This section examines the criteria for the selection of record and proxy type for the study. Their strengths and weaknesses in palaeoenvironmental and palaeoclimate reconstructions are then discussed. Instrumental records in northern Russia, typically, extend back less than a hundred years and even the longest European temperature record only covers the past 350 years (Manley 1974). Their short duration mean they provide an inadequate perspective on climatic variation where forcing mechanisms act on centennial to millennial time-scales. Many natural systems, biological and physical, are climate- dependent and therefore provide an indirect record of palaeoclimatic conditions. However the climatic signal may be weak and embedded in random (climatic or environmental) noise. Interpretation of these proxy records can be challenging and a multi-proxy approach, using mutually independent records, is often used to reinforce inferences (Lotter 2003; Birks and Birks 2006). The choice of proxy depends on the research question to be answered and study location; the strengths and weaknesses of proxy records have been reviewed by Elias (2006).
1.4.1. Lake sediments as archives of palaeoenvironmental records
In continental regions the palaeoenvironment can be reconstructed from records preserved in specific depositional environments; for example, glaciers (Nesje and Dahl 2003), peatlands (Barber and Charman 2003) or speleothems (Lauritzen 2003). The choice is determined by the environments available within the selected study area and the proxy under consideration. The semi- terrestrial nature of peatlands, for example, would affect the composition of the chironomid fauna. As the composition and ecology of these assemblages are poorly known, and peatlands respond primarily to changes in hydrology (Barber and Charman 2003), this environment would not be suitable for chironomid- based palaeoclimatic reconstructions.
Lake deposits are one of the best archives of continental climate and its environmental impact (Fritz 2003). Sedimentary sequences are often continuous and may span thousands of years. Even in environments with low sediment accumulation rates, such as tundra lakes, sediment cores can give records with multiannual or decadal temporal resolution. As lakes are widespread throughout the study area lake sediment cores were selected as the most appropriate archive. A lake’s response to regional climate is mediated by site-specific catchment and basin characteristics including the presence of inflows and outflows, soils, vegetation, topography and lake morphometry (Fritz 2003). Therefore sediment cores were collected from 2-3 lakes to discern the regional climate response.
1.4.2. The selection of chironomids for this study
To minimise the response of the biota to environmental parameters other than temperature, lakes were selected above the present-day tree line. In these tundra areas, with low local pollen production, pollen records may be compromised by long-distance or extra-regional pollen transport (Birks and Birks 2003). Although plant macrofossil data can be used to valid pollen-based climate reconstruction (Birks and Birks 2003) arctic and sub-arctic vegetation is dominated by slow-growing species which may dampen observed climate responses (Birks 1981). The location above the tree-line also precludes the use of dendrochronology. There has also been a loss of thermal response in tree
ring width since the 1960s (Vaganov et al. 1999; Jacoby et al. 2000; Briffa et al. 2002b) which has reduced the sensitivity of this technique for reconstructing recent climatic change.
Diatom-inferred temperature reconstructions show close correspondence to instrumental data, in most cases, but correlated poorly when changes in the pH of the lake water occurred (Bigler and Hall 2003). However their sensitivity to pH means changes in the diatom assemblages can give an indication of changes in pH or trophic conditions which can help in the interpretation of the chironomid response (Battarbee et al. 2001).
Coleopteran assemblages have been used to reconstruct palaeotemperatures (Coope et al. 1998; Lemdahl 2000). However the large volumes of material required for analysis, typically 0.5-1kg would require the export of large quantities of sediment which would be logistically difficult. In comparison, the larval head capsules of Chironomidae (Insecta, Diptera) are abundant, diverse, well-preserved and ubiquitous in lake sediment samples (Brooks et al. 2007). A minimum of fifty head capsules are required for estimating past temperatures using a chironomid-based inference model (Heiri and Lotter 2001; Larocque 2001; Quinlan and Smol 2001). Therefore, even from arctic lakes with low productivity, less than two grams of sediment are required for analysis so cores can be finely sliced and analysed at high temporal resolution (Brooks and Birks 2004).
Chironomids are sensitive indicators of past climates. Many taxa are stenothermic and the winged adults disperse readily across the landscape which means they effectively respond instantaneously to changes in the climate. European chironomid-mean July air temperature inference models have prediction errors of about + 1.0°C (Lotter et al. 1997; Brooks and Birks 2000a; Larocque et al. 2001; Luoto 2008) and show close agreement with instrumental records over the past 100 years (Larocque and Hall 2003).
