Individual Variability in the Movement Behaviour of Juvenile Atlantic Salmon

1Individual variability in the movement behaviour of juvenile Atlantic salmon2Mathieu L. Roy1, André G. Roy1, James W. A. Grant2, Normand Bergeron331Département de géographie, Université de Montréal, 520 Côte Ste-Catherine, Montréal, 4QC, H2B 2V8, Canada.52Department of Biology, Concordia University, 7141 Sherbrooke St. West, Montréal, QC,6H4B 1R6, Canada.73 INRS-Eau, Terre et Environnement, 490 rue de la Couronne, Québec, QC, G1K 9A9, 8Canada910111213141516171819202122231 1 2 24Abstract25Stream-dwelling salmonid populations are generally thought to be composed of both 26relatively mobile and sedentary individuals, but this conclusion is primarily based on 27results obtained from recapture methods with low temporal resolution. In this study, the 28mobility of 50 juvenile Atlantic salmon was monitored using a large array of passive 29integrated transponder antennas buried in the bed of a natural stream. Fish locations were 30recorded at a high frequency for a period of three months in a 65 m reach. Four types of 31daily behaviour were identified: stationary (detected primarily at one location), sedentary 32(limited movement between a few locations), floater (frequent movements in a restricted 33home range) and wanderer (movements across the reach). Most individuals exhibited low34mobility on most days, but also showed occasional bouts of high mobility. Between-35individual variability accounted for only 12-17% of the variability in the mobility data. 36High mobility was more frequent at low flow, but no difference was observed between 37the summer (12-18oC) and the autumn (4-12oC). Individual variation on a daily basis 38suggested that movement behaviour is a response to changing environmental conditions 39rather than an individual behavioural trait. 4041Keywords: Juvenile salmon, Mobility, Individual variability, Behaviour, Habitat. 42434445463 2 4 47Introduction48 Early studies depicted juvenile salmonids as sedentary, territorial animals 49exhibiting high site fidelity (Kalleberg 1958; Keenleyside 1962). The development of 50better tagging technology, which allowed for the tracking of individuals at a higher 51temporal resolution, revealed that the territorial mosaic of salmon parr was more flexible 52and dynamic than previously thought (Armstrong et al. 1999; Okland et al. 2004). In 53particular, Atlantic salmon parr have large, overlapping home ranges, with some 54individuals moving out of their home ranges to relocate either upstream or downstream 55(Okland et al. 2004; Ovidio et al. 2007), suggesting little fidelity to a particular 56microhabitat, sometimes undertaking habitat switches of several kilometres to lakes 57(Hutchings, 1986; Erkinaro et al. 1998) or estuaries (Cunjak et al. 1986).58 It now seems broadly accepted that both sedentary and mobile individuals occur 59within a given population (Gowan et al. 1994; Rodriguez 2002; Morrissey and Ferguson 602011). While the size of the two fractions varies considerably between sites, species, and 61life stages, the sedentary fraction tends to be larger than the mobile fraction (e.g. 62Hesthagen 1988; Steingrimsson and Grant 2003). Some individuals have been 63characterized as “movers” (i.e. cruise foragers) whereas others as “stayers” (sit-and-wait 64foragers), based on the proportion of time spent moving (Grant and Noakes 1987; 65McLaughlin et al. 1994). Although spatial behaviour might be a heritable trait (Ferguson 66and Noakes 1983), Gowan et al. (1994) suggested that individuals may switch tactics in 67response to changing environmental conditions. Although some juvenile salmonids can 68defend the same territory for extended periods (Martel 1996), a fraction can switch 69between sedentary and mobile behaviour between two subsequent years (Harcup et al. 5 3 6 701984). However, it remains unclear how common this behaviour is, and at what temporal71frequency the switching occurs. 72 Fish movements have been linked to changes in biotic and abiotic conditions 73(Gowan et al. 1994), likely due to variation in flow stage, temperature and daily light 74cycles. However, the effects of these variables on the behaviour of fish seem to be 75complex, as several studies have provided contrasting results. For instance, salmonids 76have been reported to decrease their mobility and territory size at high flows (Kemp et al. 772006), whereas others report the opposite trend (Scruton et al. 2003; Riley et al. 2009), or78no trend at all (Berland et al. 2004; Heggenes et al. 2007). Similarly, while water 79temperature affects fish metabolism (Jonsson et al. 2001), its effect on fish activity is less 80certain (Fraser et al. 1993; Breau et al. 2007). As temperature drops in the autumn, 81Atlantic salmon parr suppress their daytime activity, presumably as a result of a tradeoff 82between growth and predation risk (Fraser et al. 1995; Johnston et al. 2004). 