases with decreasing As0.25: H,,,, and the trophic state of a lake contributes to the thermocline depth. For the same lake geometry ratio, oligotrophic lakes have a greater thermocline depth than eutrophic lakes. This is

1130

Stefan et al.

4-

3“m 0 2-

l-

o I,), 0.6 0.8 1.0 5,

0)

20.0

1

0.6

1 ,I,

5

,I,

0.8 1.0 I/I,

I

I

2.0

4.0

I

I

8

10’1

6.0 1

I’I’I

8.010.0

south

20.0

i

43 N 0 0

:

5 3 CL g r

0 0 6

2-

_ -

l0;

0.6

8 ,I,

0.8 1 .O

I

2.0 A S 0.25 : H mc-Jx

I

4.0

~1’1’1

6.0

8.010.0

I

20.0

Cmmo5)

Fig. 6. Length (days) of seasonal stratification (simulated 25yr averages).

because eutrophic lakes attenuate solar radiation more readily than similar oligotrophic lakes. Seasonal stratijcation is defined herein as the condition when the temperature difference between surface and deep water is > 1°C. Although 1°C is an arbitrary criterion, it is useful to identify variations of stratification with lake geometry, trophic state, and geographic location (Fig. 6). As a result of shorter summers in the north, the length of the stratification season is also shorter there than it is in the south. It is also noteworthy that the duration of stratification in lakes with low geometry ratios is practically independent of Secchi depth. Those lakes with small surface area and(or) great depth are prone to stratify regardless of radiation attenuation. Secchi depth has, however, a strong influence on the length of the stratification periods when lakes are polymictic. A seasonal stratijkation ratio is defined as the total number of days when stratification stronger than 1°C exists, divided by the period from the earliest to the latest date of stratification (length of stratification season). A seasonal stratification ratio < 1.Oindicates polymictic behavior, and a value of 1.0 indicates a dimictic lake (i.e. once seasonal stratification is established, it lasts until fall overturn). Results indicate that lakes with geometry ratios in the range of 4 to at least 11, and probably more, are

0

0.6

#,I, 0.8 1.0

I 2.0 A 0.25. H s ’ max

I 4.0

1

1’1’1 6.0 8.010.0

20 .O

(m-0’5>

Fig. 7. Minimum daily dissolved oxygen (mg liter-l) pleths near lake bottom (simulated 25-yr averages).

iso-

polymictic. In this respect there is no strong distinction between north and south. Dissolved oxygen characteristics-parameters that can be used to characterize DO concentrations in a lake are surface DO, near-sediment DO, hypolimnetic anoxia duration, and anoxic volume percentage. In this study, these parameters are again 25-yr averages obtained by model simulations. The minimum daily surface layer DO concentrations are constant to within about +0.3 mg liter-l across a wide range of lake morphometries and trophic states and close to saturation (note that this is again a 25-yr average, not an instantaneous daily observation). The difference between north and south is primarily due to surface-water temperature because lake elevations do not vary widely in Minnesota. Northern lakes have surface-water temperatures lower than those of southern lakes; therefore oxygen solubility is higher in the northern lakes. Minimum values of daily surface layer DO are 8.0 + 0.3 mg liter- 1 in the south and 8.5 kO.3 in the north. Bottom DO concentration is reported as the lowest simulated daily DO concentration above the lake sediments. Values of this parameter given in Fig. 7 show the effect

1131

Long- term temperature and oxygen

north

01 0.6

0.8 1.0

6.0

4.0

2.0

8.010.0

-

20.0

0.6 5,

south

-11,

0

0.6

0.8 1.0

I

I

2.0

4.0

,@, 0.25 . H

s

-

max

’

8.010.0

1

1.0

,I,

I

I 11’1’1

2.0

4.0

6.0

I

I

’

I 2.0

I

8

4.0

1

8.010.0

20.0

8.010.0

20.0

1’1’1

-

III’1

6.0

0.8

20.0

( m-o.5)

ol u ,I, 0.6 0.8 1.0

A 0.25. H s - ma x

1’1’1

6.0

I

(m-09

Fig. 8. Time (days) between first and last occurrence of hypolimnetic anoxia (simulated 25-yr averages).

