The effects of soak time and depth on

The effects of soak time and depth on longline catch rates EB WP-3 EB WP-12 Peter Ward RAM Myers Dalhousie University

400 m 1950 s 1990 s 200 m 500 m Depth 25– 175 m 25– 500 m Dawn 35% 30% Dusk 0% 70% Soak time 5 hr 9 hr

Observer data from six fisheries North Pacific Swordfish (1, 539 sets) 40 N 20 N Western Pacific Bigeye (1, 915 sets) Central Pacific >500, 000 fish Bigeye (3, 243 sets) >6, 000 daily sets 0 20 S Western Pacific South Pacific Distant (234 sets) Yellowfin (1, 419 sets) 40 S South Pacific SBT (666 sets) 140 E 180 E 140 W

Data Catch rate (no. /1000 hooks) 2. 5 Observer record of time when each hook was retrieved Swordfish 2. 0 + 1. 5 Estimate of deployment time from start and end of time of set 1. 0 0. 5 0 0 5 10 15 Soak time (hr) 20

Generalized linear mixed model (1) Random effects (O) (2) Fixed effects • • • soak time (T) season (S) year (Y) dawn (A) dusk (P)

albatross Seabirds Soak time effect varies among species Billfishes swordfish Other fishes Sharks and rays blue shark skipjack Tunas bigeye -0. 2 0. 0 Soak time effect +0. 2

0. 0 blue shark skipjack tuna r = 0. 54 -0. 1 Soak time effect +0. 1 Soak time effect correlated with survival 0 20 40 60 Alive (%) 80 100

1. 0 -0. 5 0. 0 oilfish Dusk has a positive effect for many species black marlin -1. 0 Dusk effect 0. 5 Ray’s bream -1. 0 -0. 5 0. 0 Dawn effect 0. 5 1. 0

Effects make a substantial difference Swordfish 5 hr 20 hr no dawn or dusk 1 4 dawn and dusk 3 10 $1, 500 vs $5, 000

0 Depth (m) 100 Bigeye tuna Striped marlin Opah Day. Blue shark Night 200 300 400 500 0 1 2 Relative catchability Distribution of catches of most species varies with depth. . . and with time of day

Conclusions (1) Abundance indices need to be adjusted for: • soak time • dawn and dusk • depth range (2) Mortality of several species may be greater than indicated by catch records

Generalized linear mixed model Logistic regression catch y has a binomial distribution: y~b(n, π) π is the probability of catching a fish: eh p=( ) 1 + eh æ p ö ÷=h η is the ‘soak time effect’: log ç è 1 - p ø

Random effects • Operations drawn from a larger population of operations • Random effects in catch rate – soak time relationship for each operation are independent and normally distributed:

Probability of being on a hook Catch is the product of capture and loss rates No captures after β < 0 deployment e. g. seabirds Captures exceed losses e. g. blue shark β > 0 Losses eventually exceed captures Captures balance losses e. g. skipjack e. g. yellowfin 0 5 10 Soak time (hr) 15 β < 0 β = 0 20

Soak time effect generally consistent among areas swordfish escolar oilfish blue shark porbeagle

Epipelagic community 0 m 100 m striped marlin yellowfin tuna 200 m wahoo 300 m 400 m 500 m swordfish opah Mesopelagic community bigeye tuna