83Nevertheless, other proximate factors must also influence fish activity on a seasonal basis84(Bremset 2000), as other studies report either no decrease or an increase in fish mobility 85(Nykanen et al. 2004; Riley et al. 2006).86 The spatial arrangement of microhabitats might also influence mobility because 87habitat heterogeneity decreases territory size (Venter et al. 2008) and mobility (Heggenes 88et al. 2007). In less heterogeneous habitats, individuals might have to move farther to 89encounter complementary microhabitats that provide foraging opportunities and shelter 90(Venter et al. 2008). However, information on the relationship between habitat structure 91and fish mobility remains fragmentary. Furthermore, juvenile salmonids are often 92captured using methods that might be better suited for catching sedentary than mobile 7 4 8 93fish (Gowan and Fausch 1996). Therefore, if mobility affects habitat use, the estimation 94of habitat preference might be biased towards sedentary fish.95 The results from movement studies depend on how frequently fish have been 96located and for what duration (Lucas and Baras 2000). For instance, fish mobility 97estimates from radio-telemetry studies are generally greater than those obtained from 98mark-recapture studies. However, radio-telemetry suffers from the inability to sample 99small fish and from relatively large sampling units of habitat, making difficult the 100quantification of small-scale movements. Recent developments in flat-bed passive 101integrated transponder (PIT) antenna grid provide fish tracking data at both high temporal102and spatial resolutions over extended periods of time (Greenberg and Giller 2000; Riley 103et al. 2003). In this study, we used a large PIT antenna grid to monitor daily movement of104a group of individually marked Atlantic salmon parr 1+ in a natural stream. Positions of 105tagged fish were recorded continually during three months of the summer and autumn. 106While previous studies have reported a high between-individual variation in parr mobility107(Okland et al. 2004; Ovidio et al. 2007), within-individual mobility variation has received108little attention. Hence, our primary objective was to document the magnitude of 109individual variation of parr daily mobility to test the competing predictions that 110individuals: (1) adopt consistent mobile or sedentary tactics over long periods of time; or 111(2) modify their mobility based on changing biotic and abiotic conditions. Second, if the 112data support the second prediction, we tested whether changes in behaviour could be 113predicted by environmental fluctuations. In particular, we tested the predictions that parr 114will be more sedentary: (3) when flow stage increases, as both the availability of drifting 115prey and swimming energy costs increase; (4) in the autumn than in the summer; and, (5) 9 5 10 116in heterogeneous habitats, which likely provide complementary feeding and sheltering 117habitats in closer proximity.118119Material and methods120Study site121This study was carried out on the Xavier Brook, a tributary of the Ste-Marguerite River in122Saguenay, Québec, Canada (48°2591799 N; 69°5394899 W). The study reach, located 123425 m from the main river confluence, was approximately 65 x 10 m (length x width), 124composed of two pools separated by a steep riffle, providing high physical habitat 125diversity. In the thalweg at low stage (0.4 m3·s-1), depth ranged from approximately 0.1 m126in the riffle to 1.65 m in the upstream pool. Median substrate size (B-axis, i.e. particle 127width) varied from gravel-cobble in the riffle to gravel-cobble in the deep portion of the 128pools and gravel-sand in the pool recirculation zones (substrate classification according to129Wolman 1954).130131Fish tracking system132To monitor fish movements, we used a large flatbed antenna grid covering the entire 133study reach. The system was used to monitor tagged fish locations in the reach during 97 134days (24 July to 1 November 2008). The tracking system consisted of an array of 149 135circular antennas (50 cm in diameter), which were buried within the river bed and 136designed to detect the presence of 23mm PIT tags (Texas Instruments (TIRIS) model RI-137TRP-RRHP, 134 2 kHz) and other tags complying with