1. Modelling from experiments on farm and wild Atlantic salmon in nature Kjetil Hindar & Ola Diserud Norwegian Institute for Nature Research (NINA), Trondheim, Norway Phil McGinnity, Marine Institute, Ireland Ian Fleming, Ocean Sciences Centre, Nfld., Canada Genimpact, Bergen, July 2007

2. Background: wild and farm salmon  Atlantic salmon is a highly valued species  Population sizes in nature may be at an all- time low  Farm salmon production is increasing  Escaped farm salmon outnumber wild salmon in several rivers.  What are the long-term consequences?  Implications for management?

3. Wild Atlantic salmon (Salmo salar)  Consist of many semi-isolated populations in nature  Show between-population variation in molecular and quantitative genetic traits  Some differences likely represent local adaptation

4. Cultured Atlantic salmon  Outnumber wild salmon 500-fold  400 000 – 2 000 000 fish escape every year  Make up 20-40% in North Atlantic catches of salmon  Constitute ca. 20% in Norwegian rivers at spawning  Are increasingly genetically different from wild fish

5. Modelling interactions  Look at the long-term consequences using: Population vectors from surveys in Norwegian rivers Spawning success and survival vectors from experiments Input parameter values varied across likely range (Hindar et al., 2006 ICES J Mar. Sci., 63)

6. Farm and wild salmon in R. Imsa (Fleming et al., 2000 Proc. R. Soc. Lond. B, 267)  One generation: adult-to- adult in a controlled natural stream  22 farm and 18 native spawners released  16% farm success vs. native salmon  ca. 30% lower total and native smolt productivity

7. Farm and wild salmon in Burrishoole (McGinnity et al., 2003 Proc. R. Soc. Lond. B, 270)  Two generations: egg-to-adult  Farm salmon had lifetime success of 2-4% relative to Wild  ‘Hybrids’ showed intermediate fitness and decreased survival; by rank: Backcross to wild, 89% FxW hybrid (wild mother) 2nd generation hybrid Backcross to farm WxF hybrid (farm mother)27%

8. Spawning success Fish typeRelative successReference Sea-ranched male0.51 (0.29-0.71)1 Sea-ranched female0.91 (0.82-1.00)1; 2 Farm male0.13 (0.01-0.24) 3; 4 Farm female0.44 (0.20-0.82)2; 3; 4 1 Fleming et al. (1997) Behav. Ecol., 8 2 H. Lura (1995) PhD thesis, University of Bergen 3 Fleming et al. (1996) J. appl. Ecol., 33 4 Fleming et al. (2000) Proc. R. Soc. Lond B, 267

9. Spawning success: parr males Fish typeProportional successReference (% of eggs) Wild X wild 4.5 (3.0-6.0)1; 2 Wild X farm10.5 (8.0-13.0)1; 2 Farm X farm 8.5 (4.0-13.0)1; 2 1 Garant et al., (2003) Ecol. Letters, 6 2 Weir et al., (2005) Can. J. Fish. Aquat. Sci., 62 In total; parr males fertilised 23-24% of the eggs in the two experiments. Larger variation is known.

10. Survival (Fleming et al., 2000; McGinnity et al., 2003) Fish typeEgg to 0+ 0+ to smoltSmolt to adult Wild X wild111 Farm X wild; W X F0.67 (0.63-0.71)0.92 (0.54-1.13)0.85 (0.43-1.21) Farm X farm0.88 (0.49-1.43)1.06 (0.61-1.53)0.33 (0.03-0.83) Backcross to W0.75-0.801.11-1.431.03 Backcross to F0.71-0.771.03-2.070.32 2nd gen hybrid0.83-0.801.22-2.30NA

11. Basic simulation (Hindar et al., 2006. ICES J. Marine Sci., 63)  20% farm escapes each generation  Even sex ratio  Average fitness from experiments  No mature parr  Back-crosses to ’1/2 wild or farm’ and ’1/2 hybrid’  4-yr generation wild hybrid feral farm

13. Difference between rivers wild hybrid feral farm wild hybrid River Stjørdalselva, 1989-2000 12 consecutive intrusion rates: [0, 0.07, 0, 0.02, 0, 0, 0, 0.25, 0, 0, 0.03, 0.03] River Vosso, 1989-2000 [0, 0.19, 0.23, 0.46, 0.75, 0.61, 0.71, 0.59, 0.65, 0, 0, 0] Hindar et al., (2006) ICES JMS, 63

14. Recovery of high-impacted populations?  ’High’ for 10 years. Then no escapes for 100 years W H W H FF Years Proportions

15. Other effects (not modelled)  Higher growth rate of farm and hybrid offspring  Altered life-history traits Age at smoltification Age at sexual maturity  Reduced productivity: 30% fewer smolts  Increased interspecific hybridisation with trout  Diseases

16. Can we suggest limits to gene flow? Some possibilities to consider:  Intrusion rate (x modelled fitness)  Genetic difference between populations  Long-term effective population size

17. Intrusion rate of escaped farm salmon: results after 10 generations W H W FF Proportion escaped farm salmon Proportion wild hybrid feral farm

18. Genetic difference between populations  Do not allow more migrants than can be deduced from F ST = 1 / (4N e m + 1)  Limit related to difference between donor and recipient, which can be estimated by genetic analysis  F ST ~ 0.05 between rivers within continent (N e m ~ 5)  F ST > 0.25 between continents (N e m < 1)  Wild vs captive somewhere between (N e m ~ 2) Ryman (1991) J. Fish Biol., 27 (suppl A) Alternative based on quantitative trait: Tufto (2001) Amer. Natur., 158

19. Long-term effective population size (Tufto & Hindar, 2003. J. theor. Biol., 222) Two-way symmetric gene flow One-way gene flow * isolation increases total effective population size * asymmetry in gene flow decreases total effective population size Two-way asymmetric gene flow

20. Summary & recommendations  Farm escaped salmon have negative effects on wild salmon Effects cumulative over generations Population eventually composed of ‘hybrids’ and feral farm fish  Farm fish must be contained By keeping them inside the net pens By sterilization