558 



Fishery Bulletin 97(3), 1999 



mean baywide growth-in-length and growth-in- 

 weight rates for each cruise were estimated from 

 regressions fitted to pooled length and age data from 

 all stations. Because larvae collected in June were 

 <13 mm SL, growth rates of larvae collected in July 

 were estimated from two separate regressions — one 

 for larvae <13 mm SL that was directly comparable 

 to the June data and one for larvae >13 mm SL. 



Mortality rates 



Age-length keys were developed to convert larval 

 length distributions to age distributions from which 

 mortality rates then were estimated. To derive the 

 keys, linear regressions, based upon subsamples of 

 otolith-aged larvae from each cruise and region, were 

 fitted to larval age-on-length relationships. For each 

 regression, the standard error of the estimated re- 

 gression coefficient was used to calculate a standard 

 normal deviate (z-statistic) from which probabilities 

 of ages of larvae within LO-mm length classes were 

 obtained. Six age-length keys were constructed, one 

 for each region and cruise. This maintained the re- 

 gion-specific integrity of size-at-age data, allowing 

 estimation of region-specific mortality rates. 



Instantaneous daily mortality rates of larvae were 

 estimated from an exponential model of decline in 

 abundance with respect to age: 



N, 





where A'^, 



N. 



abundance ( number/m''^ 1 at age Md); 



estimated initial abundance (y-inter- 



cept of regression; number/m-); 

 M = instantaneous mortality coefficient (/d); 



and 

 t = age (d). 



The data were fitted to the log-linear form of the 

 model after log_,-transformation of the abundance data. 



Region-specific mortality coefficients were esti- 

 mated and compared within each cruise by analysis 

 of covariance (ANCOVA). Mean baywide mortality 

 rates for each cruise (month) were estimated by pool- 

 ing abundance-at-age data from all stations and then 

 compared in ANCOVA. When significant, ANCOVAs 

 were followed by a multiple comparison test (Stu- 

 dent-Newman-Keuls) to determine which mortality 

 estimates differed significantly Two separate mor- 

 tality rates were estimated for larvae collected in 

 July — one for larvae <13 mm SL (<18-day-old lar- 

 vae) that was directly comparable to the June data 

 that included only larvae of those ages, and one for 

 larvae >13 mm SL (>18-day-old larvae). Length-spe- 

 cific mortality rates also were estimated, by regress- 



ing logp-transformed abundances of larvae on 1-mm 

 length classes. 



M/G ratio and stage-specific survivorship 



Stage-specific survival can be estimated from the M / 

 G ratio, where M is the instantaneous mortality rate, 

 and G is the weight-specific growth coefficient. The 

 M/G ratios were compared between cruises and among 

 the three designated regions of Chesapeake Bay. The 

 ratio M/G is an indicator of stage-specific survivorship 

 and production potential of larval cohorts ( Houde, 1996, 

 1997a, 1997b). The ratio, sometimes termed the "physi- 

 ological mortality rate," expresses a population's mor- 

 tality per unit of individual growth (Beyer, 1989). 



Stage-specific survival of bay anchovy larval co- 

 horts was estimated as 



[W /wj-'-w/G.^ 



where S = stage-specific survival = ^J^o', 



N^ = number of survivors at the end of a 

 stage; 



Nn 



number alive at the beginning of a 



stage; 



W^ = dry weight of a 12-mm-SL bay anchovy 

 larva (1000 mg); 



Wg = dry weight of a 3-mm-SL bay anchovy 

 larva ( 10 g). [Note: This weight is more 

 accurate than weight estimated for a 

 3-mm larva from the weight-length re- 

 lationship, which overestimated weights 

 of the smallest larvae]; and 

 M/G = ratio of instantaneous mortality coef- 

 ficient (M) and weight-specific growth 

 coefficient (G). 



Survival to 12 mm SL, the largest length class fully 

 represented in collections in each of the months, was 

 calculated for each region by multiplying stage-spe- 

 cific survival rate (S) by estimated abundance of the 

 smallest fully represented length class (i.e. A^,, at 3 

 mm SL). Stage-specific survival rates also were esti- 

 mated for egg to 3-day larva, 3-day to 10-day larva, 

 and 10-day to 18-day larva. Age-specific production 

 at a station was obtained by multiplying estimated 

 larval density (number/m*) by the volume repre- 

 sented by the station. Regional productions were 

 obtained by summation of larval abundances for all 

 stations in each region. 



Correlations and predictions 



Multiple regression analyses were applied to deter- 

 mine if bay anchovy larval growth and mortality 



