Model studies and the analysis of empirical data suggest that one of the effects of climate change will be an increase in the temperature of lakes (Hondzo & Stefan 1993, Livingstone 1995, Robertson & Ragotzkie 1990, Schindler et al. 1990). Surface water temperatures tend to vary closely with air temperature in spring and summer (Livingstone 1995, Livingstone & Schanz 1994), and show a secular increase similar in magnitude to that of air temperature over a period of decades (Livingstone 1995).
Temperature is not the only climatic variable of concern. Variations in wind speed, wind direction, rain and snowfall are also important, affecting with temperature the timing and intensity of thermal stratification, ice-cover, turbidity and light penetration (Catalan 1988).
The chemical and biological status of lakes is strongly related to these weather-induced physical processes acting both on the lake and its catchment. The relative length of water column stratification and mixing periods is especially important. The most biologically productive periods occur when a long stratification period is followed by a sufficiently long mixing period, allowing nutrients to be internally recycled (Goldman et al. 1989). According to model studies (Hondzo & Stefan 1993), global warming could result in a lengthening of the period of summer stratification by more than a month in deep and medium-deep lakes.
Variations in annual weather patterns cause fluctuations in primary productivity, and if sustained over longer periods changes in the species composition of plant and animal populations can occur, both by direct (e.g. temperature) and indirect (e.g. pH) effects. Such changes are recorded by the biogenic fraction of the lake sediment, both as amorphous organic matter and by diverse sub-fossil remains. Accumulating sediments also contain mineral material derived from the lake catchment, that varies in composition and quantity according to changes in inputs associated with changing snow and ice cover.
The sediment record can then be used to evaluate both short and long-term variability in climate patterns, necessary both for the interpretation of past climates (and for palaeoclimatological research), and as a baseline to assess the impact of future warming.
3.2 Specific objectives
- to collate, harmonise and analyse appropriate long-term instrumental climate records for the regions under consideration;
- to model the relationship between climate records for mountain weather stations (at study sites) with those for lowland stations;
- to assess the physical, chemical and biological seasonal variability of non-polluted remote lakes by intensive sampling and analysis at a small number of key sites;
- to establish the relationship between the distribution of the primary biological constituents of mountain lake ecosystems that form the most important groups in sediments cores i.e. diatoms, chrysophytes, cladocerans and chironomids, and measured environmental summary variables;
- to assess underlying trends and natural variability in climate from the fine-detail analysis of the uppermost sediment of study sites;
- to compare the sediment record at study sites with instrumental records of temperature and precipitation, to calibrate and validate the DYRESM and AQUASIM models and to run the models using alternative scenarios for future climate.
The lakes in this study share the same basic limnological characteristics. They are oligotrophic, typically dimictic and have a winter ice-cover and a summer thermocline. They are all situated above or beyond the regional tree-line, and they have poorly vegetated catchments.
Most of the sites were included in the AL:PE project, and are those identified in that study to be least influenced by acid deposition. With the inclusion in the EU of Finland and Austria, and the participation of Slovenia and Switzerland, additional appropriate sites have been identified in these countries, all of which have been the subject of prior study.
The sites are shown in Figure 1. They include sites in the arctic and northern latitudes (Ovre Neadalsvatn in Norway, and Saanajarvi in Finland), sites in the Alps (Gossenkollesee in Austria, Hagelsee in Switzerland and Jezero Ledvicah in Slovenia), sites in Spain (Estany Redo in the Pyrenees, Laguna Cimera in the Gredos Mountains and La Caldera in the Sierra Nevada), and a site in the Slovakian Tatra Mountains (Terianske Pleso).
3.4 Project description and methods
3.4.1 Obtain and harmonise instrumental weather records over the last 200 years
Lead laboratory: UED
Europe is particularly fortunate in having the longest and densest network of historical climate records of any part of the world. A unique body of climate data has been assembled over the last five centuries. Much of the data, the earliest reaching back to 1525 AD, has recently been converted into machine readable form. Modern computing power gives us the opportunity of unifying, comparing, correcting and standardising the data for use in studies of climate variability. Excellent, painstaking work has taken place to standardise and harmonise mean air temperature data since 1850 AD (Jones et al. 1986, Hansen & Lebedeff 1987). Similar work is presently taking place on the full suite of standard climate parameters spanning the last 100 years as part of the North Atlantic Climatological Dataset (NACD) project led by the Danish Meteorological Institute. However, additional work is needed on the longest European instrumental records in order to allow comparison with the palaeolimnological records from lake sediments.
