Title: Stress-Induced Recombination and the Mechanism of Evolvability

Title: Stress-Induced Recombination and the Mechanism of Evolvability

Journal: Behavioral Ecology and Sociobiology

Authors: Weihao Zhong and Nicholas K. Priest

Affilitation: University of Bath

Email:

The focus of our study is on the relationship between offspring production and recombination in fruit flies. The most straightforward method for this analysis is to examine the slope of the linear regression between offspring production and recombination. However, the accuracy of each estimate of recombination, determined by the proportion of recombinant offspring produced by a single female, is greatly dependent on female fecundity. Recombination rate estimates of females with low fecundity are much less accurate than recombination rates of highly fecund females. We can account for low accuracy by using methods, such as logistic regression, which weight estimates by sample size. But, as is indicated in the text, this approach reduces our ability to test whether females with low fecundity have higher rates of recombination. Thus, we analyzed our data using both weighted and unweighted regressions. Here we illustrate the relationship between sample size and estimate accuracy by presenting Figure 2 and Figure 4 with 95 % confidence intervals for each estimate of recombination as calculated by the Wilson score interval (Agresti & Coull 1998).

Online Supplemental Reference:

Agresti, A. and Coull, B.A. (1998). Approximate is better than 'exact' for interval estimation of binomial proportions. The American Statistician 52: 119–126.

Figure 2 (with 95% confidence intervals). The effect of mating frequency on the relationships between offspring production and recombination rate within the Kr-Bc marker interval of chromosome 2 (Data from Priest et al. 2007). Each data point represents the estimated value of recombination with 95% confidence intervals (Wilson score interval) for a single female over a single three-day (day-13) egg laying period. Data points were offset by 0.2 and 0.4 offspring for clarity. Sample sizes: 41 (High mating, black square), 43 (Medium mating, grey diamond) and 42 (Low mating, light grey triangle). Fitted regression lines are calculated using general linear models on untransformed data.

Figure 4 (with 95% confidence intervals). The effect of brief and extreme changes in temperature on the relationships between female offspring production and recombination within the nub-pr marker interval of chromosome 2. Each data point represents estimated value of recombination with 95% confidence intervals (Wilsonscore interval) from a single female over a single three-day (day-13) egg laying period. Data points were offset by 0.2 and 0.4 offspring for clarity. Sample sizes: 7 (Heat shock, black square), 7 (Cold shock, grey diamond) and 10 (Control, light grey triangle). Fitted regression lines are calculated using general linear models on untransformed data.