TY - JOUR
AU - Carr,, Jonathan
AB - Abstract Acoustic telemetry of surgically tagged fish has become a powerful tool for quantifying the survival of fish as they migrate. However, when a transmitter is not retained within the body cavity, survival estimates will be underestimated. In this study, we quantify transmitter retention rates and mortality of hatchery Atlantic salmon smolts tagged with various sized transmitters during the transition from fresh to salt water. Final retention rates ranged from 34 to 85% depending on transmitter type and the surgeon performing the tagging. Transmitter expulsion occurred for all transmitter types and predominantly occurred over a 40-day period from 25 to 65 days post-tagging. Expulsion rates were significantly higher for transmitters that weighed >7.5% of total smolt weight. Mortality was only significantly higher than sham treatments for transmitters over 12% of total smolt weight. Accounting for transmitter expulsion when modelling survival for tagged fish is important, otherwise, there is potential for biased estimates. However, the degree of bias will depend on transmitter size, life stage tagged, and study duration. Introduction Acoustic telemetry involves fitting transmitters (hereafter “tags”) to fish, which can then be detected by receivers along migration routes (Thorstad et al., 2013). Telemetry is increasingly being used to monitor the temporal and spatial variation in the survival of migratory fish species (Welch et al., 2007; Brown et al., 2010; Deters et al., 2010; Sandstrom et al., 2013; Jepsen et al., 2019). Survival for tagged individuals can be estimated using the Cormack-Jolly-Seber method and its variants when one considers detections as remotely sensed recaptures (Dudgeon et al., 2015). An inherent assumption involved with remote sensing of individuals is that the mark, or acoustic tag in this case, is retained by the fish throughout the time frame of the study (Welch et al., 2007) and that the mortality and behaviour of tagged fish is representative of the population (Peven et al., 2005; Thiem et al., 2011; Brown et al., 2013). Previous studies have shown that the process of surgically implanting tags can negatively affect the survival, behaviour, performance, and growth of fish (Peake et al., 1997; Connors et al., 2002; Brown et al., 2006; Welch et al., 2007), and survival estimates can be biased downwards if the fish expels its tag and migrates undetected past receiver arrays (Lacroix et al., 2004; Ammann et al., 2013). Surgical implantation of acoustic tags is most commonly intracoelomic; the transmitter is inserted within the body cavity of a fish (Thiem et al., 2011; Ammann et al., 2013) and thus, the size of tag is limited to the amount of space available within the cavity. Tag expulsion can occur when the incision or suture punctures do not heal properly, when it is passed through the muscular tissue of the body wall or, less commonly, through the intestine (Jepsen et al., 2002; Welch et al., 2007; Thorstad et al., 2013). Previous research suggests the main factors affecting tag retention are the length and/or weight of a tag relative to the length and/or weight of the fish (Lacroix et al., 2004; Welch et al., 2007; Brown et al., 2010) and the surgeon performing the tagging (Deters et al., 2010; Sandstrom et al., 2013). Advancements in acoustic telemetry technology have resulted in smaller more powerful tags available to researchers (Cooke et al., 2003; Brown et al., 2013). These developments have allowed salmonid researchers to track smaller fish over larger spatial and temporal scales (Welch et al., 2007; Sandstrom et al., 2013), thereby providing, a more representative sample of the population migratory behaviour (Cook et al., 2014). However, the longer the time between post-surgery detections, the less likely the fish are to retain the tag and survive (Lacroix et al., 2004; Welch et al., 2007; but see Sandstrom et al., 2013). Therefore, if the movement and mortality of tagged fish is to be a reliable representation of the population they were drawn from, it must be emphasized to account for tag retention over extended periods of time (Cook et al., 2014). The effects of tagging have been well documented for anadromous salmonids as these species begin long migrations as relatively small fish (10–16 cm; Brown et al., 2006, 2010, 2013; Deters et al., 2010; Ammann et al., 2013), however, the majority of tagging studies are spanned relatively short periods of time (<221 days; Ammann et al., 2013; Sandstrom et al., 2013; but see Lacroix et al., 2004) and been conducted in fresh water (Cooke et al., 2011). During salmonid transition from fresh to salt water, there are several morphological and physiological changes that take place in preparation for life in salt water (McCormick et al., 1998, 2000) resulting in increased levels of stress (Handeland et al., 1996). During this time, salmon are more susceptible to stress from handling (Carey and McCormick, 1998). Given many telemetry studies tag salmonids during the smolt life stage, it is important to quantify tag retention rates as they transition to salt water. With the expansion of acoustic telemetry infrastructure and its adoption as a tracking method through fresh and salt water environments, there is a need to investigate tag retention rates over longer periods of time while accounting for natural conditions. In this study, Atlantic salmon (Salmo salar) smolts were tagged with dummy acoustic tags of varying sizes to quantify the effects of tag size and the surgical effects on tag retention and smolt survival. As the experiments bracketed the smolt transition from fresh to salt water and covered a period of over 400 days after tagging, this study holds important implications for long-term acoustic telemetry studies on anadromous salmonid species and will provide information to fishery managers when determining, which tags to use for specific studies. Material and methods All methodologies used during this study were approved by an animal care committee and followed the Canadian Council of Animal Care guidelines. The fish used in this study were first generation hatchery smolts originating from wild St John River Atlantic salmon broodstock collected and spawned at the Fisheries and Oceans Canada Mactaquac Biodiversity Facility in 2012 (experiment one) and 2014 (experiment two). Each batch of fish was reared from egg to smolt at the Biodiversity Facility before being transferred on a truck in oxygenated tanks to the Fisheries and Oceans Biological Station in St Andrews, NB. Rearing environment Upon arrival (experiment one: 21 May 2014; experiment two: 2 May 2016) the smolts were placed in two circular holding tanks (6000 l) with dechlorinated fresh water with a flow-through rate of 38–76 l min−1. In experiment one, all fish were moved to a single tank (13 000 l) on 12 December 2014. After surgical tag implantation, fish were transitioned from fresh to 31 ppt salt water sourced from Passamaquoddy Bay over 6–8 days (experiment one: 5–14 days after tagging; experiment two: 7–8 days after tagging). Salmon were kept on a natural photoperiod and fed daily a commercial salmonid pellet diet (Nutra spirit, Skretting, Canada) until satiation. Water temperatures were monitored hourly in both experiments. In experiment one, water temperature from May to September averaged 11.01 ± 1.58°C (range 8.0–14.0°C), and from September to January averaged 9.93 ± 3.13°C (range 2.7–14.3°C). In experiment two, water temperature averaged 11.31 ± 2.00°C (range 7.4–14.7°C) throughout the study. Tagging treatments Experiment one Smolts (N = 320) were evenly divided between four tagging treatments (Table 1). The sham treatment had no tags implanted but underwent the same surgical procedure as fish tagged with acoustic tags. The other three treatments involved the surgical insertion of one of three Amirix/Vemco (Nova Scotia, Canada) dummy acoustic tags (V8, V9-1x, and a V9-6x with the same specifications as functioning tags but without electronics; Table 2) into the coelomic cavity of the fish. For individual