NOTE Cope: Inlraspecific life history patterns in sharks 



313 



ias and P. glauca) to demonstrate the calculation of 

 life history trait values for areas with missing values. 

 These species were chosen as examples because they 

 are taxonomically different, are found in most of the 

 area designations, and are represented by at least one 

 pairwise comparison for each life history trait. Further 

 investigation of the predictive capacity of these models 

 to fecundity, size at maturity, and maximum size for 

 the two species was performed by cross-validation; for 

 each life history trait, S. acanthias or P. glauca data for 

 one area were removed, the models were re-estimated, 

 and predicted values for the newly missing area were 

 calculated. Model fits to age at maturity, longevity, and 

 male P. glauca size at maturity were not explored with 

 cross validation because these data lacked the sufficient 

 sample size needed to calculate the species effect once 

 one area was removed (at least two remaining areas 

 were needed). 



Results 



Model structure 



For each life history trait that contained gender as a 

 factor, the following final GLM model with a gamma 

 distribution and a log link was selected (i.e., had the 

 lowest AIC value): 



In yA.sp.-g = In Pa + In P, + In A,,. + In P, 

 where /3^ = area effect; 



(1) 



P., 



yA+gxsp 



= gender effect; 



= species effect; 



= gender and species interaction effect. 



= value of the response variable (age, litter 

 size, or length) for each area, species, and 

 gender, accounting for the gender-species 

 interaction. 



