An inference model for mean summer air temperatures in the Southern Alps, New Zealand, using subfossil chironomids

Quantitative paleoclimate reconstructions from Southern Hemisphere mid-latitude sites are critical for identifying spatial and temporal patterns of abrupt climate change and for testing hypotheses about possible causal mechanisms. The analysis of chironomid distributions and environmental variables from 61 lakes was undertaken to develop a robust inference model for quantitative temperature reconstruction from sites in South Island, New Zealand. Detrended correspondence analysis (DCA) and canonical correspondence analysis (CCA) indicated that mean summer air temperature (SmT), chlorophyll a, conductivity, organic content of sediment as determined by loss-on-ignition, and mean annual precipitation account for a significant portion of variance in chironomid taxa distribution. SmT was the dominant variable. The strongest performing inference model, based on weighted averaging-partial least squares (WA-PLS), was developed from 60 lakes for SmT, having a root mean square error of prediction (RMSEPjack) of 1.27 °C and a coefficient of determination of 0.80 (r2jack). These results confirm that New Zealand chironomids may be used to provide robust estimates of mean summer air temperature and are a reliable method for quantitative paleoenvironmental reconstruction.