Ten regions of Europe have been selected for new harmonisation (Alexandersson 1986) studies. In each region a local network of stations has been identified which will allow the construction of an independent climate series. Following harmonisation, time-series and spatial analysis of the data sets are being used to provide a base for validation work across the full range of remote sites.
3.4.2 Correlate records of weather patterns between lowland meteorological stations and montane sites, and measure local weather patterns
Lead laboratory: EAWAG
Relationship between local montane and lowland climate records
Correlation between on-site weather data and data from nearby lowland meteorological stations is being made to enable short and long-term instrumental data records (see 3.4.1) from lowland sites to be used to interpret data at remote MOLAR sites.
A key issue is that of establishing the relevance of the existing long-term lowland datasets (e.g. Pfister's indices) to the qualitatively dissimilar climatic situation prevailing in high mountain regions, for which there is a comparative paucity of data. A number of factors need to be considered:
- mountain regions are meteorologically heterogeneous, much more so than lowland regions;
- high-altitude meteorological stations tend to be situated in exposed locations, often on mountain-tops, and therefore reflect more the free-atmosphere situation than the high-altitude near-ground situation;
- mountain lakes, by their very nature, tend to be situated in sheltered valleys or corries. Valley stations in mountainous areas sometimes tend to behave like lowland stations and sometimes like mountain-top stations, so simple interpolation may not be the most appropriate solution.
Consequently it is imperative that an effort be made to relate the micro-climate of the lake to the macro-climate of the region and to the synoptic-scale climate represented by the long-term lowland data series. This presupposes a knowledge of the degree of meteorological heterogeneity involved, which can only be obtained by analysing data from stations from different types of mountain locations, ranging from exposed mountain-tops to plateaux and to sheltered high-altitude valleys.
Monthly, weekly and some daily series of temperature, rainfall and river discharge are being analysed from Europe's mountain ranges. Records from the Pyrenees, the western and eastern Alps, the Dinaric Alps and eastwards to the Anatolian Plateau are being compared, along with data from the Caledonian and Scandinavian mountains and the sub-arctic regions of Europe.
On-site meteorological data
The relationship between site-specific weather patterns and those recorded at national meteorological stations is being assessed using data from automatic weather stations positioned at each site. Weather Stations are used to measure wind speed and direction, air temperature, relative humidity and radiation. Precipitation gauges have been set up to measure rainfall and snowfall.
In addition, essential ground-truth work to relate data on the freezing and thawing of ice on remote mountain lakes to local and regional air temperature data is being carried out to determine historical freeze-up and break-up dates of the MOLAR lakes; and evaluate the usefulness of the freeze-up and break-up dates of high-altitude and high-latitude lakes as proxy air temperature data.
3.4.3 Assess the seasonal variability of physical, chemical and biological characteristics of sites
Lead laboratories: UCL, NIVA, UHEL, FBG, UGR.ES, EAWAG, UIBK, IZ-SAS, NIB
To assess how lake chemistry and biological communities respond to changes in ice-cover, light penetration, temperature change, stratification and mixing, essential water column data (temperature, oxygen, pH, conductivity and chlorophyll-a) are collected at bi-weekly intervals in the ice-free season, and at the beginning, middle and end of the ice-cover period. Samples for major ion (calcium, magnesium, sodium, potassium, alkalinity, chloride and sulphate) and nutrient (nitrogen, phosphorus, dissolved silica) chemistry, phytoplankton (diatoms and chrysophytes) and zooplankton analysis are collected at monthly intervals. Benthic diatoms and chironomid sampling involve one detailed transect with the use of scuba to sample deep-water habitats. In addition sediment traps have been deployed in order to assess the quantity and timing of material fluxes to the sediment, both of biogenic (autochthonous) and abiogenic (allochthonous) material. The traps are emptied on a monthly basis.
