Diatom-based conductivity and water-level inference models from eastern Tibetan (Qinghai-Xizang) Plateau lakes

Climate in central Asia is dominated by the Asian monsoon. The varying impact of the summer monsoon across the Tibetan (Qinghai-Xizang) Plateau provides a strong gradient in precipitation, resulting in lakes of different salinity. Diatoms have been shown to indicate changes in salinity. Thus, transfer functions for diatoms and salinity or related environmental variables represent an excellent tool for paleoclimatic reconstructions in the Tibetan Plateau. Forty freshwater to hypersaline lakes (salinity: 0.1 to 91.7 g l^–1) were investigated in the eastern Tibetan Plateau. The relationship between 120 diatom taxa and conductivity, maximum water depth and major ions were analyzed using an indicator value approach, ordination and taxon response models. Canonical correspondence analysis indicated that conductivity was the most important variable, accounting for 10.8% of the variance in the diatom assemblages. In addition water depth and weathering were influential. Weighted Averaging (WA) and Weighted Averaging Partial Least Square (WA-PLS) regression and calibration models were used to establish diatom-conductivity and water depth transfer functions. An optimal two-component WA-PLS model provided a high jack-knifed coefficient of prediction for conductivity (r² _jack = 0.92), with a moderate root mean squared error of prediction (RMSEP_jack = 0.22), a very low mean bias (0.0003), and a moderate maximum bias (0.26). A WA model with tolerance downweighting resulted in a slightly lower r² _jack (0.89) for water depth, with RMSEP_jack= 0.26, mean bias = –0.0103 and maximum bias = 0.26.     

A comparison between diatom-based pH inference models using Artificial Neural Networks (ANN), Weighted Averaging (WA) and Weighted Averaging Partial Least Squares (WA-PLS) regressions

We explored the possibility of using artificial neural networks (ANN) to develop quantitative inference models in paleolimnology. ANNs are dynamic computer systems able to learn the relations between input and output data. We developed ANN models to infer pH from fossil diatom assemblages using a calibration data set of 76 lakes in Quebec. We evaluated the predictive power of these models in comparison with the two most commonly methods used in paleolimnology: Weighted Averaging (WA) and Weighted Averaging Partial Least Squares (WA-PLS). Results show that the relationship between species assemblages and environmental variables of interest can be modelled by a 3-layer back-propagation network, with apparent R² and RMSE of 0.9 and 0.24 pH units, respectively. Leave-one-out cross-validation was used to access the reliabilities of the WA, WA-PLS and ANN models. Validation results show that the ANN model (R² _jackknife = 0.63, RMSE_jackknife = 0.45, mean bias = 0.14, maximum bias = 1.13) gives a better predictive power than the WA model (R² _jackknife = 0.56, RMSE_jackknife = 0.5, mean bias = –0.09, maximum bias = –1.07) or WA-PLS model (R² _jackknife = 0.58, RMSE_jackknife = 0.48, mean bias = –0.15, maximum bias = –1.08). We also evaluated whether the removal of certain taxa according to their tolerance changed the performance of the models. Overall, we found that the removal of taxa with high tolerances for pH improved the predictive power of WA-PLS models whereas the removal of low tolerance taxa lowered its performance. However, ANN models were generally much less affected by the removal of taxa of either low or high pH tolerance. Moreover, the best model was obtained by averaging the predictions of WA-PLS and ANN models. This implies that the two modelling approaches capture and extract complementary information from diatom assemblages. We suggest that future modelling efforts might achieve better results using analogous multi-model strategies.     

