Overholfz and Friedland: Recovery of the Gulf of Maine-Georges Bank complex 



595 



the differential recovery by area by using NEFSC bottom 

 trawl data from 1963 to 1998. We also wanted to examine 

 two theories regarding the recovery of Georges Bank her- 

 ring: 1) that it was due to a resurgence of extant spawners 

 (Stephenson and Kornfield, 1990) or 2) that is was due to 

 recolonization from adjacent components of the complex 

 (Smith and Morse, 1993). 



Methods 



Details of the statistical design, areal coverage, standard- 

 ization methods, gear type, and species catchability in the 

 NEFSC bottom trawl surveys are provided in Grosslein 

 ( 1969); Azarovitz ( 1981); Byrne et al. ( 1981), and a NEFSC 

 report (NEFSC, 1988). Briefly, the surveys are based 

 on a stratified random research design that covers the 

 region from Cape Hatteras to the Scotian Shelf at depths 

 between 27 and 200 m (15 to >100 fathoms), each spring 

 and autumn. 



We used data from spring ( 1968-98) and autumn ( 1963- 

 98) bottom trawl surveys to produce indices of abundance, 

 and plot distribution patterns for the Gulf of Maine- 

 Georges Bank Atlantic herring complex. For spring, we 

 used the same strata set (strata 1-30, 33-40, and 61-76) 

 (Fig. 3) as used in the assessment of the herring complex 

 (NEFSC^). We produced indices of total abundance (swept 

 area numbers Imillions]) to follow changes in the stock 

 complex over time, assuming that a standardized survey 

 tow covers 0.1 nmi and that all herring are available to 

 the gear 



For autumn we used a smaller strata set (1-30, 33-40) 

 (Fig. 3) to generate indices of abundance (area-swept num- 

 bers) because herring are seldom found south and west 

 of Long Island during this season (NEFSC^). Changes in 



abundance by subregion (reflecting abundance trends by 

 spawning components) were assessed by using indices for 

 the Gulf of Maine (strata 26-28, 37-40), Nantucket Shoals 

 (Great South Channel area) (strata 9-11, 23-25), and 

 Georges Bank (strata 13-14, 16-17, 19-22, 29-30) (Fig. 3). 



We produced box plots that represent the median sur- 

 vey catch location aggregated over a latitudinal gradient 

 (35-45°N latitude) collapsed over the entire east-west 

 longitude of the region. We repeated the same procedure 

 for the longitudinal gradient (64-76°W longitude). These 

 plots show the median, inter quartiles (25% and 75%), 

 whiskers ( 1.5 times the absolute value of the interquartile 

 range) and outliers of the data. We also constructed maps 

 of spring survey catch locations for 1970, 1975, 1980, 1985, 

 1990, and 1995 to show trends in herring distribution and 

 abundance over the time period. Changes in distribution 

 and abundance during autumn 1963-98 are depicted in 

 plots of Atlantic herring catch locations for 1965, 1970, 

 1975, 1980, 1985, 1990, and 1995. 



We also produced box plots of bottom temperature 

 for the sites where herring were captured during both 

 seasons. For spring and autumn, we also plotted the pro- 

 portion of survey tows with catches of Atlantic herring 

 as a measure of how often surveys encountered Atlantic 

 herring over the time series. We used LOWESS (locally 

 weighted regression smoothing) (Cleveland, 1979) to make 

 the temporal trends in both of these time series easier to 

 follow. 



Results 



Atlantic herring abundance indices from spring surveys 

 changed markedly during 1968-98 (Fig. 4A). A period of 

 relatively high abundance (area swept, millions offish), in 