identification, a 12 mm passive integrated transponder (PIT) tag (Biomark Inc., Idaho, USA) was inserted in the dorsal musculature between the lateral line and dorsal fin. Table 1. The mean and standard deviation (mean ± SD) for fish size and surgical effects and the sample size (n) for each treatment and surgeon combination in experiment one and two. Surgeon Surgeon experience Treatment Fork length Tag burden Handling time Surgery time n Experiment one  SA High Sham 15 ± 0.8 0 324 ± 423 125 ± 17 20  SA High V8 15.1 ± 0.7 6.9 ± 1.0 435 ± 115 130 ± 22 20  SA High V9-1x 14.9 ± 0.6 12.4 ± 1.4 503 ± 163 153 ± 27 20  SA High V9-6x 14.8 ± 0.9 10.6 ± 2.0 485 ± 204 146 ± 19 20  SB Low Sham 14.9 ± 0.7 0 554 ± 107 186 ± 57 20  SB Low V8 15.5 ± 0.8 6.5 ± 1.5 547 ± 73 183 ± 27 20  SB Low V9-1x 15.1 ± 1.0 12.8 ± 2.5 556 ± 77 203 ± 31 20  SB Low V9-6x 15.4 ± 0.8 9.6 ± 1.6 549 ± 75 194 ± 17 20  SC High Sham 15.2 ± 0.9 0 584 ± 90 205 ± 27 20  SC High V8 15.1 ± 1.0 6.9 ± 1.3 569 ± 52 215 ± 25 20  SC High V9-1x 14.9 ± 1.0 13.3 ± 2.3 661 ± 282 211 ± 30 20  SC High V9-6x 15.5 ± 0.9 9.4 ± 1.6 589 ± 135 212 ± 27 20  SD Low Sham 15.7 ± 1.0 0 510 ± 90 182 ± 32 20  SD Low V8 15.1 ± 0.8 7.3 ± 1.1 492 ± 81 191 ± 28 20  SD Low V9-1x 15.3 ± 1.0 12.3 ± 2.3 543 ± 98 212 ± 40 20  SD Low V9-6x 15.4 ± 1.0 9.9 ± 1.6 520 ± 68 209 ± 37 20 Experiment two  SA High Control 15.0 ± 1.7 0 167 ± 50 NA 31  SA High V5 15.3 ± 1.6 2.2 ± 0.7 292 ± 54 46 ± 9 31  SA High V7 15.7 ± 1.3 4.7 ± 1.3 353 ± 55 81 ± 12 31  SA High V8 16.0 ± 1.3 5.0 ± 1.4 363 ± 53 78 ± 13 32  SA High V9-6x 15.9 ± 1.2 7.2 ± 1.9 377 ± 52 92 ± 26 31  SB Low Control 15.5 ± 1.5 0 289 ± 75 NA 32  SB Low V5 16.2 ± 1.3 1.9 ± 0.5 290 ± 39 65 ± 21 31  SB Low V7 16.1 ± 1.4 4.4 ± 1.2 347 ± 49 79 ± 23 31  SB Low V8 15.7 ± 1.6 5.5 ± 1.8 374 ± 55 99 ± 14 31  SB Low V9-6x 15.8 ± 1.5 7.7 ± 2.4 383 ± 69 100 ± 21 32  SE High Control 15.8 ± 1.4 0 231 ± 40 NA 31  SE High V5 15.7 ± 1.4 2.1 ± 0.7 294 ± 68 80 ± 23 31  SE High V7 15.7 ± 1.5 4.9 ± 1.5 331 ± 54 95 ± 28 32  SE High V8 15.9 ± 1.4 5.3 ± 1.7 402 ± 58 127 ± 26 31  SE High V9-6x 15.5 ± 1.5 7.3 ± 2.1 425 ± 42 137 ± 25 31  SF Low Control 15.2 ± 1.7 0 178 ± 49 NA 31  SF Low V5 15.7 ± 1.5 2.1 ± 0.5 314 ± 74 101 ± 50 32  SF Low V7 15.8 ± 1.5 5.0 ± 1.6 374 ± 91 151 ± 24 31  SF Low V8 16.0 ± 1.5 5.3 ± 1.6 426 ± 82 156 ± 22 31  SF Low V9-6x 16.1 ± 1.3 8.3 ± 2.5 424 ± 70 163 ± 22 31 Surgeon Surgeon experience Treatment Fork length Tag burden Handling time Surgery time n Experiment one  SA High Sham 15 ± 0.8 0 324 ± 423 125 ± 17 20  SA High V8 15.1 ± 0.7 6.9 ± 1.0 435 ± 115 130 ± 22 20  SA High V9-1x 14.9 ± 0.6 12.4 ± 1.4 503 ± 163 153 ± 27 20  SA High V9-6x 14.8 ± 0.9 10.6 ± 2.0 485 ± 204 146 ± 19 20  SB Low Sham 14.9 ± 0.7 0 554 ± 107 186 ± 57 20  SB Low V8 15.5 ± 0.8 6.5 ± 1.5 547 ± 73 183 ± 27 20  SB Low V9-1x 15.1 ± 1.0 12.8 ± 2.5 556 ± 77 203 ± 31 20  SB Low V9-6x 15.4 ± 0.8 9.6 ± 1.6 549 ± 75 194 ± 17 20  SC High Sham 15.2 ± 0.9 0 584 ± 90 205 ± 27 20  SC High V8 15.1 ± 1.0 6.9 ± 1.3 569 ± 52 215 ± 25 20  SC High V9-1x 14.9 ± 1.0 13.3 ± 2.3 661 ± 282 211 ± 30 20  SC High V9-6x 15.5 ± 0.9 9.4 ± 1.6 589 ± 135 212 ± 27 20  SD Low Sham 15.7 ± 1.0 0 510 ± 90 182 ± 32 20  SD Low V8 15.1 ± 0.8 7.3 ± 1.1 492 ± 81 191 ± 28 20  SD Low V9-1x 15.3 ± 1.0 12.3 ± 2.3 543 ± 98 212 ± 40 20  SD Low V9-6x 15.4 ± 1.0 9.9 ± 1.6 520 ± 68 209 ± 37 20 Experiment two  SA High Control 15.0 ± 1.7 0 167 ± 50 NA 31  SA High V5 15.3 ± 1.6 2.2 ± 0.7 292 ± 54 46 ± 9 31  SA High V7 15.7 ± 1.3 4.7 ± 1.3 353 ± 55 81 ± 12 31  SA High V8 16.0 ± 1.3 5.0 ± 1.4 363 ± 53 78 ± 13 32  SA High V9-6x 15.9 ± 1.2 7.2 ± 1.9 377 ± 52 92 ± 26 31  SB Low Control 15.5 ± 1.5 0 289 ± 75 NA 32  SB Low V5 16.2 ± 1.3 1.9 ± 0.5 290 ± 39 65 ± 21 31  SB Low V7 16.1 ± 1.4 4.4 ± 1.2 347 ± 49 79 ± 23 31  SB Low V8 15.7 ± 1.6 5.5 ± 1.8 374 ± 55 99 ± 14 31  SB Low V9-6x 15.8 ± 1.5 7.7 ± 2.4 383 ± 69 100 ± 21 32  SE High Control 15.8 ± 1.4 0 231 ± 40 NA 31  SE High V5 15.7 ± 1.4 2.1 ± 0.7 294 ± 68 80 ± 23 31  SE High V7 15.7 ± 1.5 4.9 ± 1.5 331 ± 54 95 ± 28 32  SE High V8 15.9 ± 1.4 5.3 ± 1.7 402 ± 58 127 ± 26 31  SE High V9-6x 15.5 ± 1.5 7.3 ± 2.1 425 ± 42 137 ± 25 31  SF Low Control 15.2 ± 1.7 0 178 ± 49 NA 31  SF Low V5 15.7 ± 1.5 2.1 ± 0.5 314 ± 74 101 ± 50 32  SF Low V7 15.8 ± 1.5 5.0 ± 1.6 374 ± 91 151 ± 24 31  SF Low V8 16.0 ± 1.5 5.3 ± 1.6 426 ± 82 156 ± 22 31  SF Low V9-6x 16.1 ± 1.3 8.3 ± 2.5 424 ± 70 163 ± 22 31 Fish size is in fork lengths (cm), tag burden (tag % of fish weight in air), and surgical effects are handling time (seconds) and surgery time (seconds). View Large Table 1. The mean and standard deviation (mean ± SD) for fish size and surgical effects and the sample size (n) for each treatment and surgeon combination in experiment one and two. Surgeon Surgeon experience Treatment Fork length Tag burden Handling time Surgery time n Experiment one  SA High Sham 15 ± 0.8 0 324 ± 423 125 ± 17 20  SA High V8 15.1 ± 0.7 6.9 ± 1.0 435 ± 115 130 ± 22 20  SA High V9-1x 14.9 ± 0.6 12.4 ± 1.4 503 ± 163 153 ± 27 20  SA High V9-6x 14.8 ± 0.9 10.6 ± 2.0 485 ± 204 146 ± 19 20  SB Low Sham 14.9 ± 0.7 0 554 ± 107 186 ± 57 20  SB Low V8 15.5 ± 0.8 6.5 ± 1.5 547 ± 73 183 ± 27 20  SB Low V9-1x 15.1 ± 1.0 12.8 ± 2.5 556 ± 77 203 ± 31 20  SB Low V9-6x 15.4 ± 0.8 9.6 ± 1.6 549 ± 75 194 ± 17 20  SC High Sham 15.2 ± 0.9 0 584 ± 90 205 ± 27 20  SC High V8 15.1 ± 1.0 6.9 ± 1.3 569 ± 52 215 ± 25 20  SC High V9-1x 14.9 ± 1.0 13.3 ± 2.3 661 ± 282 211 ± 30 20  SC High V9-6x 15.5 ± 0.9 9.4 ± 1.6 589 ± 135 212 ± 27 20  SD Low Sham 15.7 ± 1.0 0 510 ± 90 182 ± 32 20  SD Low V8 15.1 ± 0.8 7.3 ± 1.1 492 ± 81 191 ± 28 20  SD Low V9-1x 15.3 ± 1.0 12.3 ± 2.3 543 ± 98 212 ± 40 20  SD Low V9-6x 15.4 ± 1.0 9.9 ± 1.6 520 ± 68 209 ± 37 20 Experiment two  SA High Control 15.0 ± 1.7 0 167 ± 50 NA 31  SA High V5 15.3 ± 1.6 2.2 ± 0.7 292 ± 54 46 ± 9 31  SA High V7 15.7 ± 1.3 4.7 ± 1.3 353 ± 55 81 ± 12 31  SA High V8 16.0 ± 1.3 5.0 ± 1.4 363 ± 53 78 ± 13 32  SA High V9-6x 15.9 ± 1.2 7.2 ± 1.9 377 ± 52 92 ± 26 31  SB Low Control 15.5 ± 1.5 0 289 ± 75 NA 32  SB Low V5 16.2 ± 1.3 1.9 ± 0.5 290 ± 39 65 ± 21 31  SB Low V7 16.1 ± 1.4 4.4 ± 1.2 347 ± 49 79 ± 23 31  SB Low V8 15.7 ± 1.6 5.5 ± 1.8 374 ± 55 99 ± 14 31  SB Low V9-6x 15.8 ± 1.5 7.7 ± 2.4 383 ± 69 100 ± 21 32  SE High Control 15.8 ± 1.4 0 231 ± 40 NA 31  SE High V5 15.7 ± 1.4 2.1 ± 0.7 294 ± 68 80 ± 23 31  SE High V7 15.7 ± 1.5 4.9 ± 1.5 331 ± 54 95 ± 28 32  SE High V8 15.9 ± 1.4 5.3 ± 1.7 402 ± 58 127 ± 26 31  SE High V9-6x 15.5 ± 1.5 7.3 ± 2.1 425 ± 42 137 ± 25 31  SF Low Control 15.2 ± 1.7 0 178 ± 49 NA 31  SF Low V5 15.7 ± 1.5 2.1 ± 0.5 314 ± 74 101 ± 50 32  SF Low V7 15.8 ± 1.5 5.0 ± 1.6 374 ± 91 151 ± 24 31  SF Low V8 16.0 ± 1.5 5.3 ± 1.6 426 ± 82 156 ± 22 31  SF Low V9-6x 16.1 ± 1.3 8.3 ± 2.5 424 ± 70 163 ± 22 31 Surgeon Surgeon experience Treatment Fork length Tag burden Handling time Surgery time n Experiment one  SA High Sham 15 ± 0.8 0 324 ± 423 125 ± 17 20  SA High V8 15.1 ± 0.7 6.9 ± 1.0 435 ± 115 130 ± 22 20  SA High V9-1x 14.9 ± 0.6 12.4 ± 1.4 503 ± 163 153 ± 27 20  SA High V9-6x 14.8 ± 0.9 10.6 ± 2.0 485 ± 204 146 ± 19 20  SB Low Sham 14.9 ± 0.7 0 554 ± 107 186 ± 57 20  SB Low V8 15.5 ± 0.8 6.5 ± 1.5 547 ± 73 183 ± 27 20  SB Low V9-1x 15.1 ± 1.0 12.8 ± 2.5 556 ± 77 203 ± 31 20  SB Low V9-6x 15.4 ± 0.8 9.6 ± 1.6 549 ± 75 194 ± 17 20  SC High Sham 15.2 ± 0.9 0 584 ± 90 205 ± 27 20  SC High V8 15.1 ± 1.0 6.9 ± 1.3 569 ± 52 215 ± 25 20  SC High V9-1x 14.9 ± 1.0 13.3 ± 2.3 661 ± 282 211 ± 30 20  SC High V9-6x 15.5 ± 0.9 9.4 ± 1.6 589 ± 135 212 ± 27 20  SD Low Sham 15.7 ± 1.0 0 510 ± 90 182 ± 32 20  SD Low V8 15.1 ± 0.8 7.3 ± 1.1 492 ± 81 191 ± 28 20  SD Low V9-1x 15.3 ± 1.0 12.3 ± 2.3 543 ± 98 212 ± 40 20  SD Low V9-6x 15.4 ± 1.0 9.9 ± 1.6 520 ± 68 209 ± 37 20 Experiment two  SA High Control 15.0 ± 1.7 0 167 ± 50 NA 31  SA High V5 15.3 ± 1.6 2.2 ± 0.7 292 ± 54 46 ± 9 31  SA High V7 15.7 ± 1.3 4.7 ± 1.3 353 ± 55 81 ± 12 31  SA High V8 16.0 ± 1.3 5.0 ± 1.4 363 ± 53 78 ± 13 32  SA High V9-6x 15.9 ± 1.2 7.2 ± 1.9 377 ± 52 92 ± 26 31  SB Low Control 15.5 ± 1.5 0 289 ± 75 NA 32  SB Low V5 16.2 ± 1.3 1.9 ± 0.5 290 ± 39 65 ± 21 31  SB Low V7 16.1 ± 1.4 4.4 ± 1.2 347 ± 49 