A chironomid-based salinity inference model from lakes on the Tibetan Plateau

Previous studies have shown chironomids to be excellent indicators of environmental change and training sets have been developed in order to allow these changes to be reconstructed quantitatively from subfossil sequences. Here we present the results of an investigation into the relationships between surface sediment subfossil chironomid distribution and lake environmental variables from 42 lakes on the Tibetan Plateau. Canonical correspondence analysis (CCA) revealed that of the 11 measured environmental variables, salinity (measured as total dissolved solids TDS) was most important, accounting for 10.5% of the variance in the chironomid data. This variable was significant enough to allow the development of quantitative inference models. A range of TDS inference models were developed using Weighted Averaging (WA), Partial Least Squares (PLS), Weighted Averaging–Partial Least Squares (WA–PLS), Maximum Likelihood (ML), Modern Analogues Technique (MAT) and Modern Analogues Techniques weighted by similarity (WMAT). Evaluation of the site data indicated that four lakes were major outliers, and after omitting these from the training set the models produced jack-knifed coefficients of determination (r ²) between 0.60 and 0.80, and root-mean-squared errors of prediction (RMSEP) between 0.29 and 0.44 log₁₀ TDS. The best performing model was the two-component WA–PLS model with r ² _jack = 0.80 and RMSEP_jack = 0.29 log₁₀ TDS. The model results were similar to other chironomid-salinity models developed in different regions, and they also showed similar ecological groupings along the salinity gradient with respect to freshwater/salinity thresholds and community diversity. These results therefore indicate that similar processes may be controlling chironomid distribution across salinity gradients irrespective of biogeographical constraints. The performance of the transfer functions illustrates that chironomid assemblages from the Tibetan Plateau lakes are clearly sensitive indicators of salinity. The models will therefore allow the quantification of long-term records of past water salinity for lacustrine sites across the Tibetan Plateau, which has important implications for future hydrological research in the region.     

A Bayesian palaeoenvironmental transfer function model for acidified lakes

A Bayesian approach to palaeoecological environmental reconstruction deriving from the unimodal responses generally exhibited by organisms to an environmental gradient is described. The approach uses Bayesian model selection to calculate a collection of probability-weighted, species-specific response curves (SRCs) for each taxon within a training set, with an explicit treatment for zero abundances. These SRCs are used to reconstruct the environmental variable from sub-fossilised assemblages. The approach enables a substantial increase in computational efficiency (several orders of magnitude) over existing Bayesian methodologies. The model is developed from the Surface Water Acidification Programme (SWAP) training set and is demonstrated to exhibit comparable predictive power to existing Weighted Averaging and Maximum Likelihood methodologies, though with improvements in bias; the additional explanatory power of the Bayesian approach lies in an explicit calculation of uncertainty for each individual reconstruction. The model is applied to reconstruct the Holocene acidification history of the Round Loch of Glenhead, including a reconstruction of recent recovery derived from sediment trap data. The Bayesian reconstructions display similar trends to conventional (Weighted Averaging Partial Least Squares) reconstructions but provide a better reconstruction of extreme pH and are more sensitive to small changes in diatom assemblages. The validity of the posteriors as an apparently meaningful representation of assemblage-specific uncertainty and the high computational efficiency of the approach open up the possibility of highly constrained multiproxy reconstructions.     

Quantitative models for reconstructing catchment ice-extent using physical-chemical characteristics of lake sediments

The physical characteristics of surface sediments from a suite of pristine lakes on Signy Island, maritime Antarctic, were used to develop a quantitative link between catchment ice-extent and lake-sediment response. Percentage dry weight, median particle size, percentage loss-on-ignition and wet density of the lakes' surface sediments were the most significant variables explaining contemporary catchment ice-extent. Two independent reconstruction models – Partial Least Squares (PLS) and a Modern Analog Technique (MAT) – were applied to dated sediment cores at two sites on Signy Island. The validity of the reconstructions was tested against historical information on catchment ice-extent. With sufficiently high sedimentation rates and sampling resolution, the models can predict sub-decadal changes in ice-extent. The model results are best regarded as indicators of erosion resulting from meltwater activity in the catchment. Comparison of results with Twentieth Century climate records affirms the hypothesis that climatic warming is the most likely cause for the ice retreat observed on Signy Island during the last 40 yrs. Similar reconstruction models using these simple sedimentary measures could be developed for analogous locations in the Antarctic and in Arctic and Alpine regions.     

