Will lake warming in Minnesota drive cold-water fish to extinction?

Cold-water fish are stressed as temperatures rise and oxygen is reduced in Minnesota's lakes

Feature Image: The Cisco, a cold-water fish also known as Tullibee or Lake Herring

Climate change is beginning to impact cold-water fishes in Minnesota. Cold-water fishes are: cisco (also known as tullibee or lake herring), burbot, whitefish, and lake trout. Cold-water fish kills in Minnesota lakes have been reported in the summer of 2006 and again in 2012.  As the water in Minnesota’s many lakes warms, living space (habitat) for cold-water fish is lost.  


In 2007, Professors Heinz Stefan and Xing Fang (Auburn University) began to collaborate with the Minnesota Department of Natural Resources (MNDNR) to identify those lakes in Minnesota in which the cold-water fish cisco would have the best chance of survival. Cisco is of some commercial value, but most importantly, it is a “canary in a mineshaft” indicator species for other cold water fishes, and a vital food source for trophy fish such as walleye and northern pike. To make these projections, Stefan and Fang relied on experience with lake process and fish habitat simulations, gained over many years.
In the 1970s fisheries biologists at the U.S. Environmental Protection Agency (EPA) in Duluth began to study the effects of industrial cooling water on fish. The effect on fish of warmer water discharged into lakes and rivers, such as from electric power-generating plants, were not well understood. Fish respond to water temperature change differently, depending on whether they are adapted to cold, cool, or warm water. Then SAFL graduate students working with Heinz Stefan (one of them Professor John Gulliver) measured and modeled the temperature and dissolved oxygen regimes in the outdoor streams of the Monticello Ecological Research Station, where the EPA field studies with fish were being conducted.  
About 20 years later, the threatening effects of global warming on inland water bodies became of increasing concern, and studies of lakes and streams were conducted to project how much water temperatures would rise if global greenhouse gas emissions continued to rise. By developing and using computer simulation models and field measurements, another group of graduate students from SAFL, including Miki Hondzo, Omid Mohseni (Barr Engineering), Roy Gu (Iowa State University) and Xing Fang (Auburn University), estimated daily water temperature, ice cover and dissolve oxygen (DO) dynamics in freshwater lakes and streams across the U.S. For example, it was determined that summer surface temperatures in many Minnesota lakes and streams would rise approximately three degrees Celsius, if atmospheric CO2 doubled. Knowing the projected water temperatures and DO in small freshwater lakes, rough projections of habitat loss for adult cold-, cool- and warm-water fishes could be made. 
The Cisco

Cisco, Coregonus artedi, lake herring, tullibee (from Mn DNR).

The cisco is more tolerant of elevated water temperatures than other cold-water fish, but it is very sensitive to changes above 20°C.  Cisco fish are found in about 600 lakes mostly in two ecoregions of Minnesota: the NLF (northern lakes and forests) and the CHF (central hardwood forests) regions.  The reaction of the cisco population is a good indicator of climate change effects.  In order to survive, cisco requires sufficiently cold temperatures and enough dissolved oxygen.  Cisco begin to experience oxythermal stress when the dissolved oxygen drops to approximately 3.0 mg/L. 


Lake temperature and DO dynamics
Critical periods for the survival of fish can be in summer or in winter. With global climate change, lakes experience an increase in surface water temperature in summer, and a decrease in dissolved oxygen (DO) at greater depths. Near the surface, and as water temperature increases, small and large plants in lakes (phytoplankton and macrophytes) produce more oxygen by photosynthesis; but below the photic zone where this occurs, light is insufficient, and decaying plant mass consumes the oxygen dissolved in the water. As decaying plants take up more of the DO, less is left for the fish.
It is especially important to consider how these two variables, temperature and DO, impact the stratification (layering) of many Minnesota lakes. In summer, deeper lakes consist of three major strata: a warmer top layer with much dissolved oxygen; a middle layer with some dissolved oxygen and a rapid change in temperature; and a cooler, less oxygenated layer all the way to the bottom. Shallow lakes, which typically are not cisco lakes, may be missing a seasonal stratification. As water temperature increases in a lake, the top layer can become an unsuitable habitat for cisco.  Meanwhile, as DO decreases, cisco cannot survive in the bottom layer. In the course of the summer, the surface  layer deepens due to wind mixing and the middle layer may thin out while it sinks deeper.  Essentially, much habitat for cold-water fish can be lostin the critical summer period.
Research Goals and Methodology
Pete Jacobson and Don Pereira from MNDNR collected and analyzed extensive water quality and cisco population data in Minnesota lakes.  For fisheries management decisions it was especially important to identify “refuge” lakes, lakes that would continue to provide habitat for cisco despite future climate changes. Their first step was to determine the attributes of an ideal cisco lake. They found that cisco lakes tend to be clearer, deeper and less trophic, meaning there is less primary productivity (plant life) in them.  The MNDNR recognized 620 lakes as cisco habitat lakes.  For the purpose of temperature and DO simulations, Stefan, Fang and their students, identified thirty virtual lakes to represent the real lakes.  The virtual lakes vary in depth, surface area, and secchi depth.  Secchi depth is the depth from a lake's surface to the deepest point at which the human eye can see a 20cm diameter disk.  In cisco lakes secchi depth can be directly correlated to primary productivity (plant growth).  
Atmospheric boundary layer transition and lake mixing diagram
The second step was to develop computational models to simulate daily water temperature and DO profiles in the thirty virtual lakes.  Canadian and Japanese models of the global atmosphere, which most accurately simulated daily weather for the past, were chosen to simulate future atmospheric conditions as input to the lake.  The lake simulation models were developed from earlier lake models.  Many processes and variables are taken into account in these lake models including heat exchange, surface re-aeration, photosynthesis, sedimentary oxygen demand, biochemical oxygen demand, vertical mixing dynamics, and plant respiration.  Heat exchange within a lake and at a lake’s surface and sediment bed, is complex. Processes that contribute to heat exchange are evaporation, short and long wave radiation, back radiation, convection, and sediment heat transfer. SAFL graduate students Corey Markfort and Emily Resseger (now with the Metropolitan Council) with help from the U.S. Geological Survey (USGS), provided information on wind sheltering of lakes.   

