Fish growth and degree-days: Advice for selecting base temperatures in both within- and among-lake studies

Abstract

Ectotherm growth and development are largely a function of temperature. Within the range of thermal tolerance, growth and development increases linearly over a mid-range of temperatures. Degree-days (DD; °C;days) are a method of quantifying the thermal experience of an organism over the linear range and are an increasingly popular method for describing growth and development in fish. To limit degree-day calculations to temperatures within the linear range, a lower temperature threshold (T_o) is incorporated that defines the onset of growth. However, there is currently no convention for choosing T_o and the implications of choosing an inappropriate T_o are largely unknown. This uncertainty has resulted in a wide range of T_o values currently in use both among- and within-species. In this thesis I explore i) how T_o affects the ability of DD to explain within-population variation in fish growth, and ii) the effect of T_o on apparent growth among populations in thermally dissimilar environments. To address these objectives, I first use simulated data to determine the theoretical relationship between T_o and growth. I then demonstrate these theoretical relationships using immature length-at-age data from 8 species and 85 waterbodies in Minnesota USA and Ontario Canada. My results show that DD at low T_o are highly correlated and that these highly correlated low T_o values explain variation in growth equally well within a single population. Furthermore, growth rates among populations of thermally dissimilar habitat become increasingly dissimilar as T_o deviates from the T_o that minimizes the variation in growth rate among populations. This effect is especially apparent when the among-population range in thermal habitats is large. These findings suggest that, although T_o matters little when accounting for variation in growth within a single population, an inappropriate T_o can lead to the appearance of among-population differences in growth. When choosing T_o, I recommend estimating T_o using the Charnov and Gillooly (2003) “10°C rule” and then rounding to the nearest T_o standard (0, 5, 10 and 15°C). Choosing a T_o standard in this way will minimize the effect of T_o error on growth bias in among-population studies. In general, standardization i) simplifies T_o identification, ii) facilitates comparative studies, and ii) promotes the use of DD in future studies.