Saunders KM, Grosjean M, Hodgson DA (2013) A 950 yr temperature reconstruction from Duckhole Lake, southern Tasmania, Australia. The Holocene 23(6) 771–783. DOI: 10.1177/0959683612470176.

Page 771: Abstract

The calibration model was based on a negative, not positive, relationship between trough area (R_570/630) and temperature. The corrected model is: R = −0.56, p_auto < 0.05, root mean squared error of prediction (RMSEP) = 0.21°C, five-year filtered data, calibration period 1911–2000.

Page 774 ‘Reflectance spectroscopy’

The correlations in this paragraph are negative (not positive as stated in the original paper). The paragraph should read as follows:

VIS-RS data are summarized in Figure 3. R_570/630 and R_660/670 fluctuated throughout the core with a slight overall decreasing trend towards the surface. R_570/630 was positively correlated to R_660/670 (R = 0.5, p < 0.05). R_650;700 and RABD_660;670 were highly significantly correlated (R = −0.85, p < 0.05). R_650;700 was weakly correlated to R_660/670 (R = 0.36). R_650;700, RABD_660;670 and R_660/670 were significantly correlated with water content (R = −0.83, 0.60 and –0.53, respectively, p < 0.05). Once corrected for water content, the negative correlation between R_650;700 and R_660/670 increased (R = −0.41, p < 0.05), while there was no longer a correlation between RABD_660;670 and R_660/670.

Page 775 ‘Calibration with meteorological data’

The R values referred to should be negative:

All VIS-RS-derived proxies (R_650;700, RABD_660;670, R_660/670 and R_570/630) were calibrated to monthly and annual temperature (mean, maximum and minimum) and precipitation (total) for the period AD 1911–2000. R_650;700 was most highly and significantly correlated to mean annual (January–December) temperature (R = −0.84, p_auto < 0.05, p_auto after detrending < 0.05).

The correlation between R_650;700 and annual temperature was highly significant after applying a five-year triangular filter (non-detrended: R = −0.85, p_auto < 0.001; Figure 4a) and remained significant when detrended (R = −0.29, p_auto < 0.05; Figure 4b).

Consequently, as R_650;700 (reflecting chlorophylls and derivatives) correlated best to temperature, it was used to develop an annual temperature calibration model for Duckhole Lake: R = −0.56, p_auto < 0.05, RMSEP = 0.21°C, where both R and p were adjusted for autocorrelation.

Page 778 Figure 4

The y axes of the lower panels in 4(a) and 4(b) were not correct. The corrected figure is above: Figure 4.

Comparison of R_650;700 and annual mean temperature: (a) non-detrended R_650;700 and mean annual temperature; (b) detrended data. All data are five-year triangular filtered. Note: reversed y axes in the lower panels (proxy data) of (a) and (b).