Instrumentation to study wind sheltering

Dr. Fang and students from Auburn University generated daily temperature and dissolved oxygen profiles from the lake simulations with the new MINLAKE2010 model for two projected climate scenarios. Next, the cisco fish habitat requirements were brought into the picture.  The survival constraints of the cisco were derived from measured temperature and DO profiles in lakes with cisco populations and applied to the simulated future temperature and dissolved oxygen profiles for all 620 existing cisco lakes in Minnesota.  
Cisco habitat was identified by a parameter called TDO3 (water temperature at the depth where DO=3mg/L. The lakes were grouped into three tiers based on TDO3 values. The first tier represents prime cisco-sustaining lakes.  The second tier represents acceptable cisco lakes.  And the third tier represents poor cisco habitat.  If the average TDO3 over any sliding one month period in the summer was less than 11°C, cisco habitat was considered excellent (tier 1), when it was less than 17°C, cisco habitat was considered as good (tier 2), and values above 17°C (tier 3) identified non-refuge lakes. Pete Jacobson determined these limits by the analysis of extensive field data. The TDO3 simulation results were plotted against two independent other parameters that characterize the trophic state and the wind mixing or stratification potential of a lake. These two parameters were the secchi depth which in Minnesota lakes is strongly correlated with the standing crop of phytoplankton (algae), and the lake geometry ratio calculated from lake surface area and maximum lake depth. The plot of simulated TDO3 was then applied to all real Minnesota lakes, which could also be characterized by secchi depth and lake geometry ratio.  
By interpretation of the simulation model results obtained as described above, it is projected that by about 2100 and assuming continued climate warming (according to the 2007 Intergovernmental Panel on Climate Change), 460 of the original 620 cisco lakes would no longer support cisco habitat (tier 3).  On the other hand, 131 lakes were classified as tier 2 lakes and 29 as tier 1 cisco refuge lakes.  This is an alarming result for the projected future of cisco, and other less tolerant cold-water fish species in Minnesota lakes, as well as for fish that feed on the cisco.  


Lakes with reported cisco mortalities in the summer of 2006 (from Mn DNR).

In the summer of 2006, Minnesota saw cisco fish mortalities in 18 lakes. The accuracy of the model predictions was validated against these events. Recorded past weather data was used as an input for the model.  The outputs showed remarkable agreement between model predictions and observed adult cisco mortality.
Next steps
Additional research is needed to refine the understanding of climatic impacts on fish habitat. For example, water temperature constraints on fish reproduction are a sensitive aspect of fish survival. Finer time or spatial scales and better data input from weather stations could improve predictions. And analysis of the changes in nutrient inputs from the watershed is also needed.
However, as a result of this significant research, MNDNR can now prioritize lakes in northern and central Minnesota for protection against adverse impacts of land development, agriculture, construction, tourism and other causes. in order to maintain viable cisco and other cold-water fish populations.  Monitoring of incoming sources of water, eliminating point and non-point pollution sources, and effective oversight for tourism are examples of necessary protection efforts. The results of this study can impact environmental policy decisions and management of lakes.  And this model can be translated to other regions to forecast future environmental conditions.  
Funded by Minnesota Department of Natural Resources