North Temperate Lakes LTER

Study of Yellow Perch Improves Natural Resources Management

At the North Temperate Lakes LTER site, we have observed cyclic dynamics in a population of yellow perch in Crystal Lake over the past 20 years. Since 1981, three large cohorts of similarly aged fish have sequentially dominated the population for roughly 5 years each (Figure 1, Sanderson et al. 1999). Understanding the causes of these repeated oscillations in fish populations is an important challenge for improving fisheries management.

Figure 1. Relative abundance (CPUE) of yellow perch (Perca flavescens) in Crystal Lake as a function of size for years 1981-1996.

Regular cyclical change is one of the most intriguing types of population fluctuations for biological study. Recurrent oscillations in natural populations generate considerable interest both in their occurrence and in their cause. While cyclic patterns of abundance have been observed in freshwater fishes, the ability to detect such cycles is often obscured by the influence of environmental factors and the lack of long-term observations. Theoretical models of populations suggest that these oscillations can result from a number of factors including combinations of fertility and survival rates, intraspecific competition and cannibalism, and predator-prey dynamics.

Understanding the factors that contribute to cyclic population dynamics in fisheries improves the ability to make sound management decisions, such as setting appropriate harvest levels. Clues to the causes of these cycles come from observation and modeling made possible by long-term research. The presence of young-of-the-year fish was observed to be negatively related to juvenile perch abundance and positively related to adult perch abundance. Modeling results suggest that oscillations in young-of-the-year perch abundance were driven by the positive effect of adult perch reproduction and the negative effect of juvenile perch via cannibalism and competition. Young-of-the-year fish were abundant primarily in years when reproductively mature fish were in the lake suggesting that the cycles are driven predominantly by pulses of abundant reproductive adult perch. As these young perch grow to juveniles, they exclude the possibility of survival by younger cohorts through cannibalistic and competitive interactions. This exclusion occurs until they themselves become reproductively mature and the cycle then repeats. The long-term pattern in Crystal Lake is an exceptional example of cyclic dynamics generated by intraspecific interactions.

Long-term research on the relationships between surface and groundwater shows how lake chemistry and climate affect organisms

Long-term research at the NTL-LTER site, which was established in 1981, provided unexpected insights into the effect of climate shifts on lake-groundwater interactions. Five years after the site was established, the region experienced a severe 4-year drought. The availability of continuous, long-term data before, during, and after the drought allowed Anderson et al. (1993) to show that local flowpaths of groundwater to lakes were much more dynamic and transient than previously thought. Crystal Lake, an NTL-LTER seepage lake located high in the landscape received up to 10% of it's water inputs from groundwater during wet periods, but became totally isolated from groundwater inputs during the drought. These switches in groundwater inputs have important implications for lake chemistry and biological communities. In the Northern Highland Lake District, groundwater is the major source of materials that support aquatic life (such as the calcium needed by snails to build shells) and buffer lakes from damage by acid rain. Long term data on lake chemistry from the NTL-LTER program demonstrated that lakes moderately high in the landscape, where reversals in groundwater inflow are likely, lose cation mass during drought (Webster et al. 1996). Under the more sustained switch to warmer and drier conditions predicted by climate change models, the concentration of biologically important cations and acid neutralizing substances could substantially decline (Kratz et al. 1997). No change was observed in lakes that are always hydraulically mounded. Lakes low in the landscape, however, accumulated cations during the drought because their groundwater inputs are dominated by regional flowpaths, which are less temporally responsive to climate shifts.

Fig. 1 Changes in cation mass of the NTL-LTER lakes during drought. Lakes are arranged along the landscape position gradient. (Adapted from Webster et al. 1996).

Without the long-term data record on chemistry and hydrology from a set of lakes ranging across the landscape, these insights into the dynamic nature of lake-groundwater interactions and resultant implications for lake chemistry and biology would have been missed. Further, because of long-term studies at NTL-LTER, we were able to take advantage of a "natural" experiment-the sustained drought - allowing us to better understand differential lake responses to regional climatic events.

Long-term lake ice data reveal warming trend in northern hemisphere

Ice freeze and breakup study quantifies long-term global responses of lakes and rivers to climate change and variability

Each year lakes and rivers at northern latitudes freeze over in autumn and breakup in spring. Taken as single points in space and time, these events lack context and reveal little to the observer. Yet generations of observers have recorded such events at sufficient regularity since the middle of the last century that long-term global changes can be analyzed and significant inferences drawn of change around the Northern Hemisphere.

Figure: Historical trends in freeze and breakup dates of lakes and rivers in the Northern Hemisphere, 37 of the 39 trend slopes are in the direction of warming (modified from Fig. 1 and Table 1 in Magnuson et. al. (2000)).

