The most parsimonious ABFT model contained three variables and four interactions
glm(numbersoffish)~(dayofyear)+(latitude )+(longitude)+(dayofyear)+(latitude*dayof year) +(latitude*year) +(longitude*year)). Three variables, (day of year, latitude and longitude) and two interactions (latitude* day of year and longitude * year) were significant (p<0.05) in the ABFT model. The effect of latitude differed depending on the day of year and the effect of longitude differed across years. Average latitude of ABFT schools in the Gulf of Maine ranged from 40.95 to 42.72°N with no north or south trend (Figure 1). Average longitude of ABFT schools shifted east > 280 kilometers (-70.95 to -68.47° W) and exhibited high inter-annual variability (Figure 1). Intra-annual distribution of schools was variable. Some years contained one or two high density aggregations while others had multiple high density aggregations distributed across a broad spatial range (Figure 2). There was a significant correlation (p<0.05) between the average position of ABFT schools and the average Atlantic herring/tow in both spring (p<0.01, r2=0.24) and autumn (p<0.01, r2=0.42) surveys from offshore strata, including Georges Bank (Figure 3).
Figure 1. Surface school positions of Atlantic bluefin tuna in the Gulf of Maine.
Each location represents the mean position of bluefin schools observed by spotter pilots, from vessels or represents captured fish from the commercial fishery. Within each year, mean positions were weighted by the number of fish in each school. Error bars represent sample variance. Schools of Atlantic bluefin tuna have shifted their longitudinal distribution approximately 280km to the east (70.95 to 68.47° W).
Figure 2. Distribution of Atlantic bluefin tuna in the Gulf of Maine.
Kernel density estimations were constructed based on the number of fish observed in each school over four selected time periods. Density estimations were normalized yielding utilization distributions that displayed probability of occurrence during four time periods, A) 1979-1985, B) 1986-1992, C) 1993-1999, and D) 2000-2005.
Figure 3. Average number of Atlantic herring/tow and position of Atlantic bluefin tuna schools.
Each bluefin tuna point represents a weighted mean of surface schools (solid black circles) and is correlated with an increase in herring/tow on Georges Bank during the spring (solid black triangles) and fall trawl survey (solid black diamonds). Atlantic herring data represented as the mean herring per tow from the (National Marine Fisheries Service) autumn bottom trawl survey for survey strata on George’s Bank.
Significant differences were observed between those distances calculated from the 95th percentile utilization distributions of Atlantic herring catch locations, position of ABFT surface schools and the randomly generated ABFT surface schools (p<0.05). Results identified schools of ABFT were located closer to high densities of Atlantic herring than a population distributed at random. From a total of one hundred two-week time periods between June 1 and October 31 1996-2005, ABFT were observed within the 95th percentile utilization distribution of Atlantic herring 17 times (Figure 4). The remaining ABFT schools were located inside or along the margin of the 50th percentile utilization, or beyond areas of high herring density (Figure 3).
Figure 4. Spatial relationship between Atlantic herring and Atlantic bluefin tuna in the Gulf of Maine.
Atlantic herring kernel density estimations were constructed based on the metric tons of herring captured in each tow or set (trawler or purse seiner, respectively). Density estimations were normalized yielding utilization distributions that displayed probability of occurrence during four two week time periods, A) June 1-15 1997, B) August 16-30 2003, C) July 16-31 2004, and August 1-15 2005. Atlantic bluefin tuna (white circles with solid black borders) were located closer to high probability regions of Atlantic herring than a randomly distributed population. While the majority of bluefin tuna schools were located within areas of high herring probability, some schools of bluefin tuna did not display any association with herring.
The whole article is in the extended text Read More...