NOTE DeVries: No evidence of bias from fisli befiavior in tfie selectivity of size and sex of Pagrus pagrus by tiook-and-llne gear 



583 



evidence that the behavior-related (as opposed to gear- 

 or fishery-related) size or sex selectivity reported for 

 gag and scamp also occurred in fishery-independent 

 hook-and-line catches of the similarly protogynous, co- 

 occurring red porgy (Pagrus pagrus) (Sparidae). In 

 other words, at a given location, were males or larger 

 individuals (more likely to be males in a protogynous 

 species) more aggressive and more likely to bite a baited 

 hook and be caught before females or smaller (more 

 likely to be female) individuals? Such selectivity has 

 never been experimentally demonstrated in any pro- 

 togynous species. 



The red porgy is found in warm temperate to sub- 

 tropical waters on both sides of the N. Atlantic, in- 

 cluding the northern GOM and Mediterranean Sea, 

 and in the S.W. Atlantic from Venezuela to Argentina 

 (Manooch and Hassler, 1978). This sparid is one of the 

 most abundant, potentially exploitable reef fishes in 

 the northeastern GOM but is only lightly fished there, 

 entirely by hook-and-line gear. Red porgy are, however, 

 often taken as bycatch by fishermen targeting other reef 

 species. In contrast, the population(s) of red porgy in 

 the Atlantic waters off the southeastern United States 

 declined so steeply during the 1980s and 1990s (89% 

 drop in spawning stock biomass and a two-orders-of- 

 magnitude drop in recruitment to age 1), presumably 

 because of overexploitation, that a one-year moratorium 

 on their harvest and possession was enacted in 1999 

 (Vaughan and Prager, 2002). 



It is important to establish whether behavior-related 

 size or sex selectivity occurs in hook-and-line catches of 

 red porgy. From a sampling (whether fishery-dependent 

 or fishery-independent) and assessment standpoint, such 

 selectivity could lead to very biased conclusions about 

 population demographics and dynamics, life history 

 traits, and stock status. In addition, the potential for 

 this sort of selectivity to rapidly skew sex ratios, espe- 

 cially if it occurs in conjunction with size-selective har- 

 vesting caused by the gear and the targeting behavior 

 of fishermen, could partly explain the apparent crash 

 of the Atlantic population(s) of red porgy off the south- 

 eastern United States. Also, given the large number of 

 exploited protogynous species among sparids and many 

 other families, and the need to predict the effects of 

 fishing, it is important to determine whether such be- 

 havior-related selectivity characterizes most fishes with 

 this mating system or whether it is restricted to certain 

 serranids. If demonstrated to occur in red porgy, this 

 would be the first experimental evidence of selectivity 

 caused by the behavior of a protogynous species other 

 than a serranid. 



Materials and methods 



Red porgy were collected with standardized hook-and- 

 line gear during February-November 2000 at seven 

 low-relief (less than 0.5 m) hard (live) bottom sites at 

 41-67 m depths in the northeastern GOM, off Panama 

 City, FL. The sites were a subset, and the deepest, of 



nine sites that had been regularly sampled for a larger 

 study of small-scale spatial variation in population traits 

 of red porgy (Fig. 3.1 in DeVries, 2006). Site selection 

 was driven primarily by logistic constraints. Locations 

 with the most consistently high catch rates were chosen 

 to maximize the amount of information collected and to 

 enable among-collection comparisons. Gear consisted 

 of one or two size-1 Mustad j-style hooks (O. Mustad 

 and Son (USA) Inc., Auburn, NY) attached to 5-10 cm 

 droppers on a 27 kg test monofilament leader with a 

 terminal lead sinker of varying (depending on current 

 and wind conditions) weight. Bait consisted of squid and 

 occasionally cut fish. All fishing was undertaken by two 

 to four anglers (with varying levels of experience) during 

 daytime hours and from an anchored boat. 



Target sample size each time a site was fished was 

 25-30 red porgy, although for logistic reasons (primar- 

 ily variable catch rates) that goal was not always met. 

 For each collection, each fish was tagged with a unique 

 sequential identification number indicating the order 

 in which it was captured, and then placed on ice. The 

 next day fish were measured to the nearest mm total 

 length (TL) and sex was determined macroscopically, 

 after which histological samples were taken from the 

 gonads of any fish identified as female or transitional 

 (DeVries, 2006). Final determination of sex was based 

 on histological findings. 



Nonparametric runs tests "above and below the me- 

 dian" and "up and down" (Sokal and Rohlf, 1981a; Zar, 

 1999) were used to look for evidence of behavior-re- 

 lated size selectivity in individual collections. Critical 

 values for determining acceptance or rejection of the 

 hypothesis that the order was random were obtained 

 from statistical tables (Sokal and Rohlf, 1981b; Zar, 

 1999). Total length was the variable used in both tests. 

 In general, runs tests are used to determine whether 

 the order of a sequence of observations is random or 

 whether each observation is independent of its prede- 

 cessor. A run is defined as a sequence of one or more 

 like elements preceded and followed by unlike elements 

 (Sokal and Rohlf, 1981a). In the "above and below the 

 median" test, observations above the median are labeled 

 pluses and those below as minuses. The order of those 

 pluses and minuses is then tested for randomness. If 

 fish caught at the beginning of the collection tended 

 to be larger, as would occur if larger individuals (more 

 likely to be males) were more aggressive and tended to 

 bite the hook before smaller ones (more likely females), 

 there would be fewer runs than expected. The "up and 

 down" test, particularly designed for trend data, exam- 

 ines the sequence of the signs of the difference from 

 the previous value (Sokal and Rohlf 1981a). Again, if 

 larger fish tended to be caught before the smaller ones, 

 the signs would be mostly negative and therefore there 

 would be fewer changes in sign than if the sequence of 

 lengths was random. 



A runs test for dichotomized data (Sokal and Rohlf, 

 1981a; Zar, 1999) was used to determine if the order 

 in which sexes were captured in a given collection was 

 random. Because it was not known if transitional-stage 