79 ± 23 31  SB Low V8 15.7 ± 1.6 5.5 ± 1.8 374 ± 55 99 ± 14 31  SB Low V9-6x 15.8 ± 1.5 7.7 ± 2.4 383 ± 69 100 ± 21 32  SE High Control 15.8 ± 1.4 0 231 ± 40 NA 31  SE High V5 15.7 ± 1.4 2.1 ± 0.7 294 ± 68 80 ± 23 31  SE High V7 15.7 ± 1.5 4.9 ± 1.5 331 ± 54 95 ± 28 32  SE High V8 15.9 ± 1.4 5.3 ± 1.7 402 ± 58 127 ± 26 31  SE High V9-6x 15.5 ± 1.5 7.3 ± 2.1 425 ± 42 137 ± 25 31  SF Low Control 15.2 ± 1.7 0 178 ± 49 NA 31  SF Low V5 15.7 ± 1.5 2.1 ± 0.5 314 ± 74 101 ± 50 32  SF Low V7 15.8 ± 1.5 5.0 ± 1.6 374 ± 91 151 ± 24 31  SF Low V8 16.0 ± 1.5 5.3 ± 1.6 426 ± 82 156 ± 22 31  SF Low V9-6x 16.1 ± 1.3 8.3 ± 2.5 424 ± 70 163 ± 22 31 Fish size is in fork lengths (cm), tag burden (tag % of fish weight in air), and surgical effects are handling time (seconds) and surgery time (seconds). View Large Table 2. Tag type specifications for experiment one and two. Tag type Diameter (mm) Length (mm) Weight in air (g) Edges/shape Experiment one  V8 8.00 20.50 2.00 Smooth/cylindrical  V9-1x 9.00 24.00 3.60 Smooth/cylindrical  V9-6x 9.00 21.00 2.90 Smooth/cylindrical Experiment two  V5 5.80 12.70 0.77 Rigid/cylindrical  V7 7.00 22.50 1.80 Smooth/cylindrical  V8 8.00 20.50 2.00 Smooth/cylindrical  V9-6x 9.00 21.00 2.90 Smooth/cylindrical Tag type Diameter (mm) Length (mm) Weight in air (g) Edges/shape Experiment one  V8 8.00 20.50 2.00 Smooth/cylindrical  V9-1x 9.00 24.00 3.60 Smooth/cylindrical  V9-6x 9.00 21.00 2.90 Smooth/cylindrical Experiment two  V5 5.80 12.70 0.77 Rigid/cylindrical  V7 7.00 22.50 1.80 Smooth/cylindrical  V8 8.00 20.50 2.00 Smooth/cylindrical  V9-6x 9.00 21.00 2.90 Smooth/cylindrical View Large Table 2. Tag type specifications for experiment one and two. Tag type Diameter (mm) Length (mm) Weight in air (g) Edges/shape Experiment one  V8 8.00 20.50 2.00 Smooth/cylindrical  V9-1x 9.00 24.00 3.60 Smooth/cylindrical  V9-6x 9.00 21.00 2.90 Smooth/cylindrical Experiment two  V5 5.80 12.70 0.77 Rigid/cylindrical  V7 7.00 22.50 1.80 Smooth/cylindrical  V8 8.00 20.50 2.00 Smooth/cylindrical  V9-6x 9.00 21.00 2.90 Smooth/cylindrical Tag type Diameter (mm) Length (mm) Weight in air (g) Edges/shape Experiment one  V8 8.00 20.50 2.00 Smooth/cylindrical  V9-1x 9.00 24.00 3.60 Smooth/cylindrical  V9-6x 9.00 21.00 2.90 Smooth/cylindrical Experiment two  V5 5.80 12.70 0.77 Rigid/cylindrical  V7 7.00 22.50 1.80 Smooth/cylindrical  V8 8.00 20.50 2.00 Smooth/cylindrical  V9-6x 9.00 21.00 2.90 Smooth/cylindrical View Large The experimental design was fully cross classified with respect to both tag type and surgeon so that each of four surgeons (SA, SB, SC, SD) performed 80 surgeries; 20 per treatment. The design was also cross classified with respect to the experience of the surgeon with two having over 15 years of experience (high experience: SA and SC) and two with <5 years of experience (low experience: SB and SD). Fish were kept off food for at least 24 h to evacuate the stomach prior to surgery to allow for more space within the cavity during tag implantation. All fish were taken off food the same day (1 June 2014), however, days-off-feed prior to surgery (DPS) ranged from 1 to 9 days and not all surgeons operated on each day (Table 3). However, weights and lengths of smolts did not differ between surgeons or DPS (Table 1). Table 3. Tagging schedule by tag and days-off-feed prior to surgery (DPS) in experiment one. Tag/DPS 1 2 3 4 5 6 7 8 9 V8 SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 – – SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – V9-1x SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 – – SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – V9-6x SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – Tag/DPS 1 2 3 4 5 6 7 8 9 V8 SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 – – SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – V9-1x SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 – – SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – V9-6x SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – Each day surgeons tagged equal numbers of each tag type (Surgeon; n). View Large Table 3. Tagging schedule by tag and days-off-feed prior to surgery (DPS) in experiment one. Tag/DPS 1 2 3 4 5 6 7 8 9 V8 SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 – – SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – V9-1x SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 – – SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – V9-6x SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – Tag/DPS 1 2 3 4 5 6 7 8 9 V8 SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 – – SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – V9-1x SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 – – SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – V9-6x SA; 10 SA; 10 SB; 10 SC; 10 SC; 5 SD; 10 SD; 10 – SB; 10 SC; 5 – – – – – – Each day surgeons tagged equal numbers of each tag type (Surgeon; n). View Large Experiment two Smolts (N = 625) were evenly allocated to five tagging treatments: (i) Control (anaesthetized but not subjected to surgery); (ii) V5; (iii) V7; (iv) V8; and (v) V9-6x (Tables 1 and 2). For treatment identification, fish were tagged with one of five Visual Implant Elastomer (VIE) colours (Northwest Marine Technology, Washington, USA). Individual fish were identified by their dummy tag IDs after tag expulsion or mortality. Fish were kept off food for 24 h to evacuate the stomach prior to surgery to allow for more space within the cavity during tag implantation. Tagging occurred on 10 and 11 May 2016. Four surgeons (SA, SB, SE, SF), two being the same as in experiment one (SA and SB), had either over 15 years of experience (high experience: SA and SE) or <5 (low experience: SB and SF). The experimental design was cross classified with respect to tag type, surgeon, and day of surgery so that each of four surgeons tagged approximately the same number of fish in each treatment on each of 2 days of surgery; 64 and 61 fish per treatment were tagged on day 1 and 2, respectively. Surgical procedure In both experiments, the surgeries were carried out using tools and equipment that were disinfected between each surgery by placing tools in anhydrous alcohol for 5 min then rinsed in distilled water. The entire surgical process was carried out on only one fish at a time. In experiment one, water temperature within tanks and during tagging was 8.5–9.5°C and fish were anaesthetized using MS222 (50 mg l−1) buffered with a sodium bicarbonate (100 mg l−1) solution in an aerated bath. In experiment two, water temperature within tanks and during tagging was 7.5–8.0°C and fish were anaesthetized using clove oil at a concentration of 40–120 mg l−1. Fish remained in the anaesthetic until loss of equilibrium and motion, and very little operculum movement was observed. Fish were weighed and measured for fork length then injected with a PIT tag (experiment one) or VIE tag (experiment two). Fish were placed ventral side up on a V-board and their gills were irrigated with water or anaesthetic throughout the entire surgical procedure. A 5–17 mm incision (depending on tag size) was made using a #12 curved blade (Fisherbrand, Fisher Scientific, Ontario, Canada) anterior to the pelvic girdle along the linea alba then a uniquely identifiable tag was inserted into the body cavity. Either one (incision ≤ 10 mm), two (incision 10 ≤ 15 mm), or three (incision > 15 mm) sutures were used to close the incision (Table 1). Sutures consisted of two surgeon knots followed by three overhand knots. Suture material was non-absorbable ethilon nylon monofilament (size: 4–0) applied with a FS-2 needle (Ethicon, Inc., Johnson & Johnson). Following surgery, a picture was taken of the sutured incision and the fish was placed in a holding tank until fully recovered then moved to a rearing tank. Handling time (the time from being placed in anaesthetic to being placed in recovery tanks after surgery), surgery time and the number of sutures were recorded for each fish (Table 1). Monitoring procedure Tanks were monitored daily for tag expulsions or mortalities. Fish that were found