FISHERY BULLETIN: VOL. 75, NO. 1 



decrease in the ratio for larvae >13.0 mm. The 

 ratio increased in other studies on clupeoid larvae 

 throughout the size range of larvae that were col- 

 lected (Ahlstrom 1954, 1959b; Lenarz 1973; Mat- 

 suura in press), and this is true for other species of 

 clupeid larvae that I have studied in the Gulf of 

 Mexico. The return of the ratio toward unity after 

 round herring larvae reached 13.0 mm must indi- 

 cate that larvae 13.0-18.0 mm became as good at 

 avoiding the gear at night as during the day. The 

 alternative explanation, which seems unlikely, is 

 that larger larvae lost the potential to avoid the 

 gear during daylight. Daylight is only one factor 

 that could allow larvae to avoid the gear and ad- 

 justment of catches to account for it can only par- 

 tially correct for avoidance losses. The correction 

 was made, however, in an attempt to get the best 

 estimate possible for round herring lar- 

 val mortality during the 1971-72 and 1972-73 

 seasons. 



Larval abundance estimates, corrected for day- 

 time avoidance, were determined by 1-mm length 

 classes for the 1971-72 and 1972-73 seasons (Fig- 

 ure 11) (Equation (10)). Except for larvae in the 

 4.1- to 5.0-mm length class, which were twice as 

 abundant in 1972-73, total abundance of larvae 

 was similar in the two seasons. The greater abun- 

 dance of 4.1- to 5.0-mm larvae in 1972-73 could 

 have reflected the reduction in towing speed from 

 the previous season. Escapement of small larvae 

 through the meshes may have been more impor- 

 tant in 1971-72 when towing speed averaged 

 about 0.7 knot faster. 



Abundance of round herring larvae decreased 

 exponentially as lengths increased during each 

 season (Figure 11). Fitted exponential functions 

 for 5.1- to 16.0-mm larvae in 1971-72 and 4.1- to 

 16.0-mm larvae in 1972-73 provided estimates of 

 the instantaneous mortality coefficients per mil- 

 limeter increase in length (Figure 11). The 

 coefficients were Z = 0.2269 in 1971-72 andZ = 

 0.3647 in 1972-73. These correspond to percentage 

 losses per millimeter increase in length of 20.3% 

 in 1971-72 and 30.5% in 1972-73. Confidence in- 

 tervals at the 0.95 probability level were Z = 

 0.2269 ± 0.0930 in 1971-72 and Z = 0.3647 ± 

 0.1179 in 1972-73. The null hypothesis of no 

 difference in mortality coefficients between years 

 was accepted at the a = 0.05 probability level 

 U-test; 0.05<P<0.10), but the t value was close to 

 the rejection region suggesting that mortality 

 may have been higher in 1972-73 than in 1971-72. 



The mortality coefficients that I obtained are 



70 



60 



50 



40 



30 



20 



s- 10 

 b 



~ 



S'20 



z 



no 



z 



D 



01 100 



< 



S 90 



< 

 I 80 



3 70 



60 



50 



40 



30 - 



20 



10 



J 



N L -(2.2799« IO l5 )e° 22G9L 

 (r 2 - 7718) 



life 



d 



■(5 8721 « IO'*)e" 



fa** 



tf>T> 



2 ,. »|. 4|. si- 6i- 7|- Bh 91- HI- 13 1- 151- 17 1- 19 1- 21. t- 23 1- 23 I- 271- 29 1- 

 3 40 30 60 70 8090(0 120 14 160 18 20 22.0 24 26 28 30 



LENGTH CLASS (mm) 



FIGURE ll. — Length-frequency distributions of annual larval 

 abundance estimates of round herring larvae collected in the 

 eastern Gulf of Mexico. Frequencies in each 1-mm length class 

 are expressed as estimated annual abundance and have been 

 corrected for daytime avoidance. Fitted exponential functions 

 provide estimates of the instantaneous coefficient of decline in 

 abundance by length, 1971-72 and 1972-73. 



similar to those reported by Lenarz (1973) from 

 several years of data on Pacific sardine and north- 

 ern anchovy, Engraulis mordax. He reported a 

 range of instantaneous coefficients of 0.15-0.33, 

 averaging 0.22 for Pacific sardine, that correspond 

 to a 20% loss per millimeter of growth. For an- 

 chovy his instantaneous coefficients ranged from 

 0.32 to 0.46, averaging 0.39, a mean decrease of 

 32% per millimeter of growth. Matsuura (in press) 

 has measured the rate of decline in catches of 

 Brazilian sardine, Sardinella brasiliensis, obtain- 

 ing an instantaneous coefficient of 0.4962, corre- 

 sponding to a 39% decrease in catch per millimeter 

 of growth. Most of the decline in catch of larger 

 round herring larvae presumably was due to lar- 

 val mortality but gear avoidance also must be 

 important. For this reason mortality curves were 

 fitted only for larvae 16.0 mm or less in length. 

 Catches of larger larvae were sporadic and possi- 

 bly greatly influenced by gear avoidance. 



Larval mortality is best expressed as a function 

 of age. If it is assumed that growth of round her- 

 ring larvae is exponential from the post yolk-sac 



82 



