Fig. 2. Iterative HPLC separation (A–E) and behavioural bioassay results (F–J) of egg foam compounds. Eggs treated with HPLC fractions flanked by broken red lines give rise to predominately gregariously behaving hatchlings (corresponding red histograms; median P_gregarious=0.5). In the coarsest (i.e. primary) separation, all egg foam compounds were parsed into four fractions (A; F1–F4) and tested in a logistic regression behavioural bioassay (see text) for gregarizing properties (F). In the secondary run, components of fraction 3 were collected individually (B) and tested for bioactivity (G). The tertiary (C,H) and quaternary (D,I) runs were analogous with still greater specificity. Peak labelling is hierarchical and reflects peak origin in terms of previous fractions. The top row of histograms for each trial (F–I) shows the effect of known gregarizing compounds, sometimes reiterating findings of previous runs, applied to eggs as positive controls. Histogram bin widths=0.11 P_gregarious units. Median P_gregarious values are indicated by dotted grey lines. Peak areas (in black, E), representing F3.2.1.2 (relabelled peak X for clarity) and close analogues peaks Y and Z, were recorded for later use (see Fig. 4). Control behavioural classifications of known crowd- and solitary-reared individuals are provided (J). Insects from F were randomly drawn from four egg pods, (G) from four egg pods, (H) from nine egg pods, (I) from 13 egg pods and (J) (solitary-reared) from four egg pods.