Supplementary Information to “A comparison of three Eurasian chironomid-climate calibration datasets on a W-E continentality gradient and the implications for quantitative temperature reconstructions”
Stefan Engels, Angela E Self, Tomi P Luoto, Stephen J Brooks, Karin F Helmens
Calibration datasets
The Norwegian calibration set was initially published by Brooks and Birks (2000) as a 44-lake dataset. It was then expanded by Brooks and Birks (2001; 2004) and Self et al. (2011) present the currently used version of the Norwegian calibration set which now includes Norway as well as Svalbard. The Norwegian calibration set spans a modern July air temperature range of 3.5-16.0ºC and it contains a total of 142 taxa and 157 lakes (Table 1). After deletion of 4 outliers, the dataset used for model development includes 140 taxa. A WA-PLS inference model with 3 components was selected as the preferred model, combining a high r2 (0.90) with a low root mean square error of prediction (RMSEP) (1.01°C) (Table 1). This Norwegian calibration model was used by Engels et al. (2008a; 2010) to quantify Tjul for the MIS 3 and MIS 5d-c intervals, respectively.
The Finnish calibration set (Luoto 2009) contains 82 lakes, of which 77 are included in the final inference model. Five lakes were excluded from the final model due totheir unusual environmental or physical characteristics (Luoto 2009). The Finnish calibration set spans a smaller modern July air temperature gradient (11.3-17.0°C; Table 1). The final inference model selected by Luoto (2009) is a 1-component WA-PLS model which has a r2jack of 0.78 and a RMSEP of 0.72ºC (Table 1).
The Russian calibration set comprises lakes that were sampled by Porinchu and Cwynar (2000), Solovieva et al. (2002; 2005), Sarmaja-Korjonen et al. (2003), Kumke et al. (2007) and Self et al. (2011). Self et al. (2011) amalgamated these samples, creating a 100-lake Russian chironomid-temperature calibration set. A subset of 81 samples witha more even distribution of lakes along the environmental gradientwas selected by deleting nineteen lakes withextremes of conductivity, pH or total phosphorus. This 81-lake subset was used to create a 2-component WA-PLS inference model that spans a Tjulrange of 8.8-19.0ºC. The Russian temperature inference model has a r2 of 0.92 and a RMSEP of 0.89ºC (Table 2).
Finally, subsets of the Norwegian and the Russian datasets were combined to form a calibrationset (Eurasian dataset; Table 1) that exploits the continentality gradient of the Eurasian continent (Fig. 1). Criteria to include lakes in the Eurasian dataset were temperature, continentality and geographic location, making sure that only a single lake was selected from a group of lakes with similar locations and climate regimes (Self et al. 2011). The Eurasian dataset contains a total of 149 samples and spans a continentality gradient (expressed as Continentality Index; see Self et al. (2011) for details) of 0-70 (mean = 26).It has anr2 of 0.73 and an RMSEP of 9.9 (Table 1).We applied all four transfer functions to the new Holocene chironomidrecord from Sokli (Electronic Supplementary Material), as well as to the MIS3 and MIS5d-c chironomid records (Engels et al. 2008a; 2010).
Chironomid records
For the Holocene sediment interval, subfossil chironomid remains were analysed applying standard methods (Brooks et al.2007), without KOHpretreatment which was unnecessary given the nature of the sediments. The identification of larval head capsules and the nomenclature follow Brooks et al. (2007). The minimum counting sum was set to 50 head capsules (after e.g. Heiri & Lotter 2001) and, subsequently, samples that failed to reach this level were rejected from further analyses. Chironomid preparation for MIS 3 and MIS 5d-c followed standard methods (Brooks et al. 2007). We refer to Engels et al. (2008a) and (2010) for more details.
Holocene chironomid record
A total of 88 chironomid taxa was identified in the Holocene record from lake Loitsana (Sokli) and an average of 58 head capsules (hcs) was counted in each sample (range: 50-72 hcs). A summarised chironomid sequence is presented in SI-Figure 1.
The lowermost samples in SI-Fig. 1 show high abundances of Tanytarsus lugens-type, a deep-water taxon (e.g. Engels and Cwynar 2011) which is associated with cold temperatures and oligotrophic conditions (Brooks et al. 2007). Heterotrissocladius grimshawi-type and Stictochironomus rosenschoeldi-type only occur in the samples below ~750 cm core depth and both these taxa are cold-stenothermic (Brooks et al. 2007).