dead with an expelled tag were assumed to have expelled the tag prior to dying. Experiment one All fish were anaesthetized, weighed, measured, and assessed for incision healing (wound closure and inflammation) and tag protrusions every 2–4 weeks but after day 153 the time between sampling was longer. Sampling events occurred on day 27, 42, 56, 69, 83, 97, 109, 124, 153, 186, 217, 246, 274, 307, 335, and 406 after tagging. Following the final sampling event on day 406 after tagging, all fish were culled. Experiment two Unlike in experiment one, after the initial surgery the fish were not handled throughout the experiment, which lasted from 10 May to 16 August 2016 (97 days). The shorter duration, compared with experiment one, was the result of fish showing signs of a bacterial infection in the form of lesions affecting the skin and muscle tissue. The severity of infection was greatest in tank one and by day 97 had spread into tank two. Consequently, the experiment was terminated, and fish were culled. Statistical analyses Cox Proportional Hazard (CPH) models were used to determine the relative importance of factors affecting the duration of tag retention in each study. Analyses were supported by the R statistical software version 3.3.2 (R Core Team, 2016) framework and the survival library (Therneau, 2015). Results were considered significant at α < 0.05. The independent variables considered were surgery time, handling time, number of sutures (treated as a factor), tag weight, tag length, tag diameter, tag type, surgeon, surgeon experience (treated as a factor: high and low experience), fork length, weight, tag burden (the weight of the tag relative to the weight of the fish in air), DPS (only experiment one), and tank (only experiment two). Variables were assessed for collinearity using Pearson’s correlation factor (r) and the variance inflation factor (VIF). If two or more variables were highly correlated (r > 0.8 and VIF > 4), only one variable was selected for inclusion in the global model. Martingale residuals were plotted against the variables excluded from the model to determine if they could inform on model fit. The global model for both experiments included surgery time, handling time, number of sutures, tag type, surgeon, fork length, and tag burden whereas the full model for experiment two also included tank. Before model selection, variables were tested for violation of the proportional hazards assumption by plotting scaled Schoenfeld residuals against time and testing for a slope of zero using χ2 tests (Bellera et al., 2010). Covariates that were time dependent and violated the proportional hazards assumption were stratified so that hazard ratios were relative across all stratified levels. Model selection was performed using AIC scores (Akaike, 1987) with an exhaustive search of all possible models using R statistical package “glmulti” (Calcagno and De Mazancourt, 2010) and the model with the lowest AIC score was selected (see Supplementary Table S1). The final fitted model for experiment one included tag type and surgeon, with surgeon being stratified; the hazard ratio of surgeon SD was higher in the first 42 days after surgery compared with the days thereafter. The stratification of surgeon was because of surgeon being confounded with DPS and the extended length of DPS. Thus, time varying effects on the hazard ratio coefficients were introduced. For experiment two, the fitted model contained tag type and handling time. Martingale residuals were examined for model fit to the data. Tag retention curves were generated using the proportional hazard ratios from the fitted models. Mortality rate differences between surgeon and tag type were also assessed. Treatment comparisons were made using log-rank tests with a χ2 statistic to test for significant differences. If the overall χ2 test statistic was significant, then post-hoc pairwise comparisons, with Bonferroni adjusted p-values, were made between all treatments. In experiment two, comparisons between surgeons did not include the control treatment as smolts did not have a unique ID and could not be attributed to a surgeon. Also, in experiment two, because the control group only underwent anaesthesia, comparisons between the tag types and the control reflect the effect of both the surgery and tagging rather than just the effect of tag insertion. Results Experiment one Of the 240 smolts that were tagged, 16 tags were never recovered (lost in drainpipes) with 5, 4, and 7 lost from the V8, V9-1x, and V9-6x treatments, respectively. Because all missing tags were only noticed after a fish died, the dates of expulsion were unknown. Therefore, those fish were removed from the study. Tag retention Because V8 tags had the lowest risk of expulsion, V8 tags were used as a baseline comparison to other tagged treatment groups. Compared with V8 tags, the risk of expulsion of V9-1x and V9-6x tags was 1.94 and 2.44 times higher, respectively (Table 4). Although surgeon was a stratified factor within the CPH fitted model and coefficient estimates could not be obtained, tag type retention rates differed primarily between surgeons (Figure 1). Table 4. Results from fitted Cox proportional hazard models for experiment one and two. Term coef exp(coef) s.e.(coef) z Pr(>|z|) Experiment one  V9-1x 0.6638 1.9421 0.3057 2.172 0.0299*  V9-6x 0.8938 2.4444 0.2860 3.125 0.0018* Experiment two  V5 0.5822 1.7899 0.3296 1.766 0.0773  V7 0.2989 1.3484 0.3192 0.936 0.3490  V9-6x 0.8507 2.3412 0.2884 2.949 0.0032*  Handling time 0.0008 1.0008 0.0015 0.512 0.6083 Term coef exp(coef) s.e.(coef) z Pr(>|z|) Experiment one  V9-1x 0.6638 1.9421 0.3057 2.172 0.0299*  V9-6x 0.8938 2.4444 0.2860 3.125 0.0018* Experiment two  V5 0.5822 1.7899 0.3296 1.766 0.0773  V7 0.2989 1.3484 0.3192 0.936 0.3490  V9-6x 0.8507 2.3412 0.2884 2.949 0.0032*  Handling time 0.0008 1.0008 0.0015 0.512 0.6083 In both experiments, comparisons were made to a baseline of V8 for tag types. * denotes significant results at an alpha of 0.05. View Large Table 4. Results from fitted Cox proportional hazard models for experiment one and two. Term coef exp(coef) s.e.(coef) z Pr(>|z|) Experiment one  V9-1x 0.6638 1.9421 0.3057 2.172 0.0299*  V9-6x 0.8938 2.4444 0.2860 3.125 0.0018* Experiment two  V5 0.5822 1.7899 0.3296 1.766 0.0773  V7 0.2989 1.3484 0.3192 0.936 0.3490  V9-6x 0.8507 2.3412 0.2884 2.949 0.0032*  Handling time 0.0008 1.0008 0.0015 0.512 0.6083 Term coef exp(coef) s.e.(coef) z Pr(>|z|) Experiment one  V9-1x 0.6638 1.9421 0.3057 2.172 0.0299*  V9-6x 0.8938 2.4444 0.2860 3.125 0.0018* Experiment two  V5 0.5822 1.7899 0.3296 1.766 0.0773  V7 0.2989 1.3484 0.3192 0.936 0.3490  V9-6x 0.8507 2.3412 0.2884 2.949 0.0032*  Handling time 0.0008 1.0008 0.0015 0.512 0.6083 In both experiments, comparisons were made to a baseline of V8 for tag types. * denotes significant results at an alpha of 0.05. View Large Figure 1. View largeDownload slide Experiment one tag retention curves from CPH model for each tag type by each surgeon with high experience (SA and SC) and low experience (SB and SD) for the first 110 days. Shaded areas represent 95% CI. Vertical dashed lines represent the beginning and end of transition to salt water. Figure 1. View largeDownload slide Experiment one tag retention curves from CPH model for each tag type by each surgeon with high experience (SA and SC) and low experience (SB and SD) for the first 110 days. Shaded areas represent 95% CI. Vertical dashed lines represent the beginning and end of transition to salt water. CPH tag retention curves for tag types between surgeons showed that tag expulsion predominantly occurred over a 30-day period between day 35 and 65 after tagging, 21 days after smolts were fully transitioned to salt water. By day 50, tag retention rates for V8, V9-1x, and V9-6x varied from 67–87%, 45–77% ,and 37–71%, respectively, depending on surgeon (Figure 1). The last tag was expelled 97 days after tagging and the final retention rates ranged from 64–83.6%, 42–70.6% and 33.6–64.5% for V8, V9-1x and V9-6x tags, respectively. Across all surgeons, only V8 tags did not