Diatom transfer-functions for quantifying past air temperature, pH and total organic carbon concentration from lakes in northern Sweden

The relationships between diatoms (Bacillariophyceae) in surface sediments of lakes and summer air temperature, pH and total organic carbon concentration (TOC) were explored along a steep climatic gradient in northern Sweden to provide a tool to infer past climate conditions from sediment cores. The study sites are in an area with low human impact and range from boreal forest to alpine tundra. Canonical correspondence analysis (CCA) constrained to mean July air temperature and pH clearly showed that diatom community composition was different between lakes situated in conifer-, mountain birch- and alpine-vegetation zones. As a consequence, diatoms and multivariate ordination methods can be used to infer past changes in treeline position and dominant forest type. Quantitative inference models were developed to estimate mean July air temperature, pH and TOC from sedimentary diatom assemblages using weighted averaging (WA) and weighted averaging partial least squares (WA-PLS) regression. Relationships between diatoms and mean July air temperature were independent of lake-water pH, TOC, alkalinity and maximum depth. The results demonstrated that diatoms in lake sediments can provide useful and independent quantitative information for estimating past changes in mean July air temperature (R² _jack = 0.62, RMSEP = 0.86 °C; R² and root mean squared error of prediction (RMSEP) based on jack-knifing), pH (R² _jack = 0.61, RMSEP = 0.30) and TOC (R² _jack = 0.49, RMSEP = 1.33 mg l^-1). The paper focuses mainly on the relationship between diatom community composition and mean July air temperature, but the relationships to pH and TOC are also discussed.     

How does taxonomic resolution affect chironomid-based temperature reconstruction?

The resolution achievable for chironomid identifications has increased in recent years because of significant improvements in taxonomic literature. However, high taxonomic resolution requires more training for analysts. Furthermore, with greater taxonomic resolution, misidentifications and the number of rare, poorly represented taxa in chironomid calibration datasets may increase. We assessed the effects of various levels of taxonomic resolution on the performance of chironomid-based temperature inference models (transfer functions) and temperature reconstruction. A calibration dataset consisting of chironomid assemblage and temperature data from 100 lakes was examined at four levels of taxonomic detail. The coarsest taxonomic resolution primarily represented identifications to genus or suprageneric level. At the highest level of taxonomic resolution, identification to genus level was possible for 37% of taxa, and identification below genus was possible for 60% of taxa. Transfer functions were obtained using Weighted Averaging (WA) and Weighted Averaging-Partial Least Squares (WA-PLS) regression. Cross-validated performance statistics, such as the root mean square error of prediction (RMSEP) and the coefficient of determination (r ²) between inferred and observed values improved considerably from the lowest taxonomic resolution level (WA: RMSEP 1.91°C, r ² 0.78; WA-PLS: RMSEP 1.59°C, r ² 0.86) to the highest taxonomic resolution level (WA: RMSEP 1.66°C, r ² 0.84; WA-PLS: RMSEP 1.41°C, r ² 0.89). Reconstructed July air temperatures during the Lateglacial period based on fossil chironomid assemblages from Hijkermeer (The Netherlands) were similar for all levels of taxonomic resolution, except the coarsest level. At the coarsest taxonomic level, reconstruction failed to infer one of the known Lateglacial cold episodes in the record. Also, the difference in reconstructed values based on lowest and highest taxonomic resolutions exceeded sample-specific estimated standard errors of prediction in several instances. Our results suggest that chironomid-based transfer functions at the highest taxonomic resolution outperform models based on lower-resolution calibration data. However, transfer functions of intermediate taxonomic resolution produced results very similar to models based on high-resolution taxonomic data. In studies that include analysts with different levels of expertise, inference models based on intermediate taxonomic resolution, therefore, might provide an alternative to transfer functions of maximum taxonomic detail in order to ensure taxonomic consistency between calibration datasets and down-core records produced by different analysts.     

