260 



Fishery Bulletin 94(2). 1996 



cause mortality estimates presumably were not af- 

 fected by consistent proportional losses. Adjustments 

 to estimate actual yolk-sac larval abundances could 

 be obtained by multiplying the uncorrected abun- 

 dances by 2.3, the adjustment factor derived from 

 paired collections in 280-/jm and 53-fum mesh nets 

 (MacGregor, 1994). 



Mortality 



Because bay anchovy eggs are abundant in Chesa- 

 peake Bay and hatch <24 h after being spawned, an 

 intensive ichthyoplankton survey could derive mor- 

 tality estimates for a daily cohort of eggs or yolk-sac 

 larvae on the basis of the decline in their abundances 

 on a single day. By using this approach, daily mor- 

 tality rates were estimated. Hourly instantaneous 

 mortality rates of eggs and yolk-sac larvae were es- 

 timated from a log-linear fit of the exponential model: 

 N - N Q e Zt , where N t is the abundance (number 

 under 1/m 2 ) of a cohort at time t, N n is its initial abun- 

 dance (number under 1 m 2 ), Z is the instantaneous 

 mortality coefficient ( per h ), and / is the time elapsed 

 (h) since the initial abundance estimate. 



Estimates of daily cohort abundances and mortal- 

 ity rates of eggs and yolk-sac larvae were compared 

 among days, an approach seldom possible in research 

 on species with eggs that have longer development 

 times. Some losses due to diffusion or transport of 

 eggs and larvae out of the sampling areas may have 

 occurred. However, such losses were minimized by 

 the short period over which each cohort was sampled 

 (<1 d) and because sampling was directed by the drift 

 of mesocosms that tracked the initially sampled wa- 

 ter mass. 



Egg production and adult biomass 



We obtained an estimate of adult biomass from esti- 

 mates of egg production, knowledge of adult fecun- 

 dity, and spawning frequency. Egg production, when 

 divided by relative fecundity (mean number of eggs 

 spawned per gram of female), provides an estimate 

 of adult female biomass (Saville, 1964). Biomass of 

 serial-spawning fishes such as bay anchovy can be 

 estimated by single-survey methods of egg produc- 

 tion (Lasker, 1985a; Parker, 1985; Eltink 1 ). Parker's 

 ( 1985) formulation to estimate adult biomass is 



B = EJ(F. xSxR) 



'Eltink, A. 1991. Horse mackerel egg production and spawn- 

 ing stock size in the North Sea in 1990. Int. Counc Explor. 

 Sea, Council Meeting 1991/H:27, 14 p. 



where B = adult biomass (g); 



E , = daily egg production; 



F bw = mean batch fecundity per female; 



S = spawning fraction (i.e. the fraction of 



females that spawned on the sampling 



date); and 

 R - the proportion of the population that 



is female. 



Batch fecundities and spawning fraction of bay an- 

 chovy were based upon estimates by Zastrow et al. 

 ( 1991) who reported that batch fecundity (F bw ) was 

 687.1 ova per gram of ovary-free female weight and 

 that spawning fraction (S) was 1.0 during July in 

 Chesapeake Bay. We assumed that the proportion of 

 adult females (i?) was 0.5, on the basis of trawling 

 data (Newberger and Houde, 1995), which indicated 

 a mean female:male ratio of 1.02:1.00. Thus, rela- 

 tive adult biomass (g/m 2 ) at each site was about twice 

 the relative female biomass. 



Predators and other factors 



Egg and yolk-sac larval abundances were examined 

 in relation to environmental variables. Possible cor- 

 relations between egg or yolk-sac larval abundances 

 and biovolumes (mL/m 2 ) of two gelatinous predators 

 (ctenophore M. leidyi and scyphomedusa C. quin- 

 quecirrha ) were analyzed because Cowan and Houde 

 ( 1992, 1993) and Purcell et al. ( 1994) indicated that 

 these dominant species were important predators on 

 bay anchovy eggs and larvae. Bay anchovy appar- 

 ently is not cannibalistic (Klebasko, 1991), and other 

 major predators have not been identified. Possible cor- 

 relations between egg or yolk-sac larval abundances 

 and zooplankton biomasses also were examined. 



Statistical analyses 



When assumptions for parametric tests were met, 

 Analysis of Variance (ANOVA) was used to test for 

 differences among daily surveys in initial egg abun- 

 dances, zooplankton biomasses, and gelatinous 

 predator biovolumes. An ANOVA also was used to 

 determine if means of those variables differed among 

 the five surveys carried out at site 5 (Fig. 1 ). When 

 ANOVA results were significant, individual means were 

 compared by using the a posteriori Student-Newman- 

 Keuls (SNK) multiple-comparison procedure. 



The estimates of hourly instantaneous mortality 

 rates of eggs and yolk-sac larvae were the regres- 

 sion coefficients in the log-linear models relating log ( , 

 abundance to elapsed time. Egg mortality was cal- 

 culated from collections made between 0200 and 2000 

 h because eggs hatch in -20 h. Yolk-sac larval mor- 