Physical and chemical analysis are the responsibility of the site operator. Analysis of biological samples and sediment trap contents is the responsibility of specialist laboratories, as in 3.4.4. Protocols for sampling and analysis are those developed for AL:PE.
3.4.4 Harmonise taxonomy of key indicator taxa and model their distribution in relation to environmental variables
Lead laboratories: UCL, UiB(ZI & BI), FSCU, ILIMNOL
The AL:PE project established an excellent taxonomic and ecological database for diatoms, cladocera and chironomids from arctic and alpine environments. In this project it is necessary to build on this knowledge to: (i) harmonise taxonomic procedures with those of new participants (from Switzerland and Finland) and (ii) develop a taxonomic system for chrysophytes and chrysophyte cysts.
A series of workshops has been organised to harmonise taxonomic conventions for diatoms (UCL), chrysophytes (ILIMNOL), cladocera (FSCU) and chironomids (UiB(ZI)/UCL) and to develop statistical models (UiB(BI)) relating the distribution of key taxa to their environmental (temperature, pH, nutrients) optima and tolerances in mountain lakes.
3.4.5 Establish long-term variability in ecosystem dynamics from recent palaeolimnological records
Lead laboratories: UCL, ULIV, UiB(BI), UHEL, FBG, ILIMNOL, EAWAG, NIB, FSCU, CNR
The longer-term (decadal) lake and climate variability will be assessed from a high resolution analysis of recent sediments. Lake sediment records are extremely rich and diverse. In this project the key biological remains (benthic and planktonic diatoms, chrysophyte cysts, cladocera and chironomids) are analysed, along with analysis of the mineral fraction of the sediment derived from the catchment.
Sediment sampling, mineralogy and dating
Core samples of the upper sediment are taken with a wide-diameter piston corer. The uppermost 20 cm of sediment are sub-sampled at contiguous 2 mm intervals to provide an average sample resolution of 5-10 years. In the laboratory the dry weight, wet density and loss-on-ignition of each sample are measured. Mineralogical and magnetic analyses are also carried out to characterise changes in catchment erosion. The sediments are being dated using 210Pb dating, supplemented by 137Cs and 241Am (ULIV). This method allows sediment that has accumulated over the last 150 years to be dated accurately, an essential prerequisite for comparison with instrumental climate records.
Diatoms and chrysophytes
The primary production of arctic-alpine lakes is dominated by diatoms (especially in the benthos) and chrysophytes (mainly as plankton) that are sensitive to pH and nutrient changes. Since pH and nutrients in unpolluted waters are controlled by climate and variations in climate (see above), the record in sediments of the changes in abundance and species composition provides an excellent means of reconstructing climate change (e.g. Psenner and Schmidt 1992). A large diatom - chemistry training set has already been developed within the AL:PE project, but there is a need to integrate the Swiss dataset and to expand the training set in the medium pH (5.5 - 6.5) range, and to include more arctic samples. Similar protocols will be developed for chrysophytes using the methods of Duff et al. (1994).
Diatom analysis is co-ordinated by UCL who are also responsible for the analysis of Norwegian sites. Cyst assemblages from Scandinavia and the UK are be studied at UCL, those from the Alps are studied at ILIMNOL and those from Spanish sites at FBG. AQC is achieved by exchange of material and workshops, coordinated by ILIMNOL (see 3.4.4).
Remains of Cladocera are abundant in lake sediments and are good indicators of environmental change (Frey 1976, Prazakova & Fott 1994, Whiteside 1970). However, their response, direct or indirect, to climate variability in alpine lake ecosystems has not yet been studied. In this project we have an opportunity to evaluate their importance using a range of data such as geographical distribution of individual species from training sets, studies of seasonal variability within and between study sites, and by comparing changes in Cladoceran assemblages with other components of the sediment record.
Cladoceran analyses is carried out by several laboratories co-ordinated by FSCU, who are responsible for protocols, quality control exercises and workshops.