Diatom-inference models for surface-water temperature and salinity developed from a 57-lake calibration set from the Sierra Nevada, California, USA

Physical, chemical, and biological data were collected from a suite of 57 lakes that span an elevational gradient of 1360 m (2115 to 3475 m a.s.l.) in the eastern Sierra Nevada, California, USA as part of a multiproxy study aimed at developing transfer functions from which to infer past drought events. Multivariate statistical techniques, including canonical correspondence analysis (CCA), were used to determine the main environmental variables influencing diatom distributions in the study lakes. Lakewater depth, surface-water temperature, salinity, total Kjeldahl nitrogen, and total phosphorus were important variables in explaining variance in the diatom distributions. Weighted-averaging (WA) and weighted-averaging partial least squares (WA-PLS) were used to develop diatom-based surface-water temperature and salinity inference models. The two best diatom-inference models for surface-water temperature were developed using simple WA and inverse deshrinking. One model covered a larger surface-water temperature gradient (13.7 °C) and performed slightly poorer (r² = 0.72, RMSE = 1.4 °C, RMSEP_jack = 2.1 °C) than a second model, which covered a smaller gradient (9.5 °C) and performed slightly better (r² = 0.89, RMSE = 0.7 °C, RMSEP_jack = 1.5 °C). The best diatom-inference model for salinity was developed using WA-PLS with three components (r² = 0.96, RMSE = 4.06 mg L^–1, RMSEP_jack = 11.13 mg L^–1). These are presently the only diatom-based inference models for surface-water temperature and salinity developed for the southwestern United States. Application of these models to fossil-diatom assemblages preserved in Sierra Nevada lake sediments offers great potential for reconstructing a high-resolution time-series of Holocene and late Pleistocene climate and drought for California.     

Chironomid-inferred late-glacial and early-Holocene mean July air temperatures for Kråkenes Lake, western Norway

A chironomid data-set calibrated to July air temperatures, based on 44 lakes in western Norway, is used to reconstruct mean July air temperatures from late-glacial and early-Holocene fossil chironomid assemblages at Kråkenes Lake. The calibration function is based on Weighted Averaging Partial Least Squares regression and has a root mean square error of prediction (RMSEP) of 1.13 °C, a r² of 0.69, and a maximum bias of 2.66 °C. All these statistics are based on leave-one-out cross-validation. A calibration function based on summer surface-water temperatures has a poorer performance (RMSEP = 2.22 °C, r² = 0.30, maximum bias = 5.29 °C). The reconstructed July air temperatures at Kråkenes rise to 10.5 °C soon after deglaciation, are about 11.5 °C in the Allerød, decrease to 9.5-10 °C in the Younger Dryas, and rise rapidly within 15 yrs to 11.5 °C at the onset of the Holocene. There is a two-step rise to 13 °C or more in the early-Holocene. The likely over-estimation of Younger Dryas temperatures and under-estimation of early-Holocene temperatures probably result from the limited temperature range represented by the existing calibration set. The data set is currently being expanded to include lakes with warmer air temperatures (> 14 °C) and with colder air temperatures (< 8 °C).     

Diatom-based models for inferring past water chemistry in western Ugandan crater lakes

Diatom surface sediment samples and corresponding water chemistry were collected from 56 lakes across a natural conductivity gradient in western Uganda (reflecting a regional climatic gradient of effective moisture) to explore factors controlling diatom distribution. Here we develop a regional training set from these crater lakes to test the hypothesis that this approach, by providing more appropriate and closer analogues, can improve the accuracy of palaeo-conductivity reconstructions, and so environmental inferences in these lake systems compared to larger training sets. We compare this output to models based on larger, but geographically and limnologically diverse training sets, using the European Diatom Database Initiative (EDDI) database. The relationships between water chemistry and diatom distributions were explored using canonical correspondence analysis (CCA) and partial CCA. Variance partitioning indicated that conductivity accounted for a significant and independent portion of this variation. A transfer function was developed for conductivity (r _jack ² ?=?0.74). Prediction errors, estimated using jack-knifing, are low for the conductivity model (0.256 log₁₀ units). The resulting model was applied to a sedimentary sequence from Lake Kasenda, western Uganda. Comparison of conductivity reconstructions using the Ugandan crater lake training set and the East Africa training set (EDDI) highlighted a number of differences in the optima of key diatom taxa, which lead to differences in reconstructed values and could lead to misinterpretation of the fossil record. This study highlights issues of how far transfer functions based on continental-scale lake datasets such as the EDDI pan-African models should be used and the benefits that may be obtained from regional training sets.     

