Fig. 4

Sex ratios increase with foundress numbers, but not enough. Observed (open shapes) and expected (solid shapes) sex ratio variation in 36 data sets from 25 fig wasp species. Within each graph, each shape refers to a different species. We obtained 95% CIs in five ways indicated by the colours of the vertical lines and in order of preference as data allowed: (1) In black, e^a/(e^a + 1) where a is given by the generalized linear model coefficient ± 1.96x(se of the coefficient). A model was fitted for each foundress number and a quasibinomial statistical model was used if data was overdispersed. (2) In brown e^a/(1 + e^a) where a is the mean of log(p/(1-p)) ± 1.96x(se of the mean). (3) In purple, (sin(a))² where a is the mean ± 1.96x(se) of arcsine transformed data. (4) In blue, mean sex ratio ± 1.96x(se). (5) In green, given. a Blastophaga nipponica (circles) and B. psenes (squares), b Ceratosolen capensis (circles), C. fusciceps (squares) and C. galili (rombi), c C. gravelyi (circles) and C. marchali (squares), d C. solmsi (circles) and Elisabethiella baijnathi (squares), e Eupristina koningsbergeri (circles, mean sex ratio and 95% CI of 5 foundresses is not plotted and is 0.46–0.47–0.49), Kradibia tentacularis (squares), f Pegoscapus aerumnosus (circles), P. franki (squares) and P herrei (rombi), g P. jimenezi (circles), P. longiceps (squares), P. lopesi (rombi), and P. piceipes (triangles), h P. silvestrii (circle), P. tonduzi (squares) and Pegoscapus sp. ex F. crocata (rombi), i Platyscapa awekei (circles), Tetrapus ecuadoranus (squares), Tetrapus sp. ex F. insipida (rombi) and Valisia javana (triangles). Expected values were calculated using Eq. (2). Details in Additional files 1 and 2 Supplementary text and figures. Genera are colour coded as in later figures