Chironomid midge larvae are ubiquitous in freshwater ecosystems, are especially abundant in montane lakes, and their head capsules are extremely well-preserved in lake sediments. They are good indicators of temperature, especially summer surface water temperature (Walker et al. 1991a,b).
Field and laboratory methods for chironomid sampling involve separation of littoral/sublittoral forms from profundal elements to exclude organisms from analysis that live below the thermocline. Chironomid assemblages from Norway are studied at UiB(ZI), from Finland at UHEL, from Austria and Switzerland at UCL, and from Spain at FBG. Taxonomic workshops, co-ordinated by UCL, are held to achieve taxonomic consistency and AQC (see also 3.4.4).
Pigments and C:N ratio
One of the expected responses of a montane lake ecosystem to climate variability is a change in productivity. As the abundance and composition of different pigments preserved in the sediments has been shown to be closely related to lake productivity (Guilizzoni et al. 1983, Guilizzoni & Lami 1992, Leavitt 1993), analyses of sedimentary pigments will be made. Climate variability may also influence the balance between autochthonous and allochthonous organic matter in lake sediments, and this can be characterised from carbon:nitrogen ratios.
3-5 g wet weight sub-samples from each core section are transferred to CNR. Samples are analysed by spectrophotometer and HPLC (Guilizzoni & Lami 1992, Lami et al. 1994). Carbon and nitrogen are analysed using a CN elemental analyser.
Data processing and comparison with instrumental climate record
Training-set data are being compiled in a relational database at UIB(BI) using ACCESS software. Transfer functions will be developed for lake-water pH, nutrients, and temperature using weighted averaging regression and calibration (Birks et al. 1990) and its extension, weighted averaging partial least squares (ter Braak & Juggins 1993), using WACALIB and CALI software. Sample-specific errors of prediction of each reconstruction will be derived by statistical cross-validation or by Monte Carlo simulation using WACALIB and WAPLS software. Statistical and ecological evaluation of all reconstructions will be made using the approaches of Birks et al. (1990) and Birks (1995), namely modern analogue matching, poorness-of-fit measures, and root mean squared errors.
Finally, climate reconstructions based on the sediment core records will be validated by correlation with long-term instrumental records (3.4.1).
3.4.6 Model the relationship between weather patterns and lake dynamics, validate the model against the sediment record, and forecast lake response to alternative climate scenarios
Lead laboratories: FBG, UCL, EAWAG, UiB(BI)
A dynamic model of mountain lake ecosystems will be developed relating weather forcing with water column chemistry and biology and the sediment record. The model will then be used to hindcast responses to climate change in the recent past. These predictions will be validated using the 200 year instrumental weather record (as model inputs) and the sediment record (3.4.1 and 3.4.5) as a proxy reference for output. The model will then be used to forecast responses to alternative future climate scenarios. A full sensitivity analysis of the model will also be carried out.
The model itself will be based on the architecture and input needs of different existing models (DYRESM and AQUASIM) using empirical relationships developed in 3.4.2, 3.4.3 and 3.4.4. The model will be constructed in several steps involving both empirical and process-based links. Initially the relationship between weather patterns and lake physics will be established by customising two physical lake models to our experimental sites. DYRESM has been developed by the Centre for Water Research in Western Australia (Imberger & Patterson 1987, Patterson & Hamblin 1988). AQUASIM is presently being developed at EAWAG in Zurich. EAWAG is directly involved in this project and the Australian Centre cooperates with the project as advisers to the FBG group. Key variables to be modelled are summer temperature, length of winter ice-cover, and stratification duration, that determine much of the chemical and biological variability in these lakes (Catalan & Fee 1994).
Three components of the chemistry, biology and sediment record will be modelled from the output of the physical model. These are particulate organic matter (or chlorophyll-a as a proxy), redox conditions, and abundance of organisms that leave remains in the sediment. The first two of these will be simulated using DYRESM and AQUASIM, that contain chemical and biological routines (Hamilton & Schladow 1993), and using empirical models developed from 3.4.2, 3.4.3 and 3.4.4. The last will be based on empirical models, derived from the multivariate statistical analysis of species-environment datasets, partialling out the influence of different physical and chemical variables on organism composition and abundance (Borcard et al. 1992).