fall below 50% retention. Tag retention did not appear to be influenced by a surgeon’s experience because the least experienced surgeons (SB and SD) had the highest and lowest retention rates, respectively (Figure 1). Surgeon B tagged fish on DPS two and three whereas SD tagged on DPS eight and nine (Table 3), suggesting that DPS could be more important than experience of the tagger. Mortality Of the 304 fish (excluding the 16 fish where a tag was never recovered), 37 died (5, 6, 16, and 10 in the sham, V8, V9-1x, and V9-6x treatments, respectively). Mortality occurred throughout the first 337 days after surgery with 59.5% of all mortalities occurring within the first 10 days after tagging and before smolts were fully transitioned to salt water (Figure 2). The mortality rate for the entire 406-day experiment was 0.030% of smolts per day, however, in the first 97 days, the mortality rate was 0.112% per day. Of the 79 smolts that expelled tags, only four fish died after tag expulsion; two died shortly after expulsion (3–5 days), and two died 85 and 292 days after expulsion. Log-rank tests for survival curves showed mortality rates were significantly different between treatments (χ2 = 12.9; df = 3; p = 0.0048). Mortality was highest in V9-1x tags followed by V9-6x tags, V8 tags and the sham treatment (Figure 2). Post-hoc pairwise comparisons between treatments showed only a significant difference between the sham and V9-1x treatments (p = 0.01). Survival curves did not significantly differ between surgeons (χ2 = 7.7; df = 3; p = 0.053). Figure 2. View largeDownload slide Survival curves for each tag type treatment compared with the sham treatment for the first 110 days of experiment one. Shaded areas represent 95% CI. Vertical dashed lines represent the beginning and end of transition to salt water. Note that the sham treatment underwent surgery but did not receive a tag. Figure 2. View largeDownload slide Survival curves for each tag type treatment compared with the sham treatment for the first 110 days of experiment one. Shaded areas represent 95% CI. Vertical dashed lines represent the beginning and end of transition to salt water. Note that the sham treatment underwent surgery but did not receive a tag. Experiment two Tag retention One smolt was removed from the study because of missing handling time information. Because V8 tags had the lowest risk of expulsion, V8 tags were used as a baseline comparison to other tagged treatment groups. Compared with the V8 tags, the risk of expulsion associated with V5, V7, and V9-6x tags was 1.79, 1.35, and 2.34 times higher, respectively (Table 4). Tag type retention rates differed significantly between V9-6x and V7 tags, and V9-6x and V8 tags. Although handling time was included within the model, its effect on tag retention was not significant. Retention curves were predicted for tag type fixing handling time constant at the mean value of 360.9 s. The curves indicate that tag expulsion occurred predominantly over a 37-day period between 25 and 62 days after tagging and 17 days after the full transition to salt water (Figure 3). Fifty days after tagging, tag retention rates were 86, 89, 92, and 82% for V5, V7, V8, and V9-6x tags, respectively. Ninety-seven days after tagging, at the experiment’s conclusion, retention rates were 75, 80, 85, and 68% for V5, V7, V8, and V9-6x tags, respectively (Figure 3). Figure 3. View largeDownload slide Experiment two tag retention curves from CPH model for each tag type. Tag retention rates were estimated keeping handling time at a constant of 360.9 s. Shaded areas represent 95% CI. Vertical dashed lines represent the beginning and end of transition to salt water. Figure 3. View largeDownload slide Experiment two tag retention curves from CPH model for each tag type. Tag retention rates were estimated keeping handling time at a constant of 360.9 s. Shaded areas represent 95% CI. Vertical dashed lines represent the beginning and end of transition to salt water. Mortality Of the 625 fish in the experiment, 80 died (13, 16, 19, 14, and 18 died in the control, V5, V7, V8, and V9-6x tagging treatments, respectively). Throughout the 97-day experiment, the mortality rate was 0.132% smolts per day. Of the 500 tagged smolts, 108 smolts expelled tags with 14 of these smolts dying and 11 of these mortalities occurring on the same day as tag expulsion. Three fish had delayed mortality after tag expulsion and mortality occurred 1, 3, and 7 days after expulsion. Mortality occurred evenly throughout the experiment, beginning on day 3 and finishing on day 93 (Figure 4). Log-rank tests on survival curves showed mortality was not significantly affected by tag type (Figure 4; χ2 = 2.2; df = 4; p = 0.699) or surgeon (χ2 = 2.7; df = 3; p = 0.434). Of tagged smolts, 20 of the 67 mortalities were attributed to the bacterial infection. Figure 4. View largeDownload slide Survival curves for each tag type treatment compared with the control in experiment two. Shaded areas represent 95% CI. Vertical dashed lines represent the beginning and end of transition to salt water. Note that the control did not undergo surgery or receive a tag. Figure 4. View largeDownload slide Survival curves for each tag type treatment compared with the control in experiment two. Shaded areas represent 95% CI. Vertical dashed lines represent the beginning and end of transition to salt water. Note that the control did not undergo surgery or receive a tag. Discussion This is the first study to examine tag retention of Atlantic salmon smolts during their transition from fresh to salt water for an extended period of time (>365 days). Results indicated that the type of tag used was the most influential variable affecting tag retention, however, this may be more associated with tag shape than tag size. Furthermore, the high survival rates in our study suggest that previously recommended tag burden limits (<2%) were underestimated and smolts of 14.5–16.5 cm with tag burdens above 7.5% are more likely to expel a tag than to suffer mortality. Tag expulsion rates were not constant throughout the study and occurred predominantly throughout a 40-day period after smolts had fully transitioned to salt water. As the majority of these tags are generally active for up to 130 days (depending on ping rates and power settings), tag retention rates are likely to bias survival estimates. However, the amount of bias introduced will depend on the length of the study, the tag type used, and the life stage/life history of the study species. Tag retention Past studies on salmonids have found larger tags are retained less (Lacroix et al., 2004; Rechisky and Welch, 2009; Sandstrom et al., 2013) and higher tag burdens usually have lower retention rates (Brown et al., 2010; Smircich and Kelly, 2014). The larger tags (V9-1x and V9-6x) in experiment one, were expelled more often implying a tag size/tag burden effect, however, this relationship was less evident in experiment two (Figure 3). This is not surprising as tag burdens were much lower in experiment two (tag burden range = 1.9–8.3%; Table 1), which likely resulted in the higher retention rates compared with experiment one (tag burden range = 6.5–13.3%). This suggests that tag retention may only be negatively affected when exceeding a certain tag size/tag burden (>7.5%) and simply having a higher tag burden, does not result in higher expulsion rates. Lacroix et al. (2004) recommended that tag burdens not exceed 8% of the total Atlantic salmon smolt weight and these limits were only exceeded in experiment one by the V9-1x and V9-6x tags. In our study, except for the V5 tags in experiment two, tag retention rates were up to 24% lower for tags that exceeded 7.5% of smolt weight, which is similar to the results reported by Lacroix et al. (2004). The higher expulsion rate of V5 tags was surprising considering their relatively small size compared with other tags (Table 1). However, this may be related to the shape of the tag. Lacroix et al. (2004) noted that tag expulsion occurred through the body wall where the edges of the tag pressed against the