314 



Fishery Bulletin 104(2) 



This model is also biologically realistic because it in- 

 cludes the possibility that males in one species may be 

 smaller than females and vice versa. For fecundity, the 

 following model was chosen: 



ln>'A«p = lnft4 + ln/3,p. 



(2) 



An assessment of Cook's statistic for all models revealed 

 no evidence of any highly influential data points. A 

 subsequent analysis of residuals by species indicated, 

 in one case, a potential departure from the assumption 

 that both genders of all species in all areas have the 

 same variance. Highest residuals were reported for age 

 at maturity in S. acanthias. Whether these high residual 

 values are truly reflective of the species or an artifact 

 of low sample size is unknown, so I carried forth with 

 the analysis using the above models. 



Area effects 



Despite the low and unbalanced numbers of comparisons 

 among coarse area designations, inter-regional varia- 

 tion and bias in sampling each life history trait, and 

 the concomitant lack of power to resolve statistically 

 significant relationships across all areas, a general and 

 consistent trend emerged among the five life history 

 traits. Intraspecifically, populations progressed from 

 larger, longer-lived, later-to-mature populations in the 

 northern-most latitudes to smaller, shorter-lived, and 

 earlier-to-mature populations in the mid and southern 

 latitudes (Fig. 2). 



Predicting missing life history information by area 



In addition to providing a comparison of the area effects 

 on the response variables, the resultant predicting model 

 offers a way to estimate missing life history values by 

 area for each species and gender: 



^ = e'/*.A+/*.+ft,+lnft„.sl_ 



(3) 



The factors /3 and /3,„ , in the above equation are not 

 present in the fecundity predictions. 



For both S. acanthias and P. glauca (Fig. 3), the pre- 

 dicted values mimicked the area trends of the reported 

 values within two standard errors in all but one case 

 (North Pacific P. glauca size at maturity was overes- 

 timated for both genders), and provided a means to 

 estimate values for each life history by species and gen- 

 der for areas not yet reported. The outlying cases may 

 indicate an area (the North Pacific) where sampling is 

 not representative of the true population (in this case, 

 of P. glauca) and is in need of further investigation. 



Cross validating models produced response variables 

 similar to those of the full models in all but two cases 

 (Fig. 4). In both cases (S. acanthias fecundity minus the 

 South Atlantic, and P. glauca size at female maturity 

 minus the North Pacific), the observed values were in 

 opposite magnitude to that predicted. This difference 

 could reflect either true relationships or possibly indi- 



cate areas that are undersampled (i.e.. North Pacific for 

 the size at maturity for female P. glauca). 



Discussion 



Knowledge of large-scale intraspecific spatial patterning 

 in life history traits may be important when considering 

 the population dynamics of a species, but such large- 

 scale patterning has seldom been formally explored. 

 Winemiller and Rose (1992) included median and range 

 latitude correlations in their consideration of several 

 life history variables of North American fishes, but 

 comparisons were only interspecific. Vila-Gispert et al. 

 (2002) demonstrated that fishes from higher latitudes 

 north of the equator matured latest and had the highest 

 fecundity, whereas fishes from South America had the 

 lowest fecundity and earliest maturation, although these 

 comparisons were again made interspecifically. Myers 

 et al. (2001) described the relationship between maxi- 

 mum reproductive rate and carrying capacity among 

 21 stocks of Atlantic cod (Gadus morhua) in the North 

 Atlantic using mixed effects models, but their analysis 

 was done for only one species in a limited region. Helser 

 and Lai (2004) also performed a similar analysis for 

 individual growth rates in North American largemouth 

 bass (Micropterus salmoicles) populations and found 

 latitudinal changes in growth rate. 



Regarding elasmobranchs, Cortes (2000) considered 

 trends in intraspecific reproductive traits for sharks but 

 did not explicitly investigate the spatial patterning of 

 those trends. Frisk et al. (2001) found regional differ- 

 ences across five areas for the spiny dogfish using three 

 life history measures (maximum size, and size and age 

 at maturity), but did not specify regional patterns. The 

 authors also performed a similar analysis with several 

 skate species, finding no difference among areas, but 

 they considered only interspecific patterns. Cortes and 

 Parsons (1996) compared the demography of two Florid- 

 ian populations of the bonnethead shark iSphyrna tibu- 

 ro}, which included several life history measures in the 

 life table analyses, but the small spatial resolution was 

 inadequate to indicate large-scale spatial life history 

 correlations within this species. Lombardi-Carlson et 

 al. (2003) extended the bonnethead shark investigation 

 to a larger portion of the eastern Gulf of Mexico and 

 found latitudinal variation in maturity and size, but 

 again the scale of this study was relatively small. Ad- 

 ditional small scale studies on intraspecific geographic 

 variation in reproductive parameters of sharks have 

 been presented by Horie and Tanaka (2002), Taniuchi 

 et al. (1993), and Yamaguchi et al. (2000). 



The results of the present study, specifically aimed at 

 sharks as an example, indicate an emerging pattern for 

 intraspecific life history variation, not unlike previously 

 recognized interspecific patterns. Generally, there is a 

 distinct difference in life history traits among areas — a 

 pattern potentially useful when considering region-spe- 

 cific population dynamics. Across taxonomic designa- 

 tions, populations in the northern latitudes tended to be 



