Midges reveal past climate

Midges (chironomids) are two-winged, non-biting insects. Most people notice them only when swarms near lake-side settlements are a nuisance. Under special circumstances, midge swarms can be so big that from a distance they appear to be clouds of smoke. In these cases, midges can indeed be a nuisance and even represent a heath risk by causing allergic reactions. The Norwegian media recently reported conditions similar to these at Gunneklevfjorden in Porsgrunn.

Hobby fishermen and aquarium owners know that midge larvae and pupae are excellent fish food and provide an important component in the diet of a number of other creatures living in or around freshwater. In Freshwater Ecology, chironomids have long been recognized as excellent indicators of lake water quality (see Armitage et al. 1995). In a single lake more than a hundred different chironomid species may occur. As many of these species have very specific ecological requirements with respect to, for example, nutrient status, pH, or oxygen availability, the chironomid fauna can be used to assess environmental conditions. Freshwater biologists have also long known that many midge species are restricted to lakes within a particular temperature-range.

More recently, paleoclimatologists (scientists studying the past climate) have also shown increasing interest in chironomids – ever since studies in the early 1990s showed strong statistical correlations between summer water temperature and the chironomid species occurring in freshwater lakes. This implies that analyses of midge remains in lake sediments can be used to provide quantitative estimates of past temperatures. Climate is a key factor affecting lake ecosystems, even though it is still uncertain how exactly it affects the chironomid fauna. This uncertainty comes from the fact that the temperature does not only affect the physiological processes of the midge directly, but also affects the nutrient status, the mixing regime (that is, how water masses are mixed), and thus also the oxygen availability in the lake.

 

In northwest Europe, the method has been used with considerable success to provide estimates of the magnitude of the large temperature fluctuations at the end of the late-glacial period (e.g. Birks and Ammann 2000; Brooks and Birks 2000a, b). Within the framework of the Norwegian Research Council Project on Past Climates of the Norwegian Region (NORPAST) and the University of Bergen’s related Norwegian Research Council Strategic University Project Norwegian Palaeoenvironments and Climate (NORPEC), a number of chironomid researchers in Bergen and London have now come together to apply and extend this approach to several Holocene and late-glacial sequences in Norway.

In order to grow, insects have to shed their stiff exoskeletons several times during development. Shed larval exoskeletons and remains of dead chironomids formerly living in a lake accumulate in the sediment, and strongly sclerotised fragments are preserved unaltered for thousands of years. Identifying these subfossil larval remains is a difficult task, as many of the fragments are damaged and lack important diagnostic features. Furthermore, for a number of species the larvae have not yet been described. Nevertheless, more than 140 species and larvae types can now be recognised in Norwegian lakes based on subfossil material, and over 90% of these show distinct and statistically significant relationships to temperature.

Reconstructing past summer temperatures based on chironomid remains is a three-step process. In the first, the surficial sediments of a large number of lakes are sampled and analysed for subfossil chironomid assemblages, and surface-water temperature for each site is measured. However, acquiring reliable summer water temperature estimates for a large number of lakes is difficult. Surface temperatures in any given lake can undergo large diurnal fluctuations or change on a day-to-day basis. In northern temperate regions water temperature is strongly linked to air temperature. Next to taking single spot measurements of lake-water temperature, an alternative is to use estimates of air temperature. By using a 30-year average of monthly temperature data from a network of meteorological stations combined with a Geographical Information System (GIS) and statistical regression procedures, estimates of the air temperature for each lake in a data set can be obtained.

 

In the second step, a numerical model is developed to predict the lake air or water temperatures based on the composition of the subfossil chironomid assemblages. The large temperature gradients in Norway call for a unimodal (bell-curve)-based rather than for a linear modelling approach. The models quantify optimal temperature and tolerance levels for each species. Experiences from diatom-based palaeolimnological studies, based on a statistical technique called WA-PLS (weighted averaging partial least squares regression), have been used to develop a chironomid-temperature calibration function for Norway. The calibration model, based on 153 Norwegian lakes ranging from Southern Norway to Svalbard, predicts July temperatures with a leave-one-out cross-validated coefficient of determination (r2) of 0.90 and root mean square error of prediction (RMSEP) of 1.01°C for air temperature and a r2 of 0.83 and RMSEP of 2.46°C for water temperature.

 

The final step of this process involves applying the calibration function for midge-inferred temperature to different levels in a lake sediment core. More specifically, a down-core chironomid stratigraphy is developed based on radiocarbon (14C) or lead (210Pb) dated sediments and the modern chironomid-temperature calibration function applied to the fossil chironomid assemblages. Given that the composition of species at each level of a sediment core reflects the temperature conditions when the exoskeleton was shed, it is possible to calculate the past temperature (within a margin of error). An example of this kind of reconstruction in Scotland is shown in Figure 1.

Within the NORPAST and NORPEC projects, chironomid analysis is being applied to a total of six Holocene sediment cores covering a transect from southernmost Norway to the Lyngen area and two late-glacial cores from the extreme south and extreme north of Norway. The Holocene cores cover approximately the past 10,000 years after the last glaciation, during which there was a relatively constant climate similar to present-day conditions. Chironomid-based temperature estimates will provide independent reconstructions of summer temperature changes in Norway at a centennial to millennial time scale. Combined with reconstructions of past climate from palaeobotanical and geological data, these estimates will provide new and important information on the late-glacial and Holocene temperature and precipitation patterns of Norway and on the natural variability of the climate system under similar boundary conditions as today.

 

Because the expected temperature changes within the Holocene are smaller than in the late-glacial and probably close to the prediction errors of the chironomid-temperature models (ca. ±1°C), an additional important task is to try to improve the reliability of the chironomid-based temperature reconstructions. This is being addressed by using new statistical calibration approaches (e.g. artificial neural networks), by investigating within-lake sedimentation patterns of subfossil chironomids, and by re-examining and partitioning the different error components of the chironomid-temperature models.