What do such long-term records tell us (Magnuson et al. 2000)? First they tell us that at the North Temperate lakes LTER site, the ice duration on Lake Mendota in the winter of 1997-98 was the shortest over the period of record beginning in the 1850s. Also, the average duration of ice cover has declined from about 4 to 3 months or by 25%. Springs with the earliest breakups occurred in the year following the onset of an El Nino. The long-term trend corresponds to a warming of about 18C in 100 years. These dynamics in the timing of ice breakup and freezing are driven by climate drivers, originating far distant from Wisconsin in the Southern and Northern Pacific, and in the case of the long-term warming trend from the globally dispersed drivers behind that warming.

The Lake Mendota patterns above are observable around the Northern Hemisphere with some variation in pattern (Magnuson et al. 2000). The ability to infer long-term and regional pattern from these events, puts the observations at one site in one year in a context more useful and meaningful to us as we attempt to deal with global change. The particular events have been moved from behind the mask of the "invisible present" (Magnuson 1990) and the "invisible place" (Swanson and Sparks 1990) and as a result can be used to understand and respond more appropriately to the changes in the world around us.

Examining the long-term relationships between invasive and native species

Fifteen years of research and observation show how invasive species out-compete natives in northern temperate lakes, Wisconsin

The rusty crayfish (Orconectes rusticus), a species native to Ohio and Indiana, has been invading lakes throughout northern Wisconsin for the past several decades (Hobbs et al. 1989). This exotic species has the ability to outcompete native crayfish species (Hill and Lodge 1999) and often replaces native crayfish species in the lakes it invades (Olsen et al. 1991). In addition, the rusty crayfish often occurs at higher population densities than the native species and has been associated with reductions in macrophyte biomass. Because macrophytes provide important foraging and reproductive habitat to fish and other aquatic organisms, the

Figure 1. Rusty crayfish abundance and macrophyte biomass at site 7 in Trout Lake from 1983-2000. Crayfish abundance is in units of numbers caught per trap set divided by two.

relationship between rusty crayfish abundance and macrophyte biomass has been of interest to scientists, lake managers, anglers, and the general public. Long-term observations of rusty crayfish abundance and macrophyte biomass in Trout Lake, one of the North Temperate Lakes LTER program's primary study lakes, have shown an inverse relationship over the past 18 years (Figure1). Interestingly, in the first seven years of record (1983-1989), when rusty crayfish abundance was increasing rapidly at this site in the lake, there was no evident decline in macrophyte biomass. The inverse relationship between rusty crayfish and macrophyte biomass, which has also been established experimentally (Lodge and Lorman 1987), was only apparent after 12-15 years of record, indicating the importance of long-term data in these lakes.

Predicting Blue-Green Algal Blooms in Lakes using Long-Term Data

Urban surface runoff causes lake eutrophication

In recent years, the unspoiled waters of Wisconsin lakes enjoyed by many residents have faced a variety of threats. The real estate boom and subsequent road construction have introduced toxic pollutants, while urbanization and growth in agriculture have increased pesticides and fertilizers which have begun contaminating streams, lakes, and groundwater.

Fig. 5. Probabilities of summer blue-green algal bloom concentrations >2 and >5 mg L-1 and spring P concentrations >0.074 mg L-1 vs. current P loading rates (0% P load change) and a proposed 50% reduction in P loading rates for Lake Mendota.

While nutrients are required for lakes to thrive, too many nutrients create overwhelming imbalances which lead to 'eutrophication.' One nutrient that occurs in excess from human development and is particularly harmful to lakes is phosphorus (P). Excessive inputs of P have long been known to cause blue-green algal blooms and other eutrophication symptoms in lakes. While some lakeside development is necessary, deciding how much P is too much for a lake requires long-term research.

To predict how an individual lake would respond to changes in P inputs, scientists frequently have relied on models that link P inputs to in-lake P concentrations, which can be linked to summer algal densities or chlorophyll (Chl) concentrations. These models were derived from cross-sectional analyses of many lakes and predict average concentrations of P or Chl at steady state conditions of P inputs - a condition that rarely occurs due to variability in runoff and other drivers. Large prediction uncertainties exist when the models are applied to any one lake thus making model predictions difficult to interpret. In addition, summer blue-green algal blooms are extreme and highly stochastic events whose occurrence can be masked in average conditions. Unfortunately, few lakes have the necessary long-term data available to accurately predict algal bloom responses to stochastic watershed P input rates under a different set of land management practices.

To illustrate the value of long-term data for lake diagnostic studies, the probabilities of summer blue-green algae exceeding bloom concentrations of >2 and >5 mg L-1 were predicted for P input loading rates for Lake Mendota, one of the North Temperate Lakes LTER study lakes (modified from Fig. 5, Lathrop et al. 1998). These analyses were possible because of a 21-year record for P input loadings, in-lake P concentrations, and blue-green algal concentrations in the lake. These analyses were conducted by a collaboration between LTER and state agency researchers and were used to justify the recommended 50% P input reduction as a target for the Lake Mendota Priority Watershed Project, which will commit over $16 million of state and local monies to improve water quality in the lake. The long-term P loading data were also used by Carpenter et al. (1999) to demonstrate that to maximize the economic benefits of improving lake water quality, P input targets should be reduced below levels derived from traditional deterministic lake models because of the uncertainties in model predictions.