176 



Fishery Bulletin 90(1). 1992 



800 

 Irowl colch (cfob/lvj) 



Figure 4 



Relationship between trawl catch and entrainment of Dunge- 

 ness crab by a hopper dredge. The line was fit by least-squares 

 and non-parametric regression. Arrow indicates two outliers 

 which were excluded from the least-squares regression. 



To relate crab entrainment to crab density, several 

 regression models were tried. The selection of a final 

 model was based on both statistical measures of fit and 

 biological reasonableness (i.e., an expectation that en- 

 trainment should increase with increasing crab den- 

 sity). First, a test for linearity ("XLOF" in the Minitab 

 package; Minitab Inc., University Park, PA) was per- 

 formed, and no significant nonlinearity was detected 

 (p>0.10). Second, a linear least-squares regression was 

 calculated; neither the slope nor the intercept were 

 significantly different from zero for this model. How- 

 ever, this relationship was heavily influenced ("Cook's 

 Distance Measure"; Weisberg 1985) by two points. 

 When these two points were excluded, the best least- 

 squares model was (Fig. 4) 



Y = 0.27X, 



(8) 



where Y is entrainment by the dredge (crab/key), and 

 X is trawl-estimated density (crab/ha). Finally, a non- 

 parametric median-slope regression (Conover 1980) 

 was calculated using all 1 4 data points. This method 

 returned the same slope as the 12-point least-squares 

 regression. 



Entrainment for the other dredge types was calcu- 

 lated from this model based on relative entrainment 

 factors given by Stevens (1981); entrainment by a 

 pipeline dredge is assumed to be 100% of the hopper 

 dredge value (this value is controversial, but is conser- 

 vative), while a clamshell dredge entrains only about 

 5% of the hopper dredge value. 



Postentrainment mortality After entrainment, crab 

 may be killed due to physical trauma during transport 

 through pipes and pumps, burial under excessive sedi- 

 ment weight, or confined disposal in landfill by a 

 pipeline dredge. Several estimates of postentrainment 

 mortality (m in Eq. 3) have been made. For a hopper 

 dredge, Stevens (1981) reported approximately 75% 

 mortality, all sizes of crab combined. Armstrong et al. 

 (1982) reported mortality rates by crab size for a hop- 

 per dredge, with 86% mortality for crab larger than 

 50 mm carapace width (CW) and 46% mortality for 

 those smaller than 50 mm CW. Other studies indicate 

 that hopper dredge mortality rates for small (< 10 mm) 

 0-1- age-class crab range from 1% to 5% (K. Larson, 

 Portland Dist., U.S. Army Corps of Eng., pers. com- 

 mun., 1987). Gross mortality observations were also 

 made during later entrainment studies (McGraw et al. 

 1988, Wainwright et al. 1990), but these recorded only 

 obvious mutilations and so underestimate total mortal- 

 ity. We adopted a set of size-dependent mortality rates 

 for a hopper dredge based on these studies (Table 2). 

 Little information is available concerning mortality 

 of crab entrained by a clamshell dredge. Stevens (1981) 

 reported an overall mortality rate of less than 10%, 

 which seems reasonable considering the operation of 

 the gear. We have used a 10% mortality rate for a clam- 

 shell dredge for all age-classes. Because its effluent 

 goes to confined upland disposal, 100% mortality was 

 assumed for all crab entrained by the pipeline dredge. 



Simulations Scheduling of dredge operations was 

 based on engineering constraints, weather limitations, 

 avoidance of salmon migration periods, and avoidance 

 of seasons and areas with high predicted crab loss. To 

 help in this planning process, loss rates (expressed as 

 crab per volume dredged) were calculated for each area 

 and each season, based on average seasonal crab den- 

 sities and age-class composition. 



