How reliable are thermal constants for insect development when estimated from laboratory experiments?

Abstract

The concept of thermal time, where a linear relationship between developmental rate (d) and environmental temperature (T) is assumed for plants and poikilothermic animals (∂ = a + b · T ; a is the intercept, b the slope, and ∂ the model estimate of d) (e.g., Bonhomme, 2000), has been widely used as an approximation of physiological time (Trudgill et al., 2005). This approach has been especially common in studies of insect development (e.g., Hon e k & Kocourek, 1990; Hone k, 1996), as temperature plays a major role in the course of their life. Although a linear approximation of the generally non-linear relationship between developmental rate and temperature (e.g., Briere & Pracros, 1998) is valid only within a limited range of temperatures, the simplicity of the thermal time approach makes it very valuable in practice. Thermal constants [lower developmental threshold (LDT) and sum of effective temperatures (SET)] can be estimated from the assumed linear relationship, usually determined from laboratory observations of insect development at different constant temperatures. The LDT, which is defined as the temperature at which the development of insect ceases (d = 0) (e.g., Jaro8 ik et al., 2004; Trudgill et al., 2005), can be estimated as LDT = − a/b. The SET, which is defined as the temperature difference above LTD when total development (d = 1) is reached in a single day (e.g., Jaro8 ik et al., 2004; Trudgill et al., 2005), can be estimated as SET = 1/b (Figure 1). Because the values d = 0 and d = 1, which determine LDT and SET, are extremes in developmental rates and therefore outside the range of values used for the development of the linear model, the estimation of thermal constants is a typical example of extrapolation. This stimulated us to investigate how uncertain the estimates of thermal constants actually are; we base our investigation on laboratory experiments, due to the extrapolation inherent to the computational procedure.