Power and May: Sea-surface temperatures and Thunnus albacares catch and effort 



433 



quarter the length of the set. We assumed that these 

 circular sampling areas represented the area fished by 

 the gear and/or traversed by the fish during the set. 

 The mean, median, and coefficient of variation of the 

 SST and MSTG pixel values contained within the three 

 concentric circular regions were computed for each set. 

 These SST and MSTG statistics computed for the cir- 

 cular regions (polygons) encompassing the sets were 

 then plotted against the corresponding yellowfin tuna 

 CPUE at that same location. The potential bias due to 

 spatial and temporal variation in cloud coverage was 

 examined by classifying each circular polygon into one 

 of three groups based on the percentage of cloud cover 

 (25-49%, 50-74%, >74%) and comparing the results 

 with CPUE and the statistics tabulated from the image 

 data. Plots of CPUE versus the circular polygon 

 statistics by month were also done to determine 

 whether associations between SST variability and 

 yellowfin tuna CPUE were present only at certain 

 times of the year. 



SEP87 NOV87 JAN88 



Date 



Figure 2 



Number of longline sets reported each day in the Gulf of Mex- 

 ico, September 1986-December 1987. Curve is fit using robust 

 locally-weighted regression (Cleveland 1979). Vertical bars at 

 the top margin of the plot are drawn at the corresponding 

 dates of the satellite images used in the analyses. 



Results 



Yellowfin tuna database 



The fishery data include only records for which geo- 

 graphic coordinates were available. In that respect, the 

 results may be biased if longline sets records lacking 

 geographic coordinates occurred in a given region, or 

 had anomalously high or low yellowfin tuna catches. 



The longline fishery operated nearly every day in the 

 Gulf of Mexico from September 1986 to December 

 1987. Usually 10-20 sets were recorded each day, with 

 very few days when no sets were reported (Fig. 2). The 

 frequency of sets appeared to increase in late 1986 to 

 midsummer 1987, followed by a decline in autumn 

 (Figure 2 and subsequent figures include a curve, 

 similar to a running mean, that was fit using Cleve- 

 land's (1979) technique of robust locally-weighted 

 regression). Fishing effort during 1986-87 was heavily 

 concentrated in the northwestern Gulf of Mexico, 

 where extensive seasonal variation in SST was ob- 

 served. Sets were mainly deployed between 26°N and 

 29°N latitude, and were fairly uniformly distributed 

 west of the Mississippi River Delta between 88° and 

 96°W longitude (cf. Fig. 3). 



Yellowfin tuna are apparently present and available 

 to the longline fishery, at least in small numbers, 

 throughout the year, since 79.2% of the sets caught 

 one or more yellowfin tuna (Fig. 4). There was a 

 noticeable decline in the daily percentage of sets 

 catching at least one yellowfin tuna in the early months 

 of 1987 (Fig. 4), but because the data set did not cover 



98 96 94 92 



Figure 3 



Geographic location of longline sets deployed in the Gulf of 

 Mexico during March 1987. 



multiple years, it is uncertain whether this is an annual 

 phenomenon. 



The mean yellowfin CPUE for the entire data set 

 (6618 sets) was 14.2 fish/1000 hooks with a standard 

 deviation of 18.9. The median CPUE was 8.9 fish/1000 

 hooks. These CPUE figures are comparable to those 

 reported by Polacheck (1989) for yellowfin tuna in the 

 Pacific. The daily mean yellowfin CPUE may be 

 elevated in the fall months and depressed during the 

 late-winter and early-spring months (Fig. 5). 



