422 



Fishery Bulletin 98(2) 



Materials and methods 



Atlantic menhaden larvae were collected while 

 ingressing into the Newport River estuary during 

 the week of 7 December 1994. In the laboratory, 

 these larvae were immersed on 8 December, in 100 

 mg/L solution of alizarin-complexone (ALC) solution 

 for 16 h (overnight) at 18°C. Nearly equal numbers 

 of marked larvae were reared in two concrete ponds 

 (6600 L). Pond one, the high-salinity pond, received 

 flow-through seawater at ambient water tempera- 

 tures. Pond two, the low-salinity pond, received flow- 

 through seawater and freshwater from a well. Pond 

 two was generally a few degrees warmer and about 

 one-half the salinity of pond one. Menhaden larvae 

 were fed powdered salmon starter, supplemented 

 with brine shrimp larvae, for the first few weeks, fol- 

 lowed by salmon starter only. 



Two to six specimens were sacrificed periodically, 

 and measured for fork length (FL) to the nearest 

 mm. Sagittal otoliths were removed, mounted in 

 droplets of epoxy resin, and sectioned obliquely with 

 a slow-speed saw with dual diamond wafering blades 

 (Ahrenholz et al., 1995). Polished otolith sections 

 (10-15 \im thick) were viewed through a compound 

 microscope (lOOOx) and on an image analysis video 

 monitor (3800x). When the otolith sample was exam- 

 ined under blue light epifluoresence, the ALC-marked 

 increment was seen as red-orange. The number of 

 increments from the ALC mark to the otolith margin 

 was estimated as the arithmetic mean of a series 

 of independent counts tallied blindly with a hand- 

 held counter. A section was counted twice when the 

 two counts differed by to 2 increments. A section 

 was counted three to five times when the difference 

 between the initial two counts was greater than two 

 increments. Otoliths from one specimen from the 

 high-salinity pond were deemed unreadable. 



An image analysis system was used to measure 

 increment widths in groups of 10, starting from the 

 innermost increment, for a select number of oto- 

 liths from each test group. However, the increments 

 measured just prior to the alizarin mark may have 

 numbered from 5 to 10. This procedure was under- 

 taken to separate pre- and postmark increments 

 from width estimates. Measurements were taken 

 from oblique-transverse sections (Ahrenholz et al., 

 1995) as opposed to similar measurements taken 

 on transverse sections, as done by Fitzhugh et al. 

 (1997). Owing to nearly concentric otolith growth for 

 the life history stage sampled, the measurements 

 taken in our study should be similar to, or slightly 

 greater than, transverse-oriented measurements. 



Estimates of the dates when eggs were spawned 

 for the test groups were obtained by sacrificing eight 



larvae on the day of ALC-marking. Their otoliths 

 were removed, mounted on microscope slides, and 

 read whole. Three independent counts were made 

 from the first increment following the first feeding 

 mark to the margin. Five days were added to the 

 mean counts to account for time from spawning to 

 first increment formation (Warlen, 1992). 



Statistical analyses of the results generally fol- 

 lowed those of Ahrenholz et al. (1995). The null 

 hypothesis tested was that the growth increments 

 in the otoliths were formed daily during the rearing 

 period. The hypothesis was not rejected if the slope 

 of a regression of increment count on number of days 

 since marking was not significantly different from 

 one and the intercept was not significantly different 

 from zero. The statistical power to test for differ- 

 ences in the slope was also calculated (Rice, 1987). 

 Student's-/ test was used to test for statistical signifi- 

 cance of the regi'ession parameters. Analysis of cova- 

 riance (ANCOVA) and regression computations were 

 performed with SAS statistical programs (SAS Insti- 

 tute, 1985). Ninety-five percent confidence intervals 

 (CI) for individual age estimations (see discussion in 

 Rice 1 1987 1 ) were calculated for an inverse prediction 

 condition (Sokal and Rohlf, 1981; note also discussions 

 of inverse regression in Draper and Smith 1 1980] ). 



Results 



Fish were sampled from pond two (the low-salinity 

 regime) on postmarking days 35, 56, 74 (and 75), 

 98. 119, and 140 (total /;=22). Fish from pond one 

 (the high-salinity regime) were sampled similarly 

 through day 75 (total « = 13). Prior to day 75 (for 

 about two weeks), many of the menhaden in pond 

 one died. The remaining fish from the high-salinity 

 regime were sampled on day 75. 



Spawned-date estimates determined from larvae 

 sacrificed before being reared ranged from 17 Octo- 

 ber to 20 October 1994, thus, verifying that the 

 experimental tests were performed on larvae derived 

 from fall-spawning activity. 



Temperature and salinity regimes during the rear- 

 ing period ( Fig. 1 ) were adequate to attain the study's 

 objectives. The cooling and warming pattern, as well 

 as the actual water temperatures were well within 

 the range observed for winters in North Carolina 

 estuaries over a 15-year period (Hettler and Chester, 

 1982). Salinity levels in both rearing regimes were 

 within ranges observed for larval or juvenile menha- 

 den in the field (Wilkenson and Lewis, 1971; Turner 

 and Johnson, 1973), although the larval forms are 

 more highly associated with lower salinities. Water 

 temperatures were relatively warm in the early por- 



