Table 1.— Sediment composition and catch data for longline sets used to establish the relationship be- 

 tween catch rates and sediment composition off Georgia. 



'Substrate samples were taken during a different cruise, however, samples were always taken within 

 65 d of each other. 



^A single depth measurement means that only one reading was taken during the longline set. This 

 depth is an approximation of longline depth. 



^Samples taken from the Delaware II. 



determine the percentage of sand and larger par- 

 ticles in the sample. The silt-clay fraction was obtain- 

 ed by subtraction. Replicate subsamples were taken 

 from six collections to establish the technique's preci- 

 sion. The mean difference in percent silt-clay frac- 

 tion among the six replicates was 2.5% (s = 1.4%). A 

 t-test for paired samples indicated that significant 

 differences did not exist among replicate determin- 

 ations for a given sample {t = 0.30, df = 5, P > 0.7). 



Statistical Analysis 



To determine the relationship between tilefish 

 444 



catch rate and sediment composition, we used the 

 silt-clay fraction of each substrate sample as an in- 

 dependent variable (X) and catch rate (i.e., tilefish 

 caught/100 hook-h per soak time) as the dependent 

 variable (Y). Exponential and polynomial regression 

 models were fit to the data using the SAS statistical 

 programs (SAS Institute Inc. 1982). The best poly- 

 nomial model was compared with the nonlinear ex- 

 ponential model using R^ as the criterion for model 

 performance. Similar patterns of variation were 

 observed in plots of residuals from all models, hence 

 R^ values are a reasonable criterion for model selec- 

 tion. 



