462 



Fishery Bulletin 103(2) 



D 



OPL 



PMXL 



E 



MXL 



DN 



D8L 



| — | = 10 mm 



Figure 1 



Diagrams of the five bluefish iPomatomus saltatrix) skull bones 

 used in this study: (A) premaxilla; (B) maxilla; (C) dentary; (D) 

 opercle; and (E) cleithrum. Bones came from a 701-mm (FL) fish 

 and are drawn to scale with respect to each other. The scale bar 

 represents 10 mm. Measurements for each bone were taken along 

 the longest axis and were given the following abbreviations: PMXL 

 (premaxilla length), MXL (maxilla length), DN (dentary length), 

 DBL (dentary body length), OPL (opercle length), CL (cleithrum 

 length). 



City, MD, north to Bayshore, NY. Upon retrieval, the 

 fork length (FL) and total length (TL) of each fish were 

 measured to the closest mm. The heads of the bluefish 

 were then removed by cutting approximately 5 cm behind 

 the pectoral girdle, and all heads were immediately 

 placed on ice. Samples were returned to the laboratory 

 and kept in a cool room on ice until the selected bones 

 could be extracted and measured (within 24 hours). 

 Bones were extracted by immersing the bluefish heads 

 in boiling water for a short period of time (between 30 

 and 180 seconds, depending on the size of the fish and 

 on the amount of musculature around the bones). The 

 dentary, maxilla, premaxilla, opercle, and cleithrum 

 were dissected from the left side of each fish and mea- 

 sured to the nearest 0.1 mm by using 0-150 mm dial 

 calipers. Measurements were taken linearly along the 



longest axis of each bone and the following abbrevia- 

 tions were used to indicate lengths: DBL (dentary body 

 length), DN (dentary length), OPL (opercle length), CL 

 (cleithrum length), MXL (maxilla length), and PMXL 

 (premaxilla length) (Fig. 1). In cases where left bones 

 were damaged, or it was determined that an accurate 

 measurement could not be retrieved, right-side bones 

 were measured in place of the damaged bones. 



Least squares regression analyses, which reveal the 

 relationship of each of the bone measurements to FL 

 and TL, were then conducted to generate predictive 

 equations. The strength of each of the correlations was 

 judged by both the r 2 values and by calculating the 

 mean percent prediction error for each model, where the 

 percent prediction error for a model (Sharf et. al., 1997) 

 is calculated by the following equation: 