muscle tissue. V5 tags have more rigid edges, compared with V7, V8, V9-1x, and V9-6x tags, which may have facilitated the expulsion of the tags through muscle tissue (Thorstad et al., 2013) and thereby explains the higher than expected rates of tag expulsion. The V5 tag has a smaller battery compared with the other tags assessed in this study and depending on tag settings (ping rate and power settings), tag expulsion may not be as important of a factor for short-term studies (<20 days). Tag retention in salmonids has been linked to many surgical, abiotic, and size related factors across many different studies (see Lacroix et al., 2004; Deters et al., 2010; Sandstrom et al., 2013) with retention rates varying from 100 (Ammann et al., 2013; Brown et al., 2013) to 0% (Lacroix et al., 2004), depending on tag type. In this study tag retention was not significantly influenced by surgical related factors like handling time, surgery time, or the number of sutures. This is not surprising, as surgery related effects were controlled so there was little variation between treatments, and thus, had little influence on results. Furthermore, evaluating the importance of these factors across studies is difficult given the lack of a standard approach when conducting intracoelomic tagging studies and speaks to the need to standardize protocols to improve surgical outcomes and safeguard fish welfare (Brown et al., 2010). However, it is important to note that in experiment one smolts were periodically re-handled after surgery, which may have increased stress and resulted in the higher degree of tag expulsion compared with experiment two. The variability between surgeons is an important factor to account for in tagging studies (Deters et al., 2012; Sandstrom et al., 2013) and appeared to be the only surgical related effect to have an influence on tag retention in our study. However, surgeon was confounded with DPS and violated the proportional hazards assumption, and therefore, hazard ratios could not be quantified. Overall, surgeons with low experience had the highest and lowest retention rates in experiment one, whereas all surgeon’s retention rates were similar in experiment two. These results infer that tag retention may be less influenced by surgeon experience and more dependent on surgeon aptitude (Deters et al., 2010). It is worth noting in experiment one that the surgeon with the lowest tag retention rate (SD) was the last to tag fish on DPS 8–9 after fish were off feed compared with DPS 1–5 for other surgeons. The length of DPS compared with the other surgeons may have increased stress levels for the last group of smolts and resulted in much higher expulsion rate. Tag retention has also been is influenced by water temperature, likely through an indirect influence on growth. Deters et al. (2010) found that fish held at higher temperatures (17°C compared with 12°C) had lower rates of tag retention. In Welch et al. (2007) and this study, tag expulsion predominantly occurred between day 21–84 and 25–65 after tagging, respectively, and at temperatures associated with spring growth between 8 and 13°C. These results are not consistent with Lacroix et al. (2004; tag burden range = 8.5–10.1%; fork length range = 14.5–14.8 cm) that found tag expulsion in Atlantic salmon began on day 139 after tagging and lasted until day 217. However, Lacroix et al. (2004) reported minimal growth within the first 4 months of their study with water temperatures never exceeding 4°C. The apparent influence of water temperature on growth and tag retention warrants further investigation as it would hold important implications for methodologies used in tagging studies. There are substantial morphological, physiological, and behavioural changes associated with spring growth, the transition from fresh to salt water, and the smolt life stage (Björnsson et al., 2011; Piironen et al., 2013). Alongside osmoregulatory changes, there is evidence that hormones promote growth, both physiologically and behaviourally, and cause the smolt and post-smolt to grow faster in length than weight, and thus, reduce body condition (Stefansson et al., 2008). Furthermore, physiological changes during smoltification can be dramatically reduced in landlocked populations and even reversed if anadromous smolts are held in fresh water (Stefansson et al., 2008; Piironen et al., 2013). Lacroix et al. (2004) held parr in fresh water for the experiment’s duration, which may help explain the differing retention rates compared with our experiment one which had similar tag burdens. As the majority of tag expulsion in this study occurred after smolts had fully transitioned to salt water (Figures 1 and 3), there is reason to believe that tag retention may also be influenced by life stage and life history and this highlights the importance of mimicking natural regimes as closely as possible when conducting in vitro retention studies. Mortality Studies have shown that the process of surgically implanting tags can negatively affect the survival, behaviour, performance, and growth of fish (Peake et al., 1997; Connors et al., 2002; Brown et al., 2006; Welch et al., 2007; Sandstrom et al., 2013; Cook et al., 2014). In nature, there are numerous types of mortality that can occur during tagging studies. Mortality can occur naturally, as a result of the tagging, or from a predation event. It is important to note that this study reflects the mortality that occurred in a lab setting using hatchery fish and results do not account for mortalities that would occur in nature because of predation, lack of foraging success, or disease. Another important consideration is the increased stress levels associated from repetitive fish handling (including times under anaesthesia), particularly in experiment one, that may have led to the higher rates of mortality. Therefore, our results should be treated with caution when applied to natural systems. In this study, there was little evidence to suggest that the survival differed between treatments; only in experiment one did tagged fish in the V9-1x treatment differ from the sham treatment. In experiment one, most of the mortality occurred prior to full transition to salt water (12 days after tagging; Figure 2), whereas in experiment two, mortality occurred more evenly throughout the entire experiment, likely because of the bacterial infection. However, the majority of non-bacterial related mortality occurred within the first 30 days post-surgery as in experiment one implying that the majority of mortality occurs relatively soon after tagging. This is not surprising as the surgery involved with tagging is very invasive (Bridger and Booth, 2003) and fish unable to recover will die soon after tagging because of stress and/or starvation. Therefore, during telemetry studies, mortalities related to surgery and tagging are likely to occur within the early portions of migration. Past studies have found up to 100% survival post-tag expulsion (Lacroix et al., 2004; Sandstrom et al., 2013), and results from our study show that mortality after tag expulsion was relatively low. In nature it is not possible to differentiate between tag loss and mortality because in both cases there are no future tag detections; consequently, high survival rates of fish after tag expulsion will introduce a negative bias into survival estimates when tag expulsion rates are high. In acoustic telemetry studies on fish species, there is a generally accepted “rule of thumb” that tags should not exceed 2% of the total weight of a tagged fish or tagging could negatively affect fish survival and performance (McCleave and Stred, 1975; Adams et al., 1998). However, more recently, the rule has been criticized for its validity (Jepsen et al., 2004) and numerous other studies have shown that there are exceptions to the 2% rule (Brown et al., 1999, 2006; Chittenden et al., 2009; Smircich and Kelly, 2014; Towne and Brandes, 2018). The survival of smolts from experiment one was only significantly different between the sham and V9-1x treatments with average tag burdens of 0 and 12.70%, respectively. Tag