Polypedilum nubeculosum-type shows high abundances between 750-700 cm core depth. Typically, P. nubeculosum-type is interpreted as an indicator of temperate climate conditions or littoral environments, which is in contrast to the habitat requirements of T. lugens-type, H. grimshawi-type, and S. roesenschoeldi-type. An increase in temperature or a decrease in lake depth might explain the change seen around 725 cm in the diagram. At approximately 700 cm core depth, a more diverse chironomid fauna establishes itself at the lake and the presence of taxa such as Cricotopus cylindraceus-type, Psectrocladius sordidellus-type, and Microtendipes pedellus-type probably indicate higher summer temperatures at the study site (Brodersen et al. 2001; Porinchu and MacDonald 2003).
Around 400 cm core depth, many of the taxa that were present between 700 and 400 cm disappear from the assemblage or decrease in abundance. Corynocera ambigua–type shows a strong increase in abundance at this point. This species has a Holarctic distribution and it commonly occurs in late-glacial sediments where it is most abundant during the cold Younger Dryas stadial (e.g. Walker and Matthewes 1989). However, in northeastern Siberia this species is exclusively found in sites south of the treeline (Porinchu and Cwynar 2000) and Porinchu and Cwynar (2002) use the presence of C. ambigua–type in a downcore record from Dologoye Ozero (northeastern Siberia) between 10-6.4 cal ka BP as an indication of warmer-than-present conditions. The interpretation of C. ambigua–type as an indicator of cold climate conditions has also been debated for European sites by Brodersen and Lindegaard (1999).The authors reported mass-occurrences of C. ambigua–type in modern sediments from shallow Danish lakes. As stated by Brodersen and Lindegaard (1999), the ecological adaptations of this species are complex and although the high abundances of C. ambigua–typein the Holocene record of Sokli (20-60%) might be related to decreasing temperatures or to lake-infilling, there are many other factors that can potentially explain the increase in the abundance by this taxon.
MIS 3 chironomid record
SI-Fig. 2a shows the chironomid taxa with a Hill’s N2 > 10 as calculated for samples in C2 (Juggins 2003) in the MIS3 record of Sokli; this cut-value was chosen to select a limited number of species. The chironomid record was initially dominated by Chironomus anthracinus-type, Procladius, and Parakiefferiella nigra-type (SI-Fig. 2a; Engels et al. 2008a). At 6.8 m core depth, Tanytarsus lugens-type reaches a maximum abundance of 40%, after which it decreases in abundance throughout the remainder of the record. Tanytarsus mendax-type, Polypedilum nubeculosum-type, and Cladotanytarsus mancus-type show high relative abundances between 6.6 and5.1 m core depth. Microtendipes pedellus-type shows abundances of 5-15% between 6.4 and 5.8 m core depth, whereas Corynocera ambigua–type only appears in the record at 6.1 m core depth, to reach an abundanceof approximately 15% for the remainder of the record. Engels et al. (2008a) concluded that a shallow lake with a diverse chironomid fauna was present at the study site throughout the period considered (although the lake might have been deeper during the initial phase of infilling) and that July air temperature was probably the most important factor influencing the composition of the fossil chironomid assemblages.
MIS 5d-c chironomid record
SI-Fig. 2b shows the chironomid taxa with a Hill’s N2 > 15 as calculated for samples in C2 (Juggins 2003) in the MIS 5d-c record of Sokli; this cut-value was chosen to select a limited number of species. Below 18.7 m core depth, only very few chironomid head capsules are present in the sediments and the ecological preferences of encountered taxa such as Gymnometriocnemus, Metriocnemus spp., and Limnophyes suggest very shallow lacustrine to semi-terrestrial conditions at the coring site (Engels et al. 2010). The concentration of chironomid remains abruptly increases at 18.7 m core depth and the presence of Tanytarsus lugens-type, T. mendax-type and Cladotanytarsus mancus-type indicates a change to a deeper water body at the coring site. Between 18.2 and 16.6 m core depth, Chironomus anthracinus-type, Cricotopus cylindraceus-type, and Metriocnemus eurynotus-type (not in SI-Fig. 2) increase to abundances of 8-20%. The combination of these taxa is striking, as M. eurynotus-type is reported to be indicative of semi-terrestrial or very shallow habitats (Cranston et al. 1983), whereas C. anthracinus-type is often abundant in relatively deep lakes in temperate regions. At 16.6 m core depth there is a large change in the lithology and pollen-content of the core (Engels et al. 2010; Helmens et al. 2012) possibly indicating the onset of MIS 5c. Microtendipes pedellus-type and Polypedilum nubeculosum-type show high abundance after this transition and, together with the presence of a high number of taxa that are associated with aquatic macrophytes and littoral habitats, they indicate shallow water and the presence of a quiet depositional environment at the coring site (Engels et al. 2010). The chironomid remains that are encountered between 14.6 and 14.1 m depth are derived from slabs of organic material that were intercalated between the clastic sediments that are deposited during this phase. They probably reflect local redeposition of peaty material in a floodplain environment. At 14.1 m core depth, the environment switches to a completely fluvial setting and no chironomid remains are encountered until 13.6 m core depth. The four uppermost samples of the record show a return to a lacustrine environment with little or no stream-influence. Engels et al. (2010) state that the chironomid assemblages encountered in the lowermost part of the record are deposited in a semi-terrestrial environment which do not have good analogues in the Norwegian calibration set, making these samples (amalgamated to a single sample in Fig. SI-2b) unsuitable for quantitative temperature inferences. However, the diverse chironomid fauna encountered between 18.6 and 14.6 m core depth reflect shallow lacustrine conditions and do provide for suitable samples for quantitative temperature reconstructions.
