Sutter et al.: Movement and stock affinities of Scomberomorus cavalla 



317 



definitions were considered in combination with other 

 published king mackerel studies (e.g., Brusher and 

 Palko 1987, Trent et al. 1987) to formulate possible 

 subareas. The larger subareas (Fig. 1) reflect little or 

 no tag releases and few returns, while the smaller 

 subareas along south Florida were utilized for locations 

 of directed tagging efforts. Returns in each subarea 

 were sorted by year and month of recapture. A con- 

 tinuous time-scale, representative of the relative time 

 of freedom and independent of actual year of release 

 or return, was created for each subarea by designating 

 the first 12 months after release as year 1, the second 

 as year 2, and continuing for up to 3-5 years. There- 

 fore, fish released in May of 1975, 1976, 1977, or 1978, 

 for example, would be assigned a May, year-1 release 

 date. 



The spatial patterns of king mackerel returns were 

 categorized according to the relative direction of move- 

 ment through coastal SE U.S. waters rather than ac- 

 cording to strict compass headings, because fish were 

 released and recaptured on both the Atlantic and Gulf 

 coasts. Fish were considered to be moving 'Atlantic- 

 ward' if they were recaptured in Atlantic waters north 

 of their release location or if they had moved toward 

 the Atlantic from a release location in the Florida Keys 

 region. Individuals were classified as moving 'Gulf- 

 ward' if they were recaptured south of their release 

 area or in the Gulf of Mexico. However, fish released 

 from the Florida Keys region had to be recaptured 

 north of the Monroe/Collier county line (B in Fig. 1) 

 to be classified as moving Gulf-ward. Distance from 

 tagging location was determined for each tag return. 

 Returns were grouped into 50-km increments (i.e., 

 0-49 km, 50-99 km, etc.) when within 800km of their 

 release site and in 100-km blocks thereafter. This 

 created a continual distance gradient for fish moving 

 either Gulf- ward or Atlantic-ward from each of the four 

 tagging regions. The 100-km increments were used 

 since only 9.5% of the recaptures occurred more than 

 800 km from the release sites. 



The relative magnitude of movement for tagged fish 

 moving at least 100 km away from each release loca- 

 tion, either Gulf-ward or Atlantic-ward, was deter- 

 mined by plotting the cumulative percentage of fish 

 moving through, or recaptured in, each respective 

 distance block. A 100% value was given to the block 

 nearest the tagging area (either Atlantic- or Gulf-ward) 

 because all fish had to move at least 100 km to be in- 

 cluded in the analysis. The 100-km limit was used to 

 prevent including in the analysis fish that were ex- 

 hibiting random movement patterns associated with 

 the release locations. Since measurements of fishing 

 effort (commercial and recreational) and catchability 

 were not available for the time-frame of our study, 

 we had to assume that the number of returns was a 



reflection of relative effort. We based this assumption 

 on the migratory nature of this species (Beaumariage 

 1973, Collette and Russo 1984), knowing that fisher- 

 men target king mackerel as they become available 

 along various coastal waters. 



Stock definitions for king mackerel in the southern 

 Atlantic and Gulf of Mexico management zones were 

 evaluated using a discriminant function technique 

 based on a measure of generalized squared distance 

 (SAS 1985). The classification variable used in the 

 analysis was membership, based on when and where 

 a specific fish was released, of the fish in either the 

 Atlantic or Gulf stock. Number of days at large, 

 distance from release to recapture locations, and month 

 and location of recapture were used as quantitative 

 variables. A test of the homogeneity of within-covar- 

 iance matrices was made to determine whether the 

 within- or pooled-covariance matrix would be used in 

 the discriminant function. The percentage of posterior 

 probability of classification of each return for its 

 original stock (i.e., Atlantic or Gulf), calculated as part 

 of this discriminant function test, was used as an in- 

 dicator of the affinity of a king mackerel for its nominal 

 stock group. The two king mackerel stocks were then 

 divided into five substocks to refine the indices of stock 

 affinities. The Atlantic stock was separated into two 

 substocks: a combination of the SAB and NEFL 

 subareas that form the SE U.S. Atlantic coast sub- 

 stock; and the southern Florida summer substock, 

 which is a conglomerate of the SEFL and FK subareas 

 from April through October. The Gulf stock was divided 

 into three substocks: the southeastern Florida winter 

 substock that consists of the SEFL subarea from 

 November through March; the Florida Keys winter 

 substock, which corresponds to the FK subarea from 

 November through March; and a combination of all Gulf 

 of Mexico subareas (NEG and NWG subareas) that 

 form the combined Gulf substock. King mackerel from 

 a one-time tagging effort off the Texas coast during 

 1977 (TV 319) by FDNR were included in the stock af- 

 finity analysis as part of those fish released as com- 

 bined Gulf substock. These Texas fish were not in- 

 cluded in any of the other analyses of the 12,493 tagged 

 fish. 



Results 



Fish length did not influence movement and was not 

 a factor in the probability of recapture. Neither length- 

 at-tagging (r 0.08; df 1,147; F 0.722) nor length-at- 

 return (r 0.075; df 1,654; F 3.688) was significantly 

 correlated with the distance that a fish traveled from 

 its release area. Bias associated with tag return relative 

 to length was examined by comparing length of re- 



