FISHERY BULLETIN: VOL. 86, NO. 2 



Most identifiable prey items fit into the nine 

 categories: dinoflagellate, tintinnid, rotifer, poly- 

 chaete larva, lamellibranch larva, crustacean 

 nauplius, copepodite and adult copepod, am- 

 phipod, and invertebrate egg (Fig. 3). The "other" 

 category applied only to size-2 larvae at 1 m (1 

 Globigerina sp.) and to size-3 larvae (1 zoea, 3 

 larvaceans, and two large [1 mm] unidentified 

 spheres). 



Guts of preflexion (sizes 1 and 2) larvae from 

 the three sampling heights contained an array of 

 small (<300 jxm) organisms that varied mainly in 

 proportions from mostly rotifers (88%) in size-1 

 larvae at 6.7 m to a diverse mix of prey numeri- 

 cally dominated by nauplii in size-2 larvae at 0.5 

 m (Fig. 3). Percent similarity (overlap) among the 

 5 groups of preflexion larvae ranged from 24 to 

 75%. The gut of the single size-2 larva at 6.7 m, 

 not included in Figure 3, contained two tintin- 

 nids. 



Size-3 larvae had a diet consisting chiefly of 

 copepodite and adult copepods that overlapped 

 only 8-9% with size-2 larvae and 1% or less with 

 the three groups of size-1 larvae. The copepods 

 eaten by size-3 larvae were mostly Corycaeus an- 

 glicus (62% of all copepods), unidentified cope- 

 podites (cyclopoid and calanoid, 25%), and Para- 

 calanus parvus (9%). Polychaete larvae were 

 identified only from the presence of setae in the 

 guts, so the proportion (nominally 16% of all prey 

 items) of this taxon in the diet is more an indica- 

 tion of incidence than of numerical importance. 

 Amphipods, mostly in the length range 1-1.5 

 mm, were found in white croaker larvae ranging 

 from 6.5 mm FL to 10.3 mm SL. The gut of the 

 flexion-stage larva at 1 m, not included in Figure 

 3, contained three C. anglicus and traces of poly- 

 chaete setae. 



While there can be no doubt that flexion and 

 postflexion larvae had a different diet than pre- 

 flexion larvae, the pattern of decreasing propor- 

 tion of rotifers with increasing size and proximity 

 to the bottom among preflexion larvae was of 

 questionable statistical significance. The first 

 question asked was whether the very high per- 

 centage of rotifers in the diet of size-1 larvae at 

 6.7 m was likely to have arisen by chance from a 

 random sampling of size-1 larvae. Formally 

 stated, Hq = "all size-1 larvae had the same per- 

 centage of rotifers". The 123 nonempty guts were 

 pooled, and random samples of 84 each were 

 drawn. In 1,000 iterations, <4% of the samples 

 had >88% rotifers, so it was concluded that lar- 

 vae at 6.7 m ate significantly more rotifers than 



similar-sized larvae near the bottom. The remain- 

 ing 75 preflexion larvae (sizes 1 and 2) are divided 

 into 4 small groups at 0.5 and 1 m, so we next 

 tested for a size effect by pooling across sampling 

 height, such that the guts of the 39 near-bottom 

 size-1 larvae contained 64% rotifers, and the 36 

 size-2 larvae had 36% rotifers. Bootstrapping as 

 before, <2% of samples of 36 had <36% rotifers, 

 so it was concluded that size-1 and size-2 larvae 

 differed in this regard. Further testing (e.g., of a 

 height effect within sizes) was not done because of 

 small sample sizes and multiple testing consider- 

 ations. 



Abundance and 

 Vertical Distribution of Prey 



Rotifers, all identified as the brachionoid Tri- 

 chocerca sp., figured importantly both in the diet 

 of preflexion larvae and in the time-related vari- 

 ance structure of the microplankton. As shown in 

 Table 3, there was a change in the size spectrum 

 of these animals that coincided approximately 

 with the time of changing from near-bottom sam- 

 pling to midwater sampling with the fish pump. It 

 was only the largest category of rotifer (200-300 

 |xm, including the "toe") that was found in the 

 guts of the larvae. The relative abundance of total 

 rotifers in the plankton at the times and heights 

 of pump sampling differed very little (25-33% of 

 all organisms in the 100-300 ^JLm size class), but 

 the percentage of rotifers in the 200-300 [xm class 

 increased from 21% (near-bottom, morning) to 

 86% of all rotifers (midwater, afternoon). The 

 dominance of rotifers in the diet of size-1 larvae in 

 midwaters is thus likely related to the larger size 

 of rotifer resident in the water column when that 

 height was sampled. 



The most notable dietary difference among the 

 larval size groups analyzed was the switch from 

 small (50-300 (xm) to larger (0.5-2.5 mm) prey, 

 principally the copepod Corycaeus anglicus (0.5- 

 0.8 mm), upon flexion of the notochord. The abun- 

 dance of Corycaeus from the 100 ixm mesh pump 

 samples (Table 4) shows that this prey item was 

 equally or more abundant in midwater than near 

 the bottom, where all the flexion and postflexion 

 larvae were captured. (Within the bottom meter, 

 the similar-sized but more transparent Para- 

 calanus parvus outnumbered C anglicus by a fac- 

 tor of 5-20.) The only prey found in numbers in 

 these larvae that was restricted to the 0.5 m sam- 

 ples was gammarid amphipods. Larger crus- 

 taceans — cumaceans, crab and shrimp zoea, 



258 



