462 



Fishery Bulletin 104(3) 



population study and was excluded from the survey. 

 Carmichael et al. (2003) used transects located on a 

 grid and visual counts in a shallow clear-water estuary 

 to estimate abundance of the population in Pleasant 

 Bay, Cape Cod, Massachusetts. Other estimates of the 

 Delaware Bay population have been based on offshore 

 surveys during nonspawning periods when populations 

 are dispersed and possibly mixed (Botton and Haskin, 

 1984; Botton and Ropes, 1987; Hata and Berkson, 2003), 

 Our mark-recapture estimates apply to adult horseshoe 

 crabs present in Delaware Bay during late May 2003 

 when spawning peaked. 



In previous horseshoe crab population estimates that 

 were based on offshore surveys (Botton and Haskin, 

 1984; Botton and Ropes, 1987; Hata and Berkson, 

 2003), capture efficiency was unknown, and adults that 

 remained in estuaries (i,e., those that did not migrate to 

 the ocean after spawning) were not sampled. Hata and 

 Berkson (2003) concluded that the capture efficiency for 

 their trawl survey was intermediate between that of the 

 trawl survey reported in Botton and Haskin (1984) and 

 that of the hydraulic dredge survey reported in Botton 

 and Ropes (1987). The most recent estimate of offshore 

 abundance of 7.1 million crabs was reported by Hata 

 and Berkson (2003). Botton and Haskin (1984) reported 

 densities that were 1.8 to 2.9 times the densities report- 

 ed by Hata and Berkson (2003), which would indicate a 

 population estimate of 12 to 20 million according to the 

 data of Botton and Haskin (1984). Botton and Ropes 

 (1987) reported a minimum population of 2.3 to 4.1 mil- 

 lion. The proportion of adults that remain in the Dela- 

 ware Bay and do not migrate to the continental shelf 

 after spawning is unknown. However, if the proportion 

 of nonmigratory adults is sizeable (e.g., on the order 

 of 0.3, which was indicated by our fishery-independent 

 catches) then that, along with gear inefficiencies in 

 trawl surveys, could explain the difference between the 

 estimates in our present study and those of Hata and 

 Berkson (2003). 



Bias due to assumption violation is another reason 

 for differences in estimates. We designed our study to 

 obtain a representative sample when horseshoe crabs 

 were concentrated and to minimize the time between 

 release and recapture periods so that only a few days 

 separated the prepeak release and recapture occasions. 

 We released 17,543 tagged horseshoe crabs over two 

 periods and examined 28,738 horseshoe crabs for tags, 

 counting them in quadrats during subsequent spawn- 

 ing surveys (29 May, 31 May, and 2 June 2003). Low 

 recapture rates (<1%) were consistent with a large 

 population. However, it was important to evaluate the 

 potential effects of assumption violations. Tag loss and 

 tag-induced mortality were likely to be trivial because 

 of the short period during which crabs were at large 

 and results from field studies and laboratory experi- 

 ments have shown no tag loss or tag-induced mortality 

 (Crawford'^; Brousseau et al., 2004). Brousseau et al. 

 (2004) attached combined acoustic and radio tags and 

 standard button tags to 24 female horseshoe crabs along 

 two beaches in Delaware Bay, and then tracked them 

 for eight days. All 24 were detected at least once, and 

 20 spawned on the beach of release within eight days 

 of release, indicating that handling and tagging had a 

 minimal effect on spawning behavior. Tags could have 

 been overlooked during the spawning survey, which 

 is why we limited our analysis to males counted and 

 recaptured within 1-m- quadrats. 



We evaluated the effect of capture and tagging on 

 spawning behavior, and found no evidence that tagged 

 males delayed spawning. Using radio-tagged horseshoe 

 crabs, we believe it is possible that tagged crabs termi- 

 nated spawning, and we adjusted abundance estimates 

 based on relocation rates of radio-tagged crabs. Thus, we 

 based our inference on abundance estimates that were 

 adjusted downward to account for that possibility. 



Recapture rates were low over the short time period 

 for our study. However, annual recapture rate was ap- 

 proximately 4% for all tag releases and recaptures. 



