Figure Captions
Figure 1. Mean and standard errors for brood sex ratios for all cross types done. Solid points are parental selection lines (female-biased selection on left-hand side; male-biased selection lines on right-hand side); each point represents block by parent/backcross parent group. Open symbols are results from crosses where n> 5 for each cross type by block combination.
Figure 2. Observed (bars) and expected (points ± SE) distribution of brood sex ratio (left panel) and expected brood sex ratio (BSR) distribution (right panel) for parental, F1 and backcross generations of Tigriopus californicus crosses among male- and female-bias selection lines. On right panel median and 95% quantile limits are shown for expected distribution of variance using dashed and dotted vertical lines, respectively; solid lines indicate observed BSR variance.
Figure 3. Change in phenotypic variance for brood sex ratio, measured as standard deviation in qnorm(proportion male) for each generation during truncation selection for biased brood sex ratios. Data from (Alexander et al. 2014). Selection lines are: C = control (no selection), F = female-biased, M = male-biased; standard least-squares lines of best fit are given for each line with gray shading indicating 95% regions. After generation 7, brood sex ratio (proportion male) was 0.75 for M line, 0.35 for F line and 0.45 for C line; all had brood sex ratio = 0.51 at generation 1.
Appendix Figure 1 . Schematic of crossing design. Two blocks were used, each with four female-biased and four male-biased selection lines. Cells filled with a number indicate we have data for a cross between these two lines. Only offspring from crosses between selection line types were carried forward for backcrosses; shaded cells indicate crosses whose offspring were used in backcrosses. Siblings from F1 crosses were used in backcrosses with 3-4 male- and female-biased parental lines in the majority of cases.
Appendix Figure 2 . Result of changing prior distribution used from uninformative (grey line) to informative (black line) (top panel) on posterior distribution of heritability on the observed scale in sex of an individual (bottom panel). The informed prior slightly decreases the width of the posterior distribution (shown in black; grey line is density for uninformative prior). UP = uninformative prior; IP = informative prior.