Stanley and Wilson: Fish abundance near oil and gas platforms 



Two multivariate analysis techniques were utilized 

 to determine the relationships between species abun- 

 dance and geological, physical, temporal, and meteor- 

 ological variables. Principal Component Analysis (PCA) 

 was used on the individual fish species as a data- 

 reduction technique. The PCA transforms the original 

 set of variables into a smaller set of orthogonal linear 

 combinations of species that account for a major por- 

 tion of the variance in the original set (Chatfield and 

 Collins 1980, Dillon and Goldstein 1984). The CPUE 

 data from 17 species or species groups were reduced 

 to 5 principal components (PC's) using the FACTOR 

 procedure (SAS 1985). Only PC loadings greater than 

 0.35 were considered; although the value of 0.35 is 

 arbitrary, it implies at least 12% of the variance of the 

 species variable was accounted for by the PC. The com- 

 ponent scores of the five PC's were used in subsequent 

 multiple-regression analyses. 



Stepwise multiple-regression analyses (MRA) were 

 performed with spotted seatrout ln(CPUE + 1) and the 

 component scores of each PC on the angler character- 

 istics, meteorological, temporal, geological, and phys- 

 ical platform data and their interactions (predictor 

 variables) (Table 1). An MRA of the predictor variables 

 and ln(CPUE + 1) of spotted seatrout was treated as 

 a separate analysis because spotted seatrout repre- 

 sented 24.8% and 28.3% of the total number of fish 

 caught by anglers and charterboat operators, respec- 

 tively (Table 2), and because they did not load positively 

 with the other species in the PCA. The MRA was ex- 

 ecuted using the STEPWISE procedure with the 

 MAXR option in SAS (1985). Unless otherwise stated, 

 all differences discussed are significant at the a = 0.01 

 level of significance. 



Results 



A total of 55 anglers and 10 charterboat operators 

 returned logbooks with usable information, a 45.8% 

 and 43.5% return rate, respectively. The participants 

 fished at 467 different oil and gas platforms a total of 

 1196 separate times. Anglers fished at platforms on 

 666 occasions and caught a total of 20,559 fish repre- 

 senting over 46 different species (Table 2). Charterboat 

 operators fished at platforms 530 times and caught a 

 total of 16,280 fish representing over 42 different 

 species (Table 2). 



A five-factor PCA explained 45.7% of the variance 

 of the original data set and allowed us to reduce the 

 data from the 17 separate species or species groups into 

 a smaller data set of presumably related species (Table 

 3). The first factor was defined as a reef fish factor 

 which included high positive loadings for greater 

 amberjack, grey triggerfish, grouper, other snapper 



Table 1 



Temporal, meteorological, angler characteristic, physical plat- 

 form, and geological variables and their interactions used in 

 the multiple regression analysis. 



Angler 



Fishing method 

 Boat length 

 Engine horsepower 

 Presence of echosounder 

 Presence of LORAN 

 Presence of graph recorder 



Geological 



Mean sediment size 



Interactions 



Boat length x Hp 

 Quadratic structure age 

 Structure age x number of legs 

 Structure age x number of crossmembers 

 Structure age x submerged surface area 

 Number of legs x number of cross members 

 Number of legs x number of wells 

 Number of legs x enclosed volume 

 Number of legs x submerged surface area 

 Structure manned x structure in production 

 Water depth x volume of water enclosed 

 Water depth x submerged surface area 

 Submerged surface area x volume of water 



Physical 



Structure age 

 Number of crossmembers 

 Number of legs 

 Number of wells 

 Water depth 

 Submerged surface area 

 Volume of water enclosed 

 Structure manned 

 Structure in production 



Temporal/meteorological 



Linear date 

 Quadratic date 

 Cubic date 

 Wind speed 

 Wind direction 

 Air temperature 



and red snapper, and a negative loading for spotted 

 seatrout (Table 3). The pelagic fish factor consisted of 

 positive loadings for dolphin, king mackerel, little tunny 

 and Spanish mackerel, and a negative loading for 

 silver/sand seatrout (Table 3). The third factor was 

 composed of high positive loadings of Atlantic croaker 

 and silver/sand seatrout (Table 3). The fourth factor 

 was composed of high positive loadings of bluefish and 

 red drum (Table 3). The fifth consisted of positive 

 loadings for cobia and sharks, and a high negative 

 loading for blue runner (Table 3). The strongest eco- 

 logical relationships within a PC existed for reef fish 

 and pelagic fish PC's. These groupings included species 

 with similar life histories, habits, and abundances. The 

 biological relationships between the species in the other 

 PC's were more tenuous; however they did provide in- 

 formation on factors relating to the species relative 

 abundance. 



Results of the MRA of ln(CPUE + 1) of spotted sea- 

 trout with the predictor variables indicated spotted 

 seatrout abundances were highest near small, non- 

 producing structures in shallow water. Fourteen 