burdens in other treatments were still well above the 2% rule at 6.90% (V8) and 9.84% (V9-6x) but did not negatively influence survival compared with the sham treatment, emphasizing that the stresses of the tagging activity on the fish are a more important consideration than strictly observing the 2% rule. Furthermore, our results suggest that smolts are more likely to expel a tag than to suffer mortality from tagging and thus tag retention is a more important consideration. Conclusion An underlying assumption of telemetry studies on migrating fish species is that the individuals retain their tags throughout the duration of the study and are not subject to increased mortality and therefore, mark recapture models assume that when a tag is not detected, there are only two possible explanations: (i) the tag passed the receiver undetected or (ii) the fish died before it reached the receiver. The results from this study show that tag expulsion does occur and if not accounted for, will result in biased and underestimated survival rates. However, the amount of bias introduced will depend on the type of tag and duration of study. The effects of tag expulsion for short-term studies (∼25 days) will not be the same as studies over extended periods of time. Furthermore, in studies partitioning survival (i.e. Chaput et al., 2018), tag retention effects will not be equal through each partition as a function of time, growth/temperature/season of tagging, life stage, and life history. Therefore, accounting for the effects of tag retention on survival estimates should be suited to each telemetry study design and diligence should be taken to determine the best tag type and protocols for the hypotheses being tested. As acoustic telemetry is increasingly being used in survival analyses there is a growing need to quantify and reduce the biases that arise from surgically tagging fish. For anadromous salmonids, tagging normally occurs during the transition from fresh to salt water, a time of high stress and significant behavioural and physiological changes. In this study, up to 35% of smolts expelled tags with the majority of them surviving an additional 85 days or longer resulting in these smolts no longer having the ability to be detected by receivers and presumed dead. Thus, survival studies of similar duration that do not recognize these potential effects will have biased survival estimates. To minimize tag expulsion, and the potential bias in survival estimates, we recommend that tags being used do not exceed a burden of 7.5%. However, for shorter term studies, when battery life is not an issue, tags with smaller tag burdens may be more desirable and tag type should be chosen based on the specific requirements for the study. Many smolt telemetry studies use tag burdens between 1 and 4%, however, even when using small tags, retention rates should still be taken into account in order to provide accurate survival estimates for the effective management of fish populations. Acknowledgements We would like to thank Amirix/Vemco for providing the dummy acoustic transmitters and the Department of Fisheries and Oceans Canada (DFO) Mactaquac Biodiversity Facility for supplying the Atlantic salmon smolts. We would also like to thank Graham Chafe [Atlantic Salmon Federation (ASF)], Leroy Anderson (DFO), Jim Hawkes [National Oceanic and Atmospheric Administration (NOAA)], Michelle Charest, Steve Tinker, Dheeraj Busawon (DFO), Stephanie Ratelle (DFO), Ross Jones (DFO), Alex Dalton (DFO), Denieve Robinson, Graham Goulette (NOAA), Mike Best (ASF), John Strøm (Tromsø University), Heather Dixon (Wilfred Laurier University), Cynthia Hawthorne (DFO), and all other DFO, ASF, and NOAA staff that aided in the tagging, monitoring, and husbandry of the tagged Atlantic salmon. References Adams N. S. , Rondorf D. W. , Evans S. D. , Kelly J. E. , Perry R. W. 1998 . Effects of surgically and gastrically implanted radio transmitters on swimming performance and predator avoidance of juvenile Chinook salmon (Oncorhynchus tshawytscha) . Canadian Journal of Fisheries and Aquatic Sciences , 55 : 781 – 787 . Google Scholar Crossref Search ADS WorldCat Akaike H. 1987 . Factor analysis and AIC . Psychometrika , 52 : 317 – 332 . Google Scholar Crossref Search ADS WorldCat Ammann A. J. , Michel C. J. , MacFarlane R. B. 2013 . The effects of surgically implanted acoustic transmitters on laboratory growth, survival and tag retention in hatchery yearling Chinook salmon . Environmental Biology of Fishes , 96 : 135 – 143 . DOI 10.1007/s10641-011-9941-9 Google Scholar Crossref Search ADS WorldCat Bellera C. A. , MacGrogan G. , Debled M. , Tunon de Lara C. , Brouste V. , Mathoulin-Pelissier S. 2010 . Variables with time-varying effects and the Cox model: some statistical concepts illustrated with a prognostic factor study in breast cancer . BMC Medical Research Methodology , 10 : 20. Google Scholar Crossref Search ADS PubMed WorldCat Björnsson B. T. , Stefansson S. O. , McCormick S. D. 2011 . Environmental endocrinology of salmon smoltification . General and Comparative Endocrinology , 170 : 290 – 298 . Google Scholar Crossref Search ADS PubMed WorldCat Bridger C. J. , Booth R. K. 2003 . The effects of biotelemetry transmitter presence and attachment procedures on fish physiology and behaviour . Reviews in Fisheries Science , 11 : 13 – 34 . Google Scholar Crossref Search ADS WorldCat Brown R. S. , Cooke S. J. , Anderson W. G. , McKinley R. S. 1999 . Evidence to challenge the “2% rule” for biotelemetry . North American Journal of Fisheries Management , 19 : 867 – 871 . Google Scholar Crossref Search ADS WorldCat Brown R. S. , Deters K. A. , Cook K. V. , Eppard M. B. 2013 . A comparison of single-suture and double-suture incision closures in seaward-migrating juvenile Chinook salmon implanted with acoustic transmitters: implications for research in river basins containing hydropower structures . Animal Biotelemetry , 1 : 10. Google Scholar Crossref Search ADS WorldCat Brown R. S. , Geist D. R. , Deters K. A. , Grassell A. 2006 . Effects of surgically implanted acoustic transmitters >2% of body mass on the swimming performance, survival and growth of juvenile sockeye and Chinook salmon . Journal of Fish Biology , 69 : 1626 – 1638 . Google Scholar Crossref Search ADS WorldCat Brown R. S. , Harnish R. A. , Carter K. M. , Boyd J. W. , Deters K. A. , Eppard M. B. 2010 . An evaluation of the maximum tag burden for implantation of acoustic transmitters in juvenile Chinook salmon . North American Journal of Fisheries Management , 30 : 499 – 505 . Google Scholar Crossref Search ADS WorldCat Calcagno V. , De Mazancourt C. 2010 . glmulti: an R package for easy automated model selection with (generalized) linear models . Journal of Statistical Software , 34 : 1 – 29C . Google Scholar Crossref Search ADS WorldCat Carey J. B. , McCormick S. D. 1998 . Atlantic salmon smolts are more responsive to an acute handling and confinement stress than parr . Aquaculture , 168 : 237 – 253 . Google Scholar Crossref Search ADS WorldCat Chaput G. , Carr J. , Daniels J. , Tinker S. , Jonsen I. , Whoriskey F. 2018 . Atlantic salmon (Salmo salar) smolt and early post-smolt migration and survival inferred from multi-year and multi-stock acoustic telemetry studies in the Gulf of St. Lawrence, northwest Atlantic . ICES Journal of Marine Science , doi: 10.1093/icesjms/fsy156. WorldCat Chittenden C. M. , Butterworth K. G. , Cubitt K. F. , Jacobs M. C. , Ladouceur A. , Welch D. W. , Mckinley R. S. 2009 . Maximum tag to body size ratios for an endangered coho salmon (O. kisutch) stock based on physiology and performance . Environmental Biology of Fishes , 84 : 129 – 140 . Google Scholar Crossref Search ADS WorldCat Connors K. B. , Scruton D. , Brown J. A. , McKinley R. S. 2002 . The effects of surgically-implanted dummy radio transmitters on the behaviour of wild Atlantic salmon smolts . Hydrobiologia , 483 : 231 – 237 . Google Scholar Crossref Search ADS WorldCat Cook K. V. , Brown R. S. , Deng Z. D. , Klett R. S. , Li H. , Seaburg A. G. , Eppard M. B. 2014 . A comparison of implantation