A model for inferring dissolved organic carbon (DOC) in lakewater from visible-near-infrared spectroscopy (VNIRS) measures in lake sediment

We developed an inference model to infer dissolved organic carbon (DOC) in lakewater from lake sediments using visible-near-infrared spectroscopy (VNIRS). The inference model used surface sediment samples collected from 160 Arctic Canada lakes, covering broad latitudinal (60–83°N), longitudinal (71–138°W) and environmental gradients, with a DOC range of 0.6–39.6 mg L⁻¹. The model was applied to Holocene lake sediment cores from Sweden and Canada and our inferences are compared to results from previous multiproxy paleolimnological investigations at these two sites. The inferred Swedish and Canadian DOC profiles are compared, respectively, to inferences from a Swedish-based VNIRS-total organic carbon (TOC) model and a Canadian-based diatom-inferred (Di-DOC) model from the same sediment records. The 5-component Partial Least Squares (PLS) model yields a cross-validated (CV) R_CV² R_{CV}^{2}  = 0.61 and a root mean squared error of prediction (RMSEP _CV ) = 4.4 mg L⁻¹ (11% of DOC gradient). The trends inferred for the two lakes were remarkably similar to the VNIRS-TOC and the Di-DOC inferred profiles and consistent with the other paleolimnological proxies, although absolute values differed. Differences in the calibration set gradients and lack of analogous VNIRS signatures in the modern datasets may explain this discrepancy. Our results corroborate previous geographically independent studies on the potential of using VNIRS to reconstruct past trends in lakewater DOC concentrations rapidly.     

Diatom-based models for reconstructing past water quality and productivity in New Zealand lakes

The trophic status of lakes in New Zealand is, on average, low compared to more densely populated areas of the globe. Despite this, trends of eutrophication are currently widespread due to recent intensification in agriculture. In order to better identify baseline productivity and establish long-term trends in lake trophic status, diatom-based transfer functions for productivity-related parameters were developed. Water quality data and surface sediment diatom assemblages from 53 lakes across the North and South Islands of New Zealand were analysed to determine species responses to the principal environmental gradients in the data set. Repeat sampling of water chemistry over a 12-month period enabled examination of species responses to annual means as well as means calculated for stratified and mixed periods. Variables found to be most strongly correlated with diatom species distributions were chlorophyll a (Chl a), total phosphorus (TP), dissolved reactive phosphorus (DRP), ionic concentration (measured as electrical conductivity (EC)) and pH. These variables were used to develop diatom-based transfer functions using weighted averaging regression and calibration (simple, tolerance down-weighted and with partial least squares algorithm applied). Overall, models derived for stratified means were weaker than those using annual or isothermal means. For specific variables, the models derived for the isothermal mean of EC (WA-tol r²_jack = 0.79; RMSEP = 0.15 log₁₀ S cm^–1),the annual mean of pH (WA r²_jack = 0.72; RMSEP = 0.25 pH units) and the isothermal mean of Chl a (WA r²_jack = 0.71; RMSEP = 0.18 log₁₀ mg m^–3 Chl a) performed best. The models derived for TP were weak in comparison (for the annual mean of TP: WA r²_jack = 0.50; RMSEP = 0.24 log₁₀ mg m^–3 TP) and residuals on estimates for this model were correlated with several other water quality variables, suggesting confounding of species responses to TP concentrations. The model derived for the isothermal mean of DRP was relatively strong (WA-tol r²_jack = 0.78; RMSEP = 0.17 log₁₀ mg m^–3 DRP); however, residual values for this model were also found to be strongly correlated with several other water quality variables. It is concluded that the poor performance of the TP and DRP transfer functions relative to that of the Chl a model reflects the coexistence of nitrogen and phosphorus limitation within the lakes in the data set. In spite of this, the suite of transfer functions developed from the training set is regarded as a valuable addition to palaeolimnological studies in NewZealand.     