Although chironomid-inferred temperature reconstructions for the Lateglacial show good agreement with other proxy records, Holocene reconstructions are
less consistent (Brooks 2006b). Chironomid distributions are influenced by many environmental factors and changes in these may invoke a stronger response in the chironomid fauna than temperature alone (Brooks et al. 2007). Re-forestation of northern Eurasia during the Holocene would have enhanced soil development, which in turn, would alter the pH, DOC and chemical characteristics of lakes within the catchment. Chironomids responding to these localised parameters may give anomalous temperature reconstructions.
1.4.3. Chironomids as a proxy in palaeoenvironmental reconstructions
Chironomids have been used in palaeoenvironmental studies in Eurasia since the 1920s (Brooks 2006b). In early work chironomids were used as qualitative indicators of trophic change (reviewed by Lindegaard 1995). Changes in faunal assemblages over the Holocene were interpreted as classical lake succession from oligotrophic to eutrophic status (for example Bryce 1962; Goulden 1964). Andersen (1938) published a Lateglacial stratigraphy from Denmark in which he suggested shifts in the faunal composition between the Older Dryas and the Allerød and again in the Younger Dryas were climate-driven. However the concept that chironomids were influenced by changing climate as well as changes in lake productivity was seldom considered until the 1980s (Brodin 1986; Hofmann 1988).
Walker and Mathewes (1987; 1989) demonstrated that chironomids were sensitive to Lateglacial and Holocene climate change and, in a pioneering step, developed chironomid-temperature inference models to quantify the climate change (Walker et al. 1991; Walker et al. 1997). The encouraging results led to the development of chironomid-inferred temperature (CI-T) models in several European countries; Finland (Olander et al. 1999), Norway (Brooks and Birks 2000a, 2001) and Sweden (Larocque et al. 2001). Initially the inference models focused on reconstructing surface-water temperature; since the majority of the chironomid life cycle (section 1.5) is aquatic, water temperature would have a significant impact on the distribution and abundance of species (Brooks 2006b). However prediction errors were relatively high. Water temperatures show wide day-to-day fluctuations and the use of 30-year mean July air temperatures, based on meteorological data corrected for altitude and distance from the coast,
improved the performance of the models (Lotter et al. 1997; Olander et al. 1999; Brooks and Birks 2001; Larocque et al. 2001). Air temperature influences the dispersal of the terrestrial adult stage and is usually closely correlated with surface-water temperature (Livingstone and Lotter 1998; Livingstone et al. 1999). However air temperatures are consistently underestimated if the lake receives substantial influxes of melt-water from glaciers or snow-beds (Brooks and Birks 2001).
The chironomid larvae are identified from characteristic features of the heavily chitinised head capsules. In most specimens these are only identifiable to a generic or morphological type level. This limits the applicability of a training set to the biogeographical region in which it was assembled as species within the same genus, which are indistinguishable as subfossils, may have different temperature optima (Brooks et al. 2007). Therefore one of the aims of the research (section 1.6) is determine whether a regional training set compiled from Russian assemblages is more appropriate for palaeoclimate reconstructions in northern Russia than either the Norwegian dataset (Brooks and Birks 2001 and unpublished data) currently used (for example Solovieva et
al. 2005) or a combined Norwegian – Russian training set. Combining the
datasets would greatly increase the number of samples and give a more even distribution along the environmental gradients but can only be justified if the faunal composition, selective pressures and taxon-responses are similar in the two geographical areas.
The relationship between chironomids and environmental variables, other than temperature, has been used to quantitatively reconstruct past total phosphorus (Lotter et al. 1998; Brooks et al. 2001), chlorophyll-a (Brodersen and Lindegaard 1999a), salinity (Walker et al. 1995; Verschuren et al. 2004; Eggermont et al. 2006; Zhang et al. 2007), hypolimnetic oxygen conditions (Quinlan et al. 1998) and water depth (Korhola et al. 2000). Birks and Birks (2006) argued that the emphasis on palaeoenvironmental reconstructions has led to a neglect of the study and understanding of lake biotic responses to changing internal or external factors, of lake dynamics and processes, and of the underlying biology and ecology of the organisms preserved in lake
sediments. These factors need to be considered in data interpretation. Multiproxy studies are often used to study these ecosystem-scale processes, with their complex networks of interactions, in order to gain a wider overview of the situations than could be acquired from a single proxy (Birks and Birks 2006). Therefore dry weight, loss-on-ignition (LOI) and stable isotopes (δ15N and δ13
C) concentrations were determined in the sediment cores to facilitate the interpretation of changes in subfossil chironomid assemblages.