Fig. 9. Maximum daily percentage of total lake volume with anoxia (simulated 25-yr averages).

of lake stratification in the low value on the left side of the graph. Oligotrophic lakes tend to have higher DO concentrations in the water near the sediments than do eutrophic lakes, primarily because of the lower SOD, and higher photosynthetic rates, especially in shallow lakes. Hypolimnetic anoxia is defined herein as the condition when DO concentration at any depth in a lake is -CO.1 mg liter-l. Although 0.1 mg liter-l is an arbitrary criterion, it is useful to identify possible low DO concentrations in lakes with different geometries, trophic state, and geographic location. For geometry ratios > 5, anoxia never occurs, regardless of lake trophic state (Fig. 8). These lakes experience sufficient vertical mixing so that oxygenrich water is frequently in contact with lake sediments, thus avoiding complete anoxia. For a geometry ratio < 5, trophic state of the lake affects duration of anoxia besides the geometry ratio. The longest duration of anoxia is estimated for eutrophic lakes, which typically have high SOD and higher water-column stability than oligotrophic lakes. Anoxic volume percentage is defined herein as the maximum daily total volume of the lake where DO is ~0.1 mg liter- 1divided by the total lake volume and multiplied

by 100. This parameter gives an estimate of the percentage of the total lake volume with anoxia. Values of anoxic volume percentage increase substantially as the lake geometry ratio decreases (Fig. 9). The rate of decrease is faster in eutrophic lakes. The numbers in Fig. 9 indicate that as much as 60% of a lake’s volume can be anoxic. The lake volume fraction unsuitable for fish would be even larger because the DO survival criterion is 2.5-3.0 mg liter- 1 instead of 0.1. Information on hypolimnetic anoxia has, however, a higher degree of uncertainty because it is directly dependent on the SOD specified in the model, and SOD is not a very precise value. Fish habitat characteristics-Fish response to habitat conditions is evaluated in terms of suitability for survival (temperature and DO) or for good growth (temperature) (Stefan et al. 1995). The good-growth season for fish begins when water temperatures exceed the minimum for good growth and continues as long as it remains below the upper threshold for good growth and DO remains adequate for survival. These limits differ by species and thus by thermal guild. The length of the good-growth season is given in Fig.

1132

Stefan et al. Northern

5

Minnesota

Lakes

Southern

Minnesota

Lakes

5

I,I,

I

I

.,.,

8

1’1’1

cold

04 0.6

south

, .( 0.6 1.0

2.0

4.0

6.0

6.010.0

5 1 cool .(.,

I 20.0

0, . ,.I 0.6

5

I.1

0.6 1.0

2.0

4.0

0.0

I

6.010.0

20.0

.,., cool

4

0

111,

0.6 0:. 0.6

_

I.1 0.6 1.0

2.0

4.0

6.0

I 20.0

6.010.0

I

0.8 1 .O

I

’

2.0 A

0.25. s

1’1’1

6.0 ’

8.010.0

20.0

H maio(m-o.5)

Fig. 12. Length (days) of no-survival conditions for cold water fish in the southern region (simulated 25-yr averages).

‘O,.O

Ap:

.

.

l-i,,,

.

.

“‘$6

‘0:6’1:0

2.0

4.0

Aso.25:

(ml-q

H,,,

6.0

I

6.010.0

20.0

(m-0.5)

Fig. 10. Length (days) of the good-growth season (simulated 25-yr averages).

Southern 5.0-

. I

1

. I

1

Minnesota

Lokes cold

:

4.0-

so2.01.o0.01

0.6

0.6 1.0

2.0

4.0

6.0

6.0,O.O

20.0

cool

0.0 I 0.6

0.0-l

0.6

1. 1 t.0

0.6

I , 0.6 I.0

4.0

2.0

2.0 ~0.25:

4.0

H,,,

6.0

6.0

(m -0.5)

6.010.0

6.6,0.0

20.0

1

20.0

0.01 . I . , 0.6 0.6 I.0

0.04 1I 0.6 0.6 1.0

2.0

4.0

2.0 A,0.25:

4.0

H,,,

6.0 6.010.0

6.0 6.010.0

:

1

20.0

1 20.0

(m-0.5)

Fig. 11. Fraction of lake volume available for good growth (simulated 25-yr averages).