Chironomid-based water depth reconstructions: an independent evaluation of site-specific and local inference models

Water depth is an important environmental variable that explains a significant portion of the variation in the chironomid fauna of shallow lakes. We developed site-specific and local chironomid water-depth inference models using 26 and 104 surface-sediment samples, respectively, from seven kettlehole lakes in the Plymouth Aquifer, southeast Massachusetts, USA. Our site-specific model spans a depth gradient of 5.6?m, has an $ {\text{r}}_{\text{jack}}^{2} $ of 0.90, root mean square error of prediction (RMSEP) of 0.5?m and maximum bias of 0.7?m. Our local model has a depth gradient of 11.7?m, an $ {\text{r}}_{\text{jack}}^{2} $ of 0.71, RMSEP of 1.6?m and maximum bias of 2.9?m. Principal coordinates of neighbourhood matrices (PCNM) analysis showed that there is no influence of spatial autocorrelation on the site-specific model, but PCNM variables explained a significant amount of variance (4.8%) in the local model. This variance, however, is unique from the variance explained by water depth. We applied the inference models to a Holocene chironomid record from Crooked Pond, a site for which multiple, independent palaeohydrological reconstructions are available. The chironomid-based reconstructions are remarkably similar and show stable water depths of ~5?m, interrupted by a 2-m decrease between 4,200 and 3,200?cal a BP. Sedimentological evidence of water level fluctuations at Crooked Pond, obtained using the so-called Digerfeldt approach, also shows a drop in water depths around that time. The period of reconstructed lower water levels coincides with the abrupt decline in moisture-dependent hemlock in this region, providing further evidence for this major palaeohydrological event. The site-specific model has the best performance statistics, but the high percent abundance of fossil taxa from the long core that are absent or rare in the training set makes the site-specific reconstruction unreliable for the period before 4,400?cal a BP. The fossil taxa are well represented in the local model, making it the preferred inference model. The strong similarity between the chironomid-based reconstructions and the independent palaeohydrological records highlights the potential for using chironomid-based inference models to determine past lake depths at sites where temperature was not an influencing factor.     

New Zealand chironomids as proxies for human-induced and natural environmental change: Transfer functions for temperature and lake production (chlorophyll a)

The analysis of chironomid taxa and environmental datasets from 46 New Zealand lakes identified temperature (February mean air temperature) and lake production (chlorophyll a (Chl a)) as the main drivers of chironomid distribution. Temperature was the strongest driver of chironomid distribution and consequently produced the most robust inference models. We present two possible temperature transfer functions from this dataset. The most robust model (weighted averaging-partial least squares (WA-PLS), n = 36) was based on a dataset with the most productive (Chl a > 10 μg l⁻¹) lakes removed. This model produced a coefficient of determination () of 0.77, and a root mean squared error of prediction (RMSEP_jack) of 1.31°C. The Chl a transfer function (partial least squares (PLS), n = 37) was far less reliable, with an of 0.49 and an RMSEP_jack of 0.46 Log₁₀μg l⁻¹. Both of these transfer functions could be improved by a revision of the taxonomy for the New Zealand chironomid taxa, particularly the genus Chironomus. The Chironomus morphotype was common in high altitude, cool, oligotrophic lakes and lowland, warm, eutrophic lakes. This could reflect the widespread distribution of one eurythermic species, or the collective distribution of a number of different Chironomus species with more limited tolerances. The Chl a transfer function could also be improved by inputting mean Chl a values into the inference model rather than the spot measurements that were available for this study.     

An expanded surface-water palaeotemperature inference model for use with fossil midges from eastern Canada

Using an expanded surface sample data set, representing lakes distributed across a transect from southernmost Canada to the Canadian High Arctic, a revised midge-palaeotemperature inference model was developed for eastern Canada. Modelling trials with weighted averaging (with classical and inverse deshrinking; with and without tolerance downweighting) and weighted averaging partial least squares (WA-PLS) regression, with and without square-root transformation of the species data, were used to identify the best model. Comparison of measured and predicted temperatures revealed that a 2 component WA-PLS model for square-root transformed percentage species data provided the model with the highest explained variance (r =0.88) and the lowest error estimate (RMSEP _jack =2.26 °C). Comparison of temperature inferences based on the new and old models indicates that the original model may have seriously under-estimated the magnitude of late-glacial temperature oscillations in Atlantic Canada. The new inferences suggest that summer surface water temperatures in Splan Pond, New Brunswick were approximately 10 to 12 °C immediately following deglaciation and during the Younger Dryas. During the Allerod and early Holocene, surface water temperatures of 20 to 24 °C were attained. The new model thus provides the basis for more accurate palaeotemperature reconstructions throughout easternmost Canada.     

