A comment on the expressed population signal

This commentary highlights a regularly occurring misinterpretation of the widely used expressed population signal (EPS). Based on thorough examination of the scientific article introducing EPS, I show that I) EPS was not meant to be a measure for the suitability of tree-ring data for climate reconstruction, and II) that the frequently used - but arbitrarily chosen - threshold of 0.85 was not meant to be used in combination with EPS. Instead, the less frequently used subsample signal strength (SSS) was intended for a respective application, i.e. as a measure of decreasing predictive power of transfer functions due to reductions in the sample size of underlying tree-ring series back in time. I conclude that classical transfer function quality statistics should be preferred over the erroneous application of EPS in the context of evaluating the suitability of dendrochronological data for climate reconstructions. (C) 2017 Elsevier GmbH. All rights reserved.