FISHERY BULLETIN; VOL. 72, NO. 2 



89°00'W 



88°30'W 



30°30'N 



30°20'N 



30°10'N 



X 



X 



a. < 

 < 



CD 



< 



BAY ST. LOUIS 



... . BILOXI •■ OCEAN 



;"< 



SPRINGS. .PASCAGOULaJ V 





HIGH POTENTIAL AREA 



30°00'N -^ **»«,>'<« flX) ^ 



[ I MODERATE POTENTIAL AREA 



LOW POTENTIAL AREA 



i 



Figure 5. — Model D8 predictions for menhaden distribution in the Mississippi Sound on 7 August 1972, between 



0900-1500 h (CDT) (based on 95 sea-truth measurements). 



size was divided into three categories: 0-50, 

 50-100, and more than 100 thousand fish, and an 

 analysis of variance applied to the categories to 

 test for differences between mean parameter con- 

 ditions. No significant differences were found be- 

 tween catch size and salinity, Forel-Ule color, and 

 depth parameters at significance levels down to 

 50%. However, a significant difference at 95% was 

 found between the first and third catch size cate- 

 gory for averaged secchi disc transparency values 



(T 



0-50K 



= 1.09 m and T 



>100K 



= 1.32 m). This sig- 



nificance probably does not have biological mean- 

 ing, however. It probably reflects changes in the 

 ability of fishermen to selectively detect and 

 capture fish schools with respect to water clarity. 



SUMMARY AND CONCLUSIONS 



The feasibility of using satellite-supported en- 

 vironmental sensors to predict fish distribution 

 was demonstrated. ERTS-1, MSS Band 5 imagery 

 was shown to contain density levels which corre- 

 lated with menhaden distribution. These density 

 levels were further shown to correlate signifi- 

 cantly with sea-truth measurements of secchi 

 disc transparency and water depth, two pa- 

 rameters which also correlated significantly 

 with menhaden distribution. Additionally, sur- 



face salinity, Forel-Ule color, and chlorophyll a 

 were found to correlate significantly with menha- 

 den distribution. Independent tests of four 

 oceanographic parameter-menhaden distribu- 

 tion relationships with oceanographic informa- 

 tion taken at or near sites of commercial menha- 

 den captures corroborated these relationships. 

 The correlation between chlorophyll a and 

 menhaden distribution could not be substantiated 

 because of insufficient data. 



Eight empirical regression models which pre- 

 dict menhaden distribution in the study area were 

 constructed from combinations of four oceano- 

 graphic parameters: water depth, secchi disc trans- 

 parency, surface salinity, and Forel-Ule color. Al- 

 though the models did not provide particularly 

 precise predictions about menhaden distributions, 

 their predictions nevertheless were statistically 

 significant. The importance of the models is that 

 they demonstrate a potential means or direction 

 through which remotely sensed oceanographic in- 

 formation can be used to provide menhaden dis- 

 tribution information on a real-time basis. This 

 information could be used by the commercial in- 

 dustry to reduce spotter-pilot search time by iden- 

 tifying likely areas for concentrations of menha- 

 den and by resource managers as an aid in plan- 

 ning assessment surveys. 



388 



