COMPARISON OF METHODS USED TO CONVERT LENGTH-FREQUENCY DATA

Three methods to convert length-frequency data from Total to Fork leng th were compared. Datasets from two teleost species, the panga Pterogy mnus laniarius (with a forked tail) (n = 1500) and lesser gumard Cheli donichthys queketti (with an emarginate tail) (n = 649) were construct ed from observed total and fork length measurements to determine the a ccuracy of the different conversion methods. Results showed that for b oth tail profiles, the conventional method of simply shifting length c lasses using a linear relationship was the most inaccurate due to over lap (and subsequent adding) of the converted size classes. The two met hods which reallocated lengths between classes, the scale reassignment and proportional methods were the most accurate and resulted in the l east data distortion. Both of these methods were shown to have limitat ions, with the size of the dataset determining which was the most suit able technique. Datasets with small sample sizes (n < 300) were most a ccurately converted using the scale reassignment method whilst the pro portional method of conversion suited the larger datasets (n greater t han or equal to 649). (C) 1997 Elsevier Science B.V.