Development of paleolimnological inference models for pH, total nitrogen and specific conductivity based on planktonic diatoms

Canonical correspondence analysis (CCA was used to explore and identify statistically significant relationships between the distributions of planktonic diatoms and the physical and chemical properties of 50 Connecticut lakes. Six variables (pH, total nitrogen, calcium, sulfate, potassium and chlorophyll- a concentrations) were found to be significantly correlated with either or both of the first two extracted axes. The pH and calcium concentration, and to a lesser extent total nitrogen concentrations, were the most important variables controlling the distributions of planktonic diatoms in this suite of lakes. Paleolimnological inference models were developed for pH, total nitrogen (TN) and specific conductivity. Weighted averaging with (WAtol) and without (WA) tolerance downweighting, with and without bootstrap resampling techniques, and using either classical or inverse deshrinking methods were used to develop inference models for each variable. The pH and TN yielded sufficiently high 1/2 ratios and a highly significant first (constrained) axis when entered as single variables in both constrained and partially constrained CCA analyses, supporting the idea that reliable inference models could be developed for these variables. The r2 and RMSE of prediction values ranged from 0.73 to 0.86 and 0.37 to 0.6, respectively for pH, and from 0.4 to 0.64 and 59 g/l to 95 g/l, respectively for TN. Inference models for specific conductivity also yielded significant goodness-of-fit statistics. However, because specific conductivity was removed from the CCA analysis due to its high variance inflation factor and did not yield a significant relationship when entered as the sole variable in a partial constrained CCA, inference models for this variable will probably not yield any additional environmental information. The use of only planktonic diatoms in construction of inference models is discussed.     

An analysis of phytolith assemblages for the quantitativereconstruction of late Quaternary environments of the Lower Taieri Plain,Otago, South Island, New Zealand I. Modern assemblages and transferfunctions

An analysis of modern phytolith assemblages is presented.Phytolith assemblages were studied in modern surface soils and sediments of 28sites from east Otago, New Zealand, within a range of vegetation types andmicroclimates. No simple distinction could be made between vegetation types onthe basis of phytolith assemblage composition. A Principal Components Analysis(PCA) of the phytolith data set revealed that festucoid, chloridoid andspherical phytolith morphotypes formed strong associations with sites fromwetland, grassland, and forest vegetation types, respectively. Moreimportantly, a comparison of sample replicates from each field site using Squared ChordDistance (SCD) assemblage analysis showed that wetland and grassland sitestended to produce more internally consistent phytolith assemblages than forestsites. Environmental variables including pH, conductivity, altitude,precipitation and temperature were also gathered for each site. The ability ofeach environmental variable to reflect variance in the entire phytolithdata set was estimated by a series of Redundancy Analyses (RDA) with MonteCarlo permutation tests of statistical significance. After a forward selectionprocess, transfer functions were generated using Partial Least Squares (PLS)regression and calibration with jack-knife validation. The final transferfunctions have root mean squared errors of prediction for pH (0.47), logconductivity (0.38 S cm), average annual precipitation (63mm), and average annual (0.28 °C), spring (0.38 °C) andautumn temperature (0.41 °C); the smallest group of environmental variablesexplaining the most variance in the modern phytolith data set. The most usefultransfer functions for application to fossil phytolith data andpaleoenvironmental interpretation are pH, log conductivity and annualprecipitation. The relationship between changes in pH and annual precipitationand phytolith assemblage composition found in this study presents aprima facie relationship with the potential to providedirect proxies for soil weathering and indirectly for paleoenvironmentalreconstruction.     

