Fig. 4

Relationship between average paternal care intensity and the change in interindividual density following the mechanical predator disturbance for (a) own and (b) adopted offspring from the opposing risk treatment, respectively. Filled dots visualize instances where high-risk males provided care, empty dots instead refer to care by low-risk males. Shaded areas (95% confidence intervals) and regression lines for the treatment-independent correlation (thick solid line) were estimated from mixed-effect models with treatment (low-risk/high-risk) as covariate and caring individual ID nested in shoal family origin as random intercept. Regression lines for treatment-specific correlations (low-risk care: dashed lines, high-risk care: dot-dashed lines) are shown without confidence intervals to avoid visual clutter. The dotted line is the zero referent and represents no change in shoaling density