506 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



standard deviation from regression. These will 

 be substituted for the mean and standard devia- 

 tion in the graphical method of Hubbs and Hubbs. 

 (The range around a point on the regression line 

 is usually not available.) Unfortunately, some 

 samples are so small and the allometric growth so 

 marked that it is necessary to consider some 

 characters at only a single size and others merely 

 from the plotted points on the graph. 



CONVERSION OF LENGTHS 



Nearly all of the measurements must be con- 

 sidered in relation to other measurements. The 

 best standard is usually length of the fish, but 

 here a difficulty arises. Conrad and LaMonte 

 (1937), Gregory and Conrad (1939), and Morrow 

 (1952a) used body length, measured from the 

 snout to the base of the tail (standard length). 

 Brock, who measured the fish in the Hawaiian 

 market where the snouts are almost always cut 

 ofiF, measured the body length from the naris to 

 the fork of the tail. Measurement from the 

 posterior edge of the orbit to the fork of the tail 

 has been commonly used by Japanese scientists. 

 Thus, a preliminary requirement for examining 

 the characters is to be able to convert from one 

 length to another. We have done this by regres- 

 sion analysis for the three species of marlins, 

 audax, marlina, and ampla, on the basis of POFI 

 measurements. In each case we converted the 

 measurement given to fork length, which is defined 

 as the straight-line distance from the tip of the 

 snout to the tip of the center rays of the tail. 

 These conversions have been made from regres- 

 sion equations (appendix tables 3-A to 3-E, p. 550) 

 on the assumption that straight-line relationships 

 exist between the length measurements. Plots 

 of all measurements for each species have sub- 

 stantiated this assumption. 



CHARACTERS 

 Weight 



The general tendency for certain species of the 

 marlins to look heavier than others suggested that 

 it might be possible to separate tlie species on the 

 basis of the length-weight relation. Nichols and 

 LaMonte (1941) attempted this for the Pacific 

 marlins and they stated that for a given length their 

 striped marlin (audax) tended to weigh the least, 

 their silver marlin (marlina) more, and their black 

 marlin (ampla) most. When the relation is plotted 



(fig. 8) for the POFI measurements from the 

 central Pacific,' it is obvious that audax weighs 

 less than the other two which are much alike, and 

 that the length-weight relation might indeed be 

 useful for distinguishing individuals of less than 

 150 pounds. At lengths of about 300 cm. and 

 weights of around 300 pounds, however, there is a 

 great deal of overlap, as the weight of audax for a 

 given length then approaches that of marlina and 

 ampla. In the larger sizes, all three species are so 

 alike that it is impossible to distinguish individuals 

 on the basis of the length-weight relation. 



A comparison of POFI data with DFG material 

 and the published data (Gregory and Conrad, 

 1939; Conrad and LaMonte, 1937; Morrow 1952a) 

 in figure 9 shows that audax from all areas is lighter 

 at a given length than the other two species. There 

 is, however, a slightly greater overlap between 

 species at the 300-cm. size, especially for the POFI 

 material in which the specimens of audax were 

 slightly heavier at a given length than were those 

 from the other areas. 

 Greatest body depth 



When this measurement is plotted against fork 

 length a marked positive allometry is obvious 

 (fig. 5) . Both figures 2 and 1 0, in which all samples 

 are compared for given lengths, show that marlina 

 is deepest bodied, ampla intermediate, and audax 

 the most slender, but there is considerable overlap 

 between the species. The species marlina and 

 audax usually can be separated on the basis of body 

 depth, but ampla cannot clearly be distinguished 

 from either. Thus, the character is of httle value 

 for taxonomic purposes. Within each species there 

 is quite close agreement of the means; and the 

 relative position of the means is almost the same 

 as the mean weights of figure 9, which indicates 

 that the local populations that are heavier for a 

 giveiv length are also deeper bodied. 

 Head length 



Head length has not been used to separate the 

 species of marlins, but Gregory and Conrad (1939, 

 fig. 1) showed that ampla has a mean head length of 

 36 percent of the body length, whereas this ratio in 

 audax is about 39 and in marlina about 38. Such a 

 difference suggests some possibility of separating 

 the species with this character, and also because 

 most head parts are compared with head length, it 

 is desirable to examine our data for allometric 



• The data used for this graph include a few specimens not listed in the 

 appendix. 