HOF, GLR and WA modeling.
Table SI-1 shows the number of occurrences, maximum abundance, Hill’s N2 and the response to July air temperature of selected chironomid taxa for the three temperature inference models presented in this study. Chironomid-temperature response data include HOF model types, Gaussian logit regression optima (GLR (ºC)) and Weighted-Averaging optima (WA (ºC).Temperature optima estimated by Gaussian logit regression (GLR) are less sensitive to uneven distribution of temperaturealong the gradient or to the length of the gradient in the calibration set than WA estimation (ter Braak and Looman 1986). As GLR assumes that all taxa have a Gaussian response to environmental variables, estimates of optimamay fall outside the environmental gradient covered by the calibration set (Birks et al. 1990; Birks 1995; Holmes et al. 2011). This may produce unrealistic optima for taxa that do not respond unimodally to the parameter of interest or that are at the edges of their range. Both unrealistically low optima (-2.4 or 0ºC for Micropsectra radialis-type) or unrealistically high optima (47.5ºC for Tanytarsus mendax-type) were produced for several of the chironomid taxa.
Tables
SI-Table 1: Number of occurrences (No occ), maximum abundance (Max), Hill’s N2 (N2) and the response to July air temperature of selected chironomid taxa for the three temperature inference models presented in this study. Chironomid-temperature response data include HOF model types (I: no response; II: sigmoidal increasing/ decreasing response; III: Plateau-ing response; IV: unimodal response; V: skewed unimodal response), Gaussian logit regression optima (GLR (ºC)) and Weighted-Averaging optima (WA (ºC)). Abbreviations: NP = Not present; x = present but too few occurrences for estimation
Figure captions
SI-Figure 1: Summary diagram of the Holocene chironomid record of the Sokli Borehole-series showing chironomid taxa with at least 2 occurrences of >2% abundance (%)
SI-Figure 2: a) Summary diagram of the MIS 3 Sokli chironomid record presented in Engels et al. (2008) showing the chironomid taxa with a Hill’s N2 in the MIS 3 record > 10; b) Summary diagram of the MIS 5d-c Sokli chironomid record presented in Engels et al. (2010) showing the chironomid taxa with a Hill’s N2 in the MIS 5d-c record > 15
References used in the Supplementary Information
Birks HJB (1995) Quantitative palaeoenvironmental reconstructions. In: Maddy D, Brew JJ (eds) Statistical modeling of quaternary science data. Quaternary ResearchAssociation,Cambridge, pp 161–254
Birks HJB, Line JM, Juggins S, Stevenson AC, ter Braak CJF(1990) Diatoms and pH reconstruction. Philos Trans RSoc Lond B 327:263–278
Brooks SJ, Birks HJB (2000) Chironomid-inferred late-glacial and early-Holocene mean July air temperatures for Kråkenes Lake, western Norway. J Paleolimnol 23: 77-89
Brooks SJ, Birks HJB (2004)The dynamics of Chironomidae (Insecta: Diptera) assemblages in response toenvironmental change during the past 700 years on Svalbard. J Paleolimnol 31: 483-498
Brooks SJ, Langdon PG, Heiri O (2007) The identification and use of Palaeoarctic Chironomidae larvae in palaeoecology. Quaternary Research Association Technical Guide 10, 276 pp
Brodersen KP, Lindegaard C (1999) Mass occurrence and sporadic distribution of Corynocera ambigua Zetterstedt (Diptera, Chironomidae) in Danish lakes. Neo- and palaeolimnological records. J Paleolimnol 22: 41-52
Brodersen KP, Odgaard BV, Vestergaard O, Anderson NJ (2001) Chironomid stratigraphy in the shallow and eutrophic Lake Søbygaards, Denmark: chironomid-macrophyte co-occurrence. Freshw Biol 46: 253-267
