Methods

Data used in this section was also collected during the experiments described in Chapter 2 (see 2.3 Methods, p21) but analysed with the inclusion of the non-standard mesh types. Mesh size variation only occurred in straight pots; Z pots have been excluded from this analysis to keep the catch rates comparative as design has a significant effect on catch rate (see Chapter 2). There were three mesh types;

Type A: 14 of the 19 straight pots use this mesh type. It is a flexible plastic diamond-shaped aperture mesh, with a maximum diameter of 60 mm (see Figure 3.1).

Type B: Three of the straight rigid pots were constructed using this mesh type. It consisted of two layers, an inner metal mesh with rigid square apertures, with a maximum diameter of 100 mm, and an outer layer of flexible plastic diamond-shaped aperture mesh, with a maximum diameter of 35 mm (see Figure 3.1).

Type C: Two of the straight collapsible pots were constructed using this mesh type. It consisted solely of the same metal mesh as the inner layer of pot type B, with rigid square apertures, and a maximum diameter of 100 mm (see Figure 3.1).

Table 3.1 Experimental clustering of fish pot designs and mesh types.

Cluster Pot Design Mesh type

1 2 collapsible A 3 rigid A 2 4 collapsible C 5 rigid A 3 7 collapsible A 8 rigid A 4 10 collapsible A 11 rigid A 5 13 rigid B 14 collapsible A 6 17 rigid B 18 collapsible A 7 20 collapsible A 21 rigid A 8 22 collapsible A 24 rigid B 9 25 rigid A 26 collapsible A 10 32 collapsible C

Figure 3.1 Three different mesh types. Top: mesh type A, plastic diamond mesh with a maximum aperture width of 60 mm, and fibre diameter of 2 mm. Mid: Mesh type B, double layered rigid metal and flexible plastic mesh, with a maximum aperture of 35 mm, and fibre diameter of 3 mm. Bottom: Mesh type C, rigid metal wire with a maximum aperture of 100 mm, wire diameter of 44 mm.

3.2.1.1 Mesh type analysis

Catch values (commercial value of all species caught in a deployment) were calculated using size:weight relationships for each species, and the same fish prices at a prior stage of the study were used, to keep pot revenues comparable to the model predictions of this prior stage

(Ogilvie, et al., 2010).

In order to gauge the impact of mesh size on the retention of undersized fish, thelegal size limit of tarakihi (25 cm) was used to divide all potted fish into generic undersized and legal sized categories. Two species in the fishery have no lower size limit (leatherjacket and sea perch), so any fish smaller than this 25 cm length were still of commercial value, and for blue cod and red cod the minimum size is 30 cm, also making the 25 cm cut-off inappropriate for these two species. Because the catch was mainly composed of tarakihi, and also because the WFC would prefer to catch less small fish of any species for ethical reasons (WFC, pers comm), 25 cm was still seen as a practical dividing point of fish sizes in order to gauge some idea of the proportion of smaller fish that were retained, to determine the ecological impact of mesh type. We calculated the proportion of all fish species that were undersized (i.e. < 25 cm) by dividing these by the total number of fish caught.

Inspection of the data indicated that the distribution of both catch value/pot and the proportion of undersized fish were not normally distributed. Accordingly, the effect of the mesh type (A versus B versus C) and location (Cape Campbell versus Cape Jackson) and their 2nd order interaction were analysed using a generalised linear model with a negative binomial error distribution and a log-link function (a test for catch value, and a test for proportion of undersized fish). Cluster was also included to account for variation in fish density between clusters. The tests were run using the GenStat statistical package (Version 13). When the model indicated significant differences post-hoc, pairwise comparisons of mean values (i.e. mesh types) were undertaken using Fishers restricted LSD test at α=0.05.

3.2.2 Soak time

Data used in this section was collected during the experiment described in Chapter 2 (see 2.3 Methods). Catch values from Z pots (equipped with standard mesh type A) were analysed; straight pots were excluded from analyses because of the significant effect of pot design on catch value.

3.2.2.1 Soak time analysis

Catch values (commercial value of all species caught in a deployment) were calculated using size:weight relationships for each species, and the same fish prices at a prior stage of the study were used, to keep pot revenues comparable to the model predictions of this prior stage

(Ogilvie, et al., 2010). To calculate the catch per hour values, each catch value was divided by its soak time in hours. Soak times were recorded as an exact number of hours, but for analysis were categorised into 5 levels (1, 2, 3, 4, and 6), corresponding to the soak time in days. Inspection of the data indicated that the distribution of catch value/pot and catch/hour were not normally distributed. Accordingly, the effect of the soak time (1, 2, 3, 4 and 6) and location (Cape Campbell versus Cape Jackson) were analysed using a generalised linear model with a negative binomial error distribution and a log-link function. The test was run using the GenStat statistical package (Version 13). When the model indicated significant differences post-hoc, pairwise comparisons of mean values (i.e. soak times) were undertaken using Fishers restricted LSD test at α=0.05.

The presence of a conger eel was recorded for each deployment. The effect of soak time on the presence/absence of a conger was analysed using a generalised linear model with a Bernoulli error distribution and a logit function. The test was run using the GenStat statistical package (Version 13). When the model indicated significant differences post-hoc, pairwise comparisons of mean values (i.e. soak times) were undertaken using Fishers restricted LSD test at α=0.05.

3.3

Results