**2.4 Histology**

Standard hematoxylin and eosin staining was performed at Oregon State University on both phenotypic and control juvenile zebrafish in order to ascertain the nature of the jaw dysmorphogenesis. Sagittal jaw sections were cut 4 micron thick, fixed in paraformaldehyde and stained with Alizarin Red according to previously published protocols (Cailliet et al., 1986). Adult fish were cleared and stained with Alizarin Red dye in order to visualize bone and pigment structure. Fish were fixed for 3 days in 10% phosphate buffered formalin. Fish were first rinsed in deionized water, bleached in 3% hydrogen peroxide / 2% potassium hydroxide to remove pigment, rinsed a second time in DI water, trypsin digested for eight days (3375 U/ml in 30% Sodium Borate) until a majority of the spine was visible, soaked in 0.5% KOH, soaked in Alizarin Red for several hours and destained in 2% KOH overnight. Stained fish were then transferred stepwise into increasing glycerol concentrations (90% final) for imaging and long-term storage.

#### **2.5 Morphometrics**

The effects of diet on normal fish growth were determined with standard growth metrics. Body mass measurements were made on anesthetized (125 mg/l MS-222) fish. Water accumulated on the body surface can result in significant overestimation of body mass and so it was removed by gently blotting the animal with absorbent paper (Kimwipe). Individual fish were then weighed three times on a 5-place analytical balance (Mettler-Toledo, Inc., Columbus, Ohio) and the average calculated. Total lengths (tip of the nose to the tip of the tail fin) were obtained from still images using a calibrated measurement tool in Rincon 7.1 software (Optronics Inc., Goleta, California) or using a manual (dial) caliper. Craniofacial development was similarly assessed from still images of Alizarin Red bone stained adult fish by measuring the lengths and widths of the Meckel's and ceratohyal complexes in fish exposed to the different dietary regimens.

### **2.6 Statistics**

All statistics were performed in SigmaStat XI (Systat Software Inc., San Jose, California). Means were analyzed using Student two-tailed unpaired t-tests or nonparametric rank sum tests as needed. Probability (P) values of <0.05 were considered significant. Incidence levels were compared using a Yates-corrected Chi-square test.