Cranston PS, Oliver DR, Saether OA (1983) The larvae of theOrthocladiinae (Diptera: Chironomidae) of the Holarctic region.Keys and Diagnoses. Entomol Scand 19: 149–291
Engels S, Cwynar LC (2011) Changes in fossil chironomid remains along a depth gradient: evidence for common faunal thresholds within lakes. Hydrobiologia 665: 15-38
Engels S, Bohncke SJP, Bos JAA, Brooks SJ, Heiri O, Helmens KF (2008a) Chironomid-based palaeotemperature estimates for northeast Finland during Oxygen Isotope Stage 3. J Paleolimnol 40: 49-61
Engels S, Helmens KF, Väliranta M, Brooks SJ, Birks HJB (2010) Early Weichselian (MIS-5d and 5c) temperatures and environmental changes as recorded by chironomids and macroremains at Sokli (northern Fennoscandia). Boreas 39: 689-704
Heiri O, Lotter AF (2001) Effect of low count sums on quantitative environmental reconstructions: an
example using subfossil chironomids. J Paleolimnol 26: 343-350
Helmens KF, Väliranta M, Engels S, Shala S (2012) Large shifts in vegetation and climate during the Early Glacial (MIS 5d-c) inferred from multi-proxy evidence at Sokli (Northern Finland). Quat Sci Rev 41, 22-38
Holmes N, Langdon PG, Caseldine C, Brooks SJ, Birks HJB (2011) Merging chironomid calibration sets: implications for palaeoclimate reconstructions. Quat Sci Rev 30: 2793-2804
Juggins S (2003) C2 user guide. Software for ecological and palaeoecological data analysis and visualisation. University of Newcastle, Newcastle upon Tyne, UK
Kumke T, Ksenofontova M, Pestryakova L, Nazarova L, Hubberten H.-W (2007) Limnological characteristics of lakes in the lowlands of Central Yakutia, Russia. J Limnol 66: 40-53
Luoto TP (2009) Subfossil Chironomidae (Insecta: Diptera) along a latitudinal gradient in Finland: development of a new temperature inference model. J Quat Sci 24: 150-158
Porinchu DF, Cwynar LC (2000)The distribution offreshwater Chironomidae (Insecta: Diptera) across treelinenear the lower Lena River, northeast Siberia. Arct Antarct Alp Res 32: 429-437
Porinchu DF,Cwynar LC (2002) Late-Quaternary history of midge communities and climate from a tundrasite near the lower Lena River, Northeast Siberia. J Paleolimnol 27: 59-69
Porinchu DF, MacDonald GM (2003) The use and application of freshwater midges (Chironomidae: insecta: diptera) in geographical research. Prog Phys Geogr 27: 378–422
Sarmaja-Korjonen K, Kultti S, Väliranta M, Solovieva N (2003) Mid-Holocene palaeoclimatic and palaeohydrological conditions in northeastern European Russia: a multi-proxy study of Lake Vankavad. J Paleolimnol 30: 415-426
Self AE, Brooks SJ, Birks HJB, Nazarova L, Porinchu D, Odland A, Yang H, Jones VJ (2011) The distribution and abundance of chironomids in high-latitude Eurasian lakes with respect to temperature and continentality: development and application of new chironomid-based climate-inference models in northern Russia. Quat Sci Rev 30: 1122-1141
Solovieva N, Jones VJ, Appleby PG, Kondratenok BM (2002) Extent, environmental impact and long-term trends in atmospheric contamination in the USA basin of east-European Russian Arctic. Water Air Soil Pollut 139: 237-260
Solovieva N, Jones VJ, Nazarova L, Brooks SJ, Birks HJB, Grytnes J-A, Appleby PG, Kauppila T, Kondratenok B, Renberg I, Ponomarev V (2005) Palaeolimnological evidence for recent climatic change in lakes from the northern Urals, arctic Russia. J Paleolimnol 33: 463-482
ter BraakCFJ, Looman CWN (1986) Weighted averaging,logistic regression and the Gaussian response model. Vegetatio 65: 3–11
Walker IR, Matthewes RW (1989) Early postglacial chironomid succession in southwestern British Columbia, Canada, and its paleoenvironmental significance. J Paleolimnol 2: 1–14