methods for large PIT tags or injectable acoustic transmitters in juvenile Chinook salmon . Fisheries Research , 154 : 213 – 223 . Google Scholar Crossref Search ADS WorldCat Cooke S. J. , Graeb B. D. S. , Suski C. D. , Ostrand K. G. 2003 . Effects of suture material on incision healing, growth and survival of juvenile Largemouth bass implanted with miniature radio transmitters: case study of a novice and experienced fish surgeon . Journal of Fish Biology , 62 : 1366 – 1380 . Google Scholar Crossref Search ADS WorldCat Cooke S. J. , Woodley C. M. , Eppard M. B. , Brown R. S. , Nielsen J. L. 2011 . Advancing the surgical implantation of electronic tags in fish: a gap analysis and research agenda based on a review of trends in intracoelomic tagging effects studies . Reviews in Fish Biology and Fisheries , 21 : 127 – 151 . Google Scholar Crossref Search ADS WorldCat Deters K. A. , Brown R. S. , Carter K. M. , Boyd J. W. , Eppard M. B. , Seaburg A. G. 2010 . Performance assessment of suture type, water temperature, and surgeon skill in juvenile Chinook salmon surgically implanted with acoustic transmitters . Transactions of the American Fisheries Society , 139 : 888 – 899 . Google Scholar Crossref Search ADS WorldCat Deters K. A. , Brown R. S. , Boyd J. W. , Eppard M. B. , Seaburg A. G. 2012 . Optimal suturing technique and number of sutures for implantation of acoustic transmitters in juvenile salmonids . Transactions of the American Fisheries Society , 141 : 1 – 10 . Google Scholar Crossref Search ADS WorldCat Dudgeon C. L. , Pollock K. H. , Braccini J. M. , Semmens J. M. , Barnett A. 2015 . Integrating acoustic telemetry into mark-recapture models to improve the precision of apparent survival and abundance estimates . Oecologia , 178 : 761 – 772 . Google Scholar Crossref Search ADS PubMed WorldCat Handeland S. O. , Järvi T. , Fernö A. , Stefansson S. O. 1996 . Osmotic stress, antipredator behaviour, and mortality of Atlantic salmon (Salmo salar) smolts . Canadian Journal of Fisheries and Aquatic Sciences , 53 : 2673 – 2680 . Google Scholar Crossref Search ADS WorldCat Jepsen N. , Flavio H. , Koed A. 2019 . The impact of cormorant predation on Atlantic salmon and sea trout smolt survival . Fisheries Management and Ecology , 26 : 183 – 186 . Google Scholar Crossref Search ADS WorldCat Jepsen N. , Koed A. , Thorstad E. B. , Baras E. 2002 . Surgical implantation of telemetry transmitters in fish: how much have we learned? Hydrobiologia , 483 : 239 – 248 . Google Scholar Crossref Search ADS WorldCat Jepsen N. , Schreck C. , Clements S. , Thorstad E. B. , 2004 . A brief discussion on the 2% tag/bodyweight rule of thumb. In Aquatic Telemetry: Advances and Applications , pp. 255 – 260 . Ed. by Spedicato M. T. , Lembo G. , Marmulla G. . Food and Agriculture Organization of the United Nations , Rome . Google Preview WorldCat Lacroix G. L. , Knox D. , McCurdy P. 2004 . Effects of implanted dummy acoustic transmitters on juvenile Atlantic salmon . Transactions of the American Fisheries Society , 133 : 211 – 220 . Google Scholar Crossref Search ADS WorldCat McCleave J. D. , Stred K. A. 1975 . Effect of dummy telemetry transmitters on stamina of Atlantic salmon (Salmo salar) smolts . Journal of the Fisheries Research Board of Canada , 32 : 559 – 563 . Google Scholar Crossref Search ADS WorldCat McCormick S. D. , Hansen L. P. , Quinn T. P. , Saunders R. L. 1998 . Movement, migration, and smolting of Atlantic salmon (Salmo salar) . Canadian Journal of Fisheries and Aquatic Sciences , 55 ( Suppl. 1 ): 77 – 92 . Google Scholar Crossref Search ADS WorldCat McCormick S. D. , Moriyama S. , Björnsson B. T. 2000 . Low temperature limits photoperiod control of smolting in Atlantic salmon through endocrine mechanisms . American Journal of Physiology-Regulatory, Integrative and Comparative Physiology , 278 : R1352 – R1361 . Google Scholar Crossref Search ADS PubMed WorldCat Peake S. , McKinley R. S. , Scruton D. A. , Moccia R. 1997 . Influence of transmitter attachment procedures on swimming performance of wild and hatchery reared Atlantic salmon smolts . Transaction of the American Fisheries Society , 126 : 707 – 714 . Google Scholar Crossref Search ADS WorldCat Peven G. , Giorgi A. , Skalski J. , Langeslay M. , Grassell A. , Smith S. , Counihan T. et al. 2005 . Guidelines and Suggested Protocols for Conducting, Analyzing, and Reporting Juvenile Salmonid Survival Studies in the Columbia River Basin . University of Washington School of Aquatic & Fishery Sciences , Seattle, Washington . Google Preview WorldCat Piironen J. , Kiiskinen P. , Huuskonen H. , Heikura-Ovaskainen M. , Vornanen M. 2013 . Comparison of smoltification in Atlantic salmon (Salmo salar) from anadromous and landlocked populations under common garden conditions . Annales Zoologici Fennici , 50 : 1 – 15 . Google Scholar Crossref Search ADS WorldCat R Core Team. 2016 . R: A Language and Environment for Statistical Computing [online]. http://www.R-project.org/ (last accessed 6 January 2017). Rechisky E. L. , Welch D. W. 2009 . Surgical implantation of acoustic tags: influence of tag loss and tag—induced mortality on free—ranging and hatchery—held spring Chinook (O. tschawytscha) smolts. In Tagging Telemetry and Marking Measures for Monitoring Fish Populations. A Compendium of New and Recent Science for Use in Informing Technique and Decision Modalities , pp. 69 – 94 . Ed. by Wolf K. S. , O’ Neal J. S. . The Pacific Northwest Aquatic Monitoring Partnership and KWA Ecological Sciences , Duvall, Washington . Google Preview WorldCat Sandstrom P. T. , Ammann A. J. , Michel C. , Singer G. , Chapman E. D. , Lindley S. , MacFarlane R. B. , Klimley A. P. 2013 . Growth, survival, and tag retention of steelhead trout (Oncorhynchus mykiss) and its application to survival estimates . Environmental Biology of Fishes , 96 : 145 – 164 . Google Scholar Crossref Search ADS WorldCat Smircich M. G. , Kelly J. 2014 . Extending the 2% rule: the effects of heavy internal tags on stress physiology, swimming performance, and growth in brook trout . Animal Biotelemetry , 2 : 16. Google Scholar Crossref Search ADS WorldCat Stefansson S. O. , Björnsson B. Th. , Ebbesson L. O. E. , McCormick S. D. 2008 . Smoltification. In Fish Larval Physiology , pp. 639 – 681 . Ed. by Finn R. N. , Kapoor B. G. . Science Publishers , Enfield, New Hampshire . Google Preview WorldCat Therneau T. 2015 . A Package for Survival Analysis in S. Version 2.38 [online]. https://CRAN.R-project.org/package=survival (last accessed 17 June 2018). Thiem J. D. , Taylor M. K. , McConnachie S. H. , Binder T. R. , Cooke S. J. 2011 . Trends in reporting of tagging procedures for fish telemetry studies that have used surgical implantation of transmitters: a call for more complete reporting . Reviews in Fish Biology and Fisheries , 21 : 117 – 126 . Google Scholar Crossref Search ADS WorldCat Thorstad E. B. , Rikardsen A. H. , Alp A. , Økland F. 2013 . The use of electronic tags in fish research—an overview of fish telemetry methods . Turkish Journal of Fisheries and Aquatic Sciences , 13 : 881 – 896 . WorldCat Towne K. , Brandes P. L. 2018 . Survival, tag retention, growth, and wound healing of juvenile Chinook salmon (Oncorhynchus tshawytscha) surgically implanted with a dummy acoustic tag . San Francisco Estuary and Watershed Science , 16 . WorldCat Welch D. W. , Batten S. D. , Ward B. R. 2007 . Growth, survival, and tag retention of steelhead trout (O. mykiss) surgically implanted with dummy acoustic tags . Hydrobiologia , 582 : 289 – 299 . Google Scholar Crossref Search ADS WorldCat © International Council for the Exploration of the Sea 2019. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Tag retention and survival of Atlantic salmon (Salmo salar) smolts surgically implanted with dummy acoustic transmitters during the transition from fresh to salt water
JF - ICES Journal of Marine Science
DO - 10.1093/icesjms/fsz139
DA - 2010-08-01
UR - https://www.deepdyve.com/lp/oxford-university-press/tag-retention-and-survival-of-atlantic-salmon-salmo-salar-smolts-0xD01U0r9H
SP - 1
VL - Advance Article
IS - 
DP - DeepDyve
ER -