FISHERY BULLETIN: VOL 76. NO. 1 



Station 2(N ^ 22) and November and July collec- 

 tions from Station 1 [N = 122). A total of 469 

 stomachs (72%) from all stations contained food 

 items. April and May collections at Station 3 were 

 made during the day; all other collections were 

 from 1 to 5 h after sunset which was near the end of 

 the greatest diel feeding activity. Ross (1977) 

 demonstrated that searobins from the West 

 Florida Shelf, including P. scitulus, had their 

 greatest feeding activity during the day, but re- 

 tained full stomachs through midnight. 



Collection depths averaged 5, 5, and 7 m, respec- 

 tively, for Stations 1-3. Sampling gear was a 3.6-m 

 otter trawl with 2.5-cm stretched mesh and a 

 0.5-cm cod end liner. Upon capture I injected all 

 specimens intraperitoneally with \Q'7( Formalin. ^ 

 Fish were fixed for 2 wk in 10'7( Formalin and then 

 washed and transferred to 40% isopropanol for 

 storage. 



I sorted prey by taxa from each 10-mm size class 

 offish and measured a random sample (/?ss25) of 

 each prey kind to the nearest 0.1 mm along the 

 axis of greatest dimension. The level of prey iden- 

 tification used in comparisons of size groups was 

 the lowest taxon which was regularly identifiable 

 for each prey kind. Since polychaetes were gener- 

 ally fragmented, they were not measured. Mean 

 number of prey per fish was based only on fish 

 which contained food items. 



I used a volume displacement technique to mea- 

 sure food items >0.05 cm^ and a squash technique, 

 modified from Hellawell and Abel (1971), to mea- 

 sure volume of food items <0.05 cm^ (Ross 1974). 

 To establish minimum sample sizes for description 

 of the ration I used the criterion t, obtained by 

 plotting cumulative trophic diversity {H ), ) against 

 cumulative stomachs examined ik). Actual num- 

 bers of stomachs ik) varied between samples but 

 had a lower limit of 17. The value of ^ was greater 

 when specimens varied more in date or location of 

 capture. Trophic diversity was determined by the 

 Brillouin information function {H) according to 

 Pielou (1966) and Hurtubia (1973). A horizontal 

 asymptote, beginning at t, indicated a sufficient 

 sample size so that examination of stomachs in 

 excess of t would not yield an increase in trophic 

 diversity. 



To compare trophic differences of size groups of 

 P. scitulus I used an unweighted pair group, 

 arithmetic average (UPGMA) cluster analysis 



^Reference to trade names does not imply endorsement by the 

 National Marine Fisheries Service, NOAA. 



226 



(Sneath and Sokal 1973) based on a Czechanowski 

 similarity matrix (Bray and Curtis 1957). All 

 linear regressions were based on the Berkson case 

 of a model I regression (Sokal and Rohlf 1969). 



All fish lengths reported are standard length 

 (SL), measured to the nearest 0.1 mm. Mouth 

 width was measured externally between the pos- 

 terior maxillary processes with the mouth fully 

 closed. Internal mouth width was not routinely 

 measured because of difficulty in working with 

 preserved specimens. However, there was no dif- 

 ference between external mouth width with the 

 mouth fully closed and internal mouth width with 

 the mouth fully opened on 36 specimens favorable 

 to such a comparison (two-tailed paired t = 1.88; 

 P>0.05). Measurement of mouth length followed 

 Hubbs and Lagler (1958). 



To measure intestinal length I cut the hindgut 

 distally at the anus and freed the intestine from 

 the investing mesentery. Length (to the nearest 

 millimeter) was measured from the stomach with 

 the intestine fully extended, but not stretched. 



Wet weights of P. scitulus were taken to the 

 nearest 0. 1 g after removing stomach contents and 

 blotting the specimens with absorbent paper. 

 Ovaries and testes were removed, blotted, and 

 weighed to the nearest 0.001 g. To compare levels 

 of gonadal activity I used a gonadosomatic index 

 (GSI = (gonad weight/somatic weight) x 100). 



RESULTS 



Food Habits 



The dominant prey of P. scitulus based on per- 

 cent occurrence and percent volume was the lance- 

 let, Branchiostoma floridae, which composed 61% 

 of the food volume and occurred in 60% of the fish 

 examined (Table 1). Numerically, cumaceans 

 were dominant, making up 40% of the total 

 number of prey. On the basis of percent number, 

 volume, and occurrence, the ration of P. scitulus 

 was composed primarily of lancelets, polychaetes, 

 natantians, brachyurans, gammarid amphipods, 

 cumaceans, pelecypods, copepods, and larval crus- 

 taceans. Ninety percent of the number of prey 

 items and volume of prey items were accounted for 

 by 6 and 7, respectively, of the 22 major food 

 categories. 



I examined seasonal feeding patterns of P. sci- 

 tulus from Station 3, using fish 100 mm or larger to 

 eliminate effects of fish size. Branchiostoma 

 floridae occurred in over 50% of the fish in 8 of the 



