622 



Fishery Bulletin 93(4), 1995 



provide a consistent basis for determining whether 

 an annulus should be counted as an annual mark. 

 For the 4-5 annual marks closest to the core, a notch 

 was usually present on the ventral edge of the sulcal 

 ridge (Fig. IB). The notch was typically accompanied 

 by an annulus extending outward to the otolith's 

 ventral margin. If a distinct notch was present, the 

 annulus was counted even if the mark extending 

 outward was indistinct. Confluent annuli were 

 counted as a single annual mark unless the two an- 

 nuli were confluent for only a short distance (Fig. 

 1C). If many confluent marks were present, the 

 otolith was rejected as unreadable. 



Annulus counts for individual otoliths often showed 

 some level of variation among readings. We estab- 

 lished criteria for accepting or rejecting individual 

 otoliths by calculating a coefficient of variation 

 (CV) = (S/yx 100%), where S = the standard error 

 of counts for a given otolith, and y~ = the mean annu- 

 lus count for a given otoli th. CV pr ecision criteria 

 were calculated as CV = ^J(nxd 2 )/t 2 , where n=the 

 number of readings for a given otolith (6), c?=the de- 

 viation allowed between the estimated mean incre- 

 ment count and the true count for a given otolith, 

 and t-a one-tailed Student's ^-statistic with a =0.05 

 and re-1 degrees of freedom. We allowed a maximum 

 deviation (d) of 10%, which corresponds to a CV of 

 12.16%. After six readings were completed, otoliths 

 for which there were significant disagreements 

 among readings (CV>12.16%) were again examined 

 by both readers in an attempt to reconcile differences. 

 After discussing possible explanations for the vari- 

 ability among readings, a decision was made regard- 

 ing the readability of the otolith. If both readers 

 judged the otolith to be readable, it was again read 

 independently by each reader without knowledge of 

 the previous readings. The reading showing the larg- 

 est difference from the mean of all readings was then 

 discarded and replaced by a new reading. This pro- 

 tocol was repeated twice, and if the CV remained 

 >12.16%, the otolith was rejected. 



Tarpon typically spawn during May-August 

 (Crabtree et al., 1992; Crabtree, 1995), and annulus 

 formation took place during January-May. Conse- 

 quently, annulus counts were not always equivalent 

 to age in years. To resolve this discrepancy, fish col- 

 lected before 1 July (the approximate middle of the 

 spawning season) that had recently formed an an- 

 nulus during the winter or spring (determined on 

 the basis of the proximity of the annulus to the 

 otolith's margin) were assigned an age one less than 

 the annulus count. Fish collected after 1 July were 

 assigned an age equal to the annulus count. 



The von Bertalanffy (1957) growth equation 

 FL t =LJl-e { ~ Kl '- to>) ) was fit to observed age-length 



data with the nonlinear regression procedure of 

 Statgraphics. Likelihood-ratio tests were used to 

 compare parameter estimates (Kimura, 1980; 

 Cerrato, 1990). Length-weight regressions were cal- 

 culated by linear regression of log 10 -transformed data 

 and were compared with a £-test (Zar, 1984). 



Tarpon were captured from the Sebastian River, 

 located on Florida's Atlantic coast, by electroshocking 

 or with trammel nets for age-validation experiments 

 (Table 1). After capture, fish were sedated with MS- 

 222, measured for fork length, and tagged with dart- 

 type tags. After tagging, tarpon were injected with 

 Liquamycin LA-200 (200-mg oxytetracycline [OTC]/ 

 mL) in the dorsal musculature at a dosage of 100- 

 mg OTC per kg fish weight. Fish weight was esti- 

 mated with a length-weight equation. Tarpon were 

 then transported to one of three holding facilities 

 located in Florida, where they were held for 13 to 50 

 months (Table 1). Two fish were held in a 25-m by 

 13-m by 2.7-m deep public aquarium at Mote Ma- 

 rine Laboratory in Sarasota, six were held in a 33.5- 

 m by 5.5-m by 0.75-m deep pond at the Keys Marine 

 Laboratory in Long Key, and 10 were held in a 9.1-m 

 diameter by 2.0-m deep tank at the Florida Marine 

 Research Institute's Stock Enhancement Research 

 Facility (SERF) at Port Manatee. Fish were held at 

 ambient temperatures in all facilities except in SERF, 

 where heaters were used during the winter to pre- 

 vent temperatures from dropping below 14°C. Tar- 

 pon were fed as much frozen fish as they would con- 

 sume at least three times a week. Otolith sections 

 were examined with a compound microscope (40- 

 lOOx) equipped with ultraviolet light so that the fluo- 

 rescent OTC marks could be detected. 



Results 



The 1,469 tarpon we examined ranged from 102 to 

 2,045 mm in length; 740 (50.4%) of these were YOY 

 or 1-year-old fish (<400 mm). Of these 740 small fish, 

 we examined 179 histologically but could sex only 11 

 (6.1%); consequently, the sex of most YOY and 1-year- 

 old tarpon was unknown and they were excluded 

 from sex-specific regressions. Neither slopes U-test, 

 df=602, £=0.039, P=0.484) nor elevations (t-test, 

 df=603, £=0.205, P=0.419) of the length-weight equa- 

 tions for male and female tarpon were significantly 

 different. The pooled length-weight equation for 

 sexed and unsexed fish and the relationships between 

 SL, FL, and TL are presented in Table 2. 



Female tarpon attained larger sizes than did 

 males. Among the fish that we sexed, females ranged 

 from 331 to 2,045 mm in length (median=l,635 mm, 

 upper quartile= 1,752 mm, n=412) and were signifi- 



