Figure 3. Time and pattern of xanthophore differentiation is critical for pattern outcome.

(a–d) When Csf1 was induced and extra xanthophores developed early in D. rerio (c; stage AR+ ⁴²), stripes extended to the ventral margin of the flank, and secondary interstripes failed to develop similar to D. albolineatus (d). When induced late (b; SA+), stripes widths were similar to controls (a).

(e–h) Higher magnification images showing organized, interstripe iridophores (arrowheads in e, f) or their absence (g, h); scattered iridophores (arrows) are present in all panels. Yellow-orange xanthophores were segregated from melanophores in e, but intermingled with melanophores in f–h.

(i–k) bnc2 mutants fail to develop secondary interstripe iridophores and Csf1 overexpression either early or late resulted in a uniform pattern.

(l) Melanophore–xanthophore segregation was less in Csf1-overexpressing D. rerio and D. al-bolineatus (“wt,” non-transgenic siblings heat shock similarly) regardless of stripe presence (“C-l,” Csf1 late-overexpression) or absence (“C-e,” Csf1 early overexpression; alb, D. albolineatus). Common letters above bars indicate groups not significantly different (P>0.05) in Tukey Kramer post hoc comparisons; overall F_3,27=9.2, P<0.0005). Shown are means±s.e.m.; sample sizes are indicated within each bar.

(m) Melanophore frequencies at dorsal–ventral positions on the ventral flank (0.5=horizontal myoseptum, 1=ventral margin of myotomes). Vertical bars of the same color indicate positions not significantly different in Tukey Kramer post hoc comparisons (P>0.05): a distinct discontinuity in melanophore distribution, representing the stripe–interstripe boundary, was evident in control and Csf1 late-overexpressing D. rerio but not other backgrounds. For clarity only comparisons across positions 0.7–0.9 are shown. Shown are means±s.e.m.; sample sizes are indicated at the lower right of each plot.

Scale bars, 400 μm (a, for a–g), 400 μm (d); 60 μm (a′ for a′–c′); 60 μm (d′).