Figure 1.
Krill sliced at cross section 1 (CS1) and cross section 2 (CS2).
Figure 2.
Shapes of CS1, CS2 and CS3. CS1 is described by a flexellipse_1, CS2 is described by a flexellipse_3 and CS3 is described by a flexdrope_2 (see appendix S1).
Figure 3.
Scanned pictures of krill with markings to illustrate the width and height at CS1 and CS2.
Table 1.
Model parameters and fit-statistics for model enabling that descriptions of CS1 and CS2 can be created based on height and width values of the cross sections alone.
Figure 4.
Image of the netting wall (mesh size15.4 mm) of the krill trawl captured during fishing operations (A).
Digitizition of selected meshes to establish realistic values for mesh opening angles (oa) (B). The camera is located 10 m from the codline-end, pointing backwards. The intire 15.4 mm trawl was covered with 200 mm double 5 mm PE diamond netting for protection.
Figure 5.
Illustration of how the effective mesh openings in a trawl can vary depending of the attack angle of krill (5–90°) (A).
The light mesh is the real mesh in 90° view and the dark mesh is the effective mesh opening when projecting the mesh opening to a plan perpendicular to the towing direction. The lower mesh series (B) shows the potential effect of rotation of the cross section (CS1_CS2) in a 30° open mesh.
Table 2.
Mesh size measurements based on image analysis (mesh id 1–5) and fit statistics for using a diamond mesh description (R2) and measured mesh opening angles (oa values in degrees) based on underwater video recordings during commercial fishing (mesh id 6–10).
Table 3.
Dates, body length and carapax width measurements of 30 juvenile, sub adult and adult krill collected fresh from the trawl catch (25. January) off South Orkney Islands and after two and 10 months preserved in borax-buffered formalin (4%).
Table 4.
Fit statistic for the used models to describe CS1, CS2 and CS3.
Table 5.
Estimated regression coefficients for CS1 and CS2.
Figure 6.
Diamond mesh design guide for Krill, based in combination of CS1 and CS2.
The plot gives iso-L50 curves as a function of mesh size (mm) and mesh opening angle (oa).
Figure 7.
Diamond mesh design guide for CS3.
The plot gives iso-L50 curves as a function of mesh size (mm) and mesh opening angle (oa).
Figure 8.
Top plot show paired gear fit of the experimental data.
The population structure of the measured individuals are indicated (solid line = 7 mm survey trawl (Macroplankton trawl); broken line 15.4 mm commercial trawl). The lower plot is the size selection curve including 95% confidence limits.
Figure 9.
The experimentally obtained L50 value is indicated with the solid line (exp L50).
L50 predictions for CS1_CS2 and for CS3 are indicated for the commercial mesh size (15.4 mm). The realistic mesh opening angles during commercial fishing is indicated with the vertical gray interval. The horizontal interval indicated the selective range for the 15.4 mm commercial mesh size.
Table 6.
Values for the length of krill (L5-95) with different fixed retention likelihoods based on selectivity data from experimental fishing.
Table 7.
Selectivity estimates for the commercial 15.4-statistics based on experimental fishing.
Figure 10.
The effect of krill encountering meshes with different orientation.
The penetration model (CS1_CS2) is rotated, at optimal attack angle perpendicular to the netting from 0–90°. 0° is a dorso ventral orientation equal to normal swimming orientation.
Figure 11.
The effect of varying attack angles in the penetration model for CS1-CS2 in the range of oa-values that were found relevant in the 15.4 mm trawl.
Figure 12.
Underwater images captured during fishing indicating escaping krill in the 15.4
The escapees are marked with red arrows in the lower photo and clearly demonstrate an optimal orientation of the krill escaping. The photo is taken 10 meters in front of the cod line during commercial fishing.
Figure 13.
Experimentally obtained data (black line) with 95% confidence limits (broken line).
Thick line (gray) is the predicted selectivity curve based on morphological based measurements (FISHSELECT) and the distribution of opening angle (oa)-values given in table 8.
Figure 14.
Predicted selectivity of krill in different mesh sizes based on the weight factors of the different opening angle (oa)-values for optimal CS1_CS2.
Predictions are made from 6 to 28
Table 8.
Simulated distribution of opening angles (oa) grouped in 5 cm. intervals that will result in identical selectivity curves for krill between simulated and experimental data.
Table 9.
Predicted selectivity parameters (L50 and SR) for diamond mesh size ranging from 6 mm to 28 mm.