10. The lake geometry ratio, as well as lake trophic state, have an influence on the cold water fish habitat. The goodgrowth season for cold water fish is given only for northern lakes because in almost all southern Minnesota lakes cold water fish cannot sustain themselves beyond the cooler water seasons. For the cold water guild, the goodgrowth season lengthens as the geometry ratio decreases (i.e. as lakes are more likely to stratify). For cool water and warm water fish this trend is not apparent. Instead, contours indicate nearly constant lengths of the good growth seasons regardless of lake geometry ratio and Secchi depth. In northern Minnesota, the length of the goodgrowth season for warm water fish is - 59d and - 134 d for cool water fish. In contrast, it is only - 100 d for warm water fish in southern Minnesota and - 106 d for cool water species. Good-growth average volumefraction indicates the fraction of the total lake volume available for good growth. The highest volume fractions for good growth (Fig. 11) seem to be available in polymictic lakes and in oligotrophic lakes for all fish guilds. In strongly stratified and eutrophic lakes, the fraction can be 0.4 or even lower. The no-survival conditions for cold water fish in the southern region expressed as the number of days during which a lake has no habitat for cold water fish are presented in Fig. 12. Because the values are 25-yr averages, cold water fish may survive in some years in a few lakes, especially those where the lake geometry ratio is near 2. Indeed the most tolerant species of the cold water guild (cisco or lake herring) have been observed in many of those lakes with only occasional partial fish kills. Eficts of climate change-Effects of climate change on the daily water temperature and DO profiles that are the basis for the foregoing graphical results have also been investigated. The output from several global circulation models (GCMs) for a doubling of atmospheric CO2 was obtained and used to modify the weather database. Simulations of projected water temperature and DO profiles

Long-term temperature and oxygen MAXIMUM HYPOLlMNETlC

TEMPERATURE

UAXIMUM

DIFFERENCE OF TEMPERATURE

(OC) BETWEEN SURFACE AND BOlTOh

5

-4 E

GISS

MODEL

-

25 Ii

-4 E zi

3

ES B O2 P

2 I 8 WC

zi El

0

0

GISS -

0.6

1.0

2.0 A&25:H,nox

LO

6.0 (m-0.5)

6.010.0

20.0

0.8 1.0 b&=:Hmox

(,n-0.5)

Fig, 13. Maximum daily hypolimnetic temperature (“C) isotherms and minimum hypolimnetic DO (mg liter-l) isopleths (simulated 25yr averages) for two northern Minnesota lakes. Past conditions (top), projected 2 x CO2 GISS climate scenario (middle), and differences between future and past climate conditions (bottom).

were then made with this modified weather data as input parameters. The results of these projections cannot be presented here in their entirety but examples are given in Figs. 13 and 14. More complete information has been reported elsewhere (Hondzo and Stefan 199 3b; Stefan and Fang 1993; Stefan et al. 1993). An example of the projected changes in two water-quality characteristics of northern Minnesota lakes due to climate change is given in Fig. 13. The climate parameter changes predicted for Duluth by the GISS 2 x CO2 model (Table 3) were used as input to the simulations. Maximum water temperatures and minimum DO concentrations in the hypolimnion of lakes before and after projected climate change are shown in Fig. 13. Either parameter can become the controlling factor for the survival of fish. Changes from past to projected future conditions are plotted at the bottom of Fig. 13. Two results stand out in Fig. 13: hypolimnetic water temperatures change less (0-3°C) than air temperatures (3.8”C average) and water temperature changes depend on lake geometry (e.g. eutrophic lakes with a geometry ratio between 2 and 4 and oligotrophic lakes with a geometry ratio between 1 and 2 are projected to have l0°C hypolimnetic temperature changes). Minimum hypolimnetic DO is not affected by climate change if lakes already experience anoxia under past climate conditions. Eutrophic and oligotrophic lakes with lake geometry ra-

2.0

hO.25:

4.0

t+mox

8.0

(m-0.5)

8.010.0

20.0

0.8

1.0

2.0

4.0

AsOs25 : Hmax

6.0

8.010.0

PAST.

20.0

(m-O.5)

Fig. 14. Maximum daily difference of temperature between surface and bottom (simulated 25-yr averages) for southern and northern Minnesota lakes. Past conditions (top), projected 2 x CO, GISS climate scenario (middle), and differences between future and past climate conditions (bottom).

tios >4 and 2, respectively, are projected to experience a drop in minimum hypolimnetic DO. The reduction will be up to 3 mg liter-‘, resulting in anoxic conditions in many eutrophic lakes with geometry ratios up to at least 20. This loss of DO, essential for fish survival and well being, will also be accompanied by a lengthening of the period of hypolimnetic anoxia, particularly in stratified lakes. The lengthening of the anoxic period could be as long as 60 d and is expected to be strongest in lakes with geometry ratios < 5 for eutrophic lakes and