Winter-duration treatments used in our experiment were based on historical (1994–2010) field measurements of water temperature from a central Lake Erie water intake located near Cleveland, OH, USA (41° 32′ 53′′ N, 81° 44′ 60′′ W). The short winter duration (52 days) was similar to the number of days below 5 °C recorded during winter 1999 (N=59 d below 5 °C), which was, until 2012, the warmest winter on record for Ohio (1895–2013)56. Our long winter duration (107 days) was equal to the mean number of days below 5 °C observed during 1994–2010. The duration of spawning is shown for each treatment. Labels indicate the dates and water temperatures during which females spawned. Despite earlier arrival of suitable spawning temperatures (8–14 °C within grey shading18; this study) following a short winter, females did not initiate spawning at these ‘normal’ temperatures in our experiment, and instead spawned at warmer temperatures during the ‘normal’ spawning period (that is, mid-April through May14,18).