The potential of site-specific and local chironomid-based inference models for reconstructing past lake levels

We examined the relationship between three key environmental variables (water depth, loss-on-ignition, and bottom-water temperature) and fossil chironomid distributions sampled from within-lake gradients in three small, moderately deep (18–35 m), maar lakes on St Michael Island, western Alaska. Site-specific (one lake, 29 samples) and local (three lakes, 87 samples) inference models for reconstructing water depth were developed using partial least squares regression and calibration. These models and a previously published regional model (136 lakes, one central-lake sample from each) are used to infer water depths from 78 fossil samples spanning the last ~30,000 ¹⁴C years B.P. at Zagoskin Lake. Although the site-specific [r ² _boot = 0.90, root mean square error of prediction (RMSEP) = 1.76] and local (r _boot² = 0.68, RMSEP = 4.36) inference models have better performance statistics than the regional model, few clear trends among all three models exist in the lake-level reconstruction. We propose that multiple, within-lake sampling of gradients can be used to improve the performance statistics of water-depth transfer functions and ultimately reconstruct paleohydrology in regions known to exhibit large fluctuations in moisture balance through time given that: (1) adequate analogs are established and (2) taphonomic processes important to benthic invertebrate remains are more fully understood.     

Environmental controls on modern chironomid faunas from NW Iceland and implications for reconstructing climate change

Reconstructing climate change quantitatively over millennial timescales is crucial for understanding the processes that affect the climate system. One of the best methods for producing high resolution, low error, quantitative summer air temperature reconstructions is through chironomid analyses. We analysed over 50 lakes from NW and W Iceland covering a range of environmental gradients in order to test whether the distribution of the Icelandic chironomid fauna was driven by summer temperature, or whether other environmental factors were more dominant. A range of analyses showed the main environmental controls on chironomid communities to be substrate (identified through loss-on-ignition and carbon content) and mean July air temperature, although other factors such as lake depth and lake area were also important. The nature of the Icelandic landscape, with numerous volcanic centres (many of which are covered by ice caps) that produce large quantities of ash, means that relative lake carbon content and summer air temperature do not co-vary, as they often do in other chironomid datasets within the Arctic as well as more temperate environments. As the chironomid–environment relationships are thus different in Iceland compared to other chironomid training sets, we suggest that using an Icelandic model is most appropriate for reconstructing past environmental change from fossil Icelandic datasets. Analogue matching of Icelandic fossil chironomid datasets with the Icelandic training set and another European chironomid training set support this assertion. Analyses of a range of chironomid-inferred temperature transfer functions suggest the best to be a two component WA-PLS model with r ² _jack = 0.66 and RMSEP = 1.095°C. Using this model, chironomid-inferred temperature reconstructions of early Holocene Icelandic sequences show the magnitude of temperature change compared to contemporary temperatures to be similar to other NW European chironomid sequences, suggesting that the predictive power of the model is good.     

Development of a diatom-based model for inferring total phosphorus in southeastern Australian water storages

Diatom-based transfer functions for inferring epilimnetic total phosphorus (TP) have been developed from a data set of 33 southeastern Australian water storages. Regular institutional monitoring of these sites has allowed comparison of models developed from TP data covering different time periods. A model based on mean annual TP performs better than models derived from winter maximum TP, spring minimum TP or TP nearest the time of diatom sampling. A mean annual TP model (WA-PLS 2 component) has a jack-knifed diatom-inferred versus measured TP correlation coefficient (r ² _jack) of 0.69 and a root-mean-square-error of prediction (RMSEP) of 0.246 log₁₀g TP l^–1, while alternative models have RMSEP > 0.27. Deletion of two samples with uncharacteristic species composition and environmental conditions improved performance of the mean annual TP model (r ² _jack= 0.74; RMSEP = 0.233 log₁₀g TP l^–1). Comparison with other published diatom-TP calibration models indicates that this model performs relatively well, with possible contributing factors including the extensive characterisation of TP (with an average 15 determinations making up the annual mean) and the dominance of planktonic diatoms in most sites. Downcore application of the model will allow the reconstruction of reservoir nutrient histories since commissioning, and thus provide a basis for understanding and management of reservoirs.