the ISO 11784/11785 138international standards. Each antenna was interrogated for fish presence every 34 s (i.e. 11 6 12 1390.03 Hz). Antennas were distributed along cross-channel transects each composed of five 140antennas. Overall, the detection field of the antenna grid covered 19% of the wetted area 141of the site at a discharge of 0.4 m3· s-1.142 Each group of five antennas was linked to a tuning capacitor, which was wired to 143a CYTEK multiplexer (JX/256 series, mercury wetted 256 single poles relay, www.cytec-144ate.com). The multiplexer was connected to an Aquartis controller (custom made by 145Technologie Aquartis; www.aquartis.ca) composed of a TIRIS S-2000 reader, a 146datalogger and a custom-made controller unit. The system was powered by three solar 147panels connected to four 6V batteries plugged in series and two 12V batteries plugged in 148parallel. Each antenna was activated successively for the detection of PIT tag presence. 149When a PIT tagged fish was detected, the date (dd/mm/yy), time (hh/mm/ss), antenna ID 150(multiplexer card and port number) and fish ID (tag number) were recorded. Detection 151range varied from 300-400 mm above the bed surface and 600-800 mm in diameter. 152During the study period, all antennas detected at least one fish. For more technical details 153on the antenna grid, see Johnston et al. (2009).154155Fish capture and tagging156 We captured 69 Atlantic salmon parr (1+) in the study reach on two occasions 157using a backpack electrofishing device: 44 fish were caught on 24 July 2008, and 25 on 15828 August 2008. During the second electrofishing session, fish were captured 159immediately upstream of the reach to avoid re-capturing tagged fish. Only parr of body 160length >80 mm were kept for the experiment to avoid the potential negative effects of PIT161tagging on parr survival and growth (Sigourney et al. 2005); smaller juveniles as well as 13 7 14 162juvenile brook trout (Salvelinus fontinalis) and longnose dace (Rhinichthys cataractae) 163were released at individual capture locations. Fish were then anesthetised in a clove oil 164solution (3 ml/10 L) and implanted with 23-mm PIT tags (Texas Instruments) in the 165abdominal cavity secured with surgical tissue adhesive (Vetbond©). Tagged fish were 166allowed a recovery period of approximately 2 hours in a fish tank before being released 167on the study site. A total of three fish died during tagging, two during the first tagging 168session and one during the second. Average fork length (L ± SD) and average mass (M± 169SD) of tagged fish were: LA: 98 ± 7.4 mm; MA= 9.7 ± 1.7 g; LB: 109 ± 8.3 mm; MB: 10.7 170± 2.3 g. The fish captured in August were larger than those captured one month earlier 171(fork length t = -5.73, df =64, p < 0.001 (mass t = -1.92, df = 64, p=0.06).172173Habitat characterization174 Flow stage and water temperature fluctuations were recorded every 15 min using 175a pressure transducer (Level logger) installed at the bottom of the upstream pool. Water 176stage was estimated by correcting the recorded pressure values for changes in 177atmospheric data obtained from the closest meteorological station, and then subtracting 178the minimum value observed during the study period. Therefore, stage was defined as the179water level above the minimum summer low flow level. The study period was 180characterized by substantial discharge variability (Fig. 1). A high magnitude flow event 181occurred at the beginning of August, followed by a stage decrease in the following 182month. Then, a prolonged low flow lasted until the end of October when it was 183interrupted by several precipitation events. Base flow between these events was 184approximately 10 cm over the minimum flow, which corresponded roughly to the median15 8 16 185flow recorded during the study period. Flow stage values were categorized as low (0-10 186cm, 35% of days), medium (10-15 cm, 35% of days), high (15-25 cm, 13% of days) and 187very high (25 cm and higher, 17% of days). Using a field based digital elevation model of188the reach, bankful flow was estimated to occur at a stage of 60 cm. 189 During the same period, water temperature decreased from 19.0oC to 2.8oC (Fig. 1901). From 24 July to 1 September, daily average water temperature fluctuated around 19115oC. After 1 September, water temperature decreased linearly. Water temperature 192reached 12oC on 9 September, which corresponds to the upper boundary of the 193temperature range at which parr suppress their daytime activity (Valdimarsson et al. 1941997). The study period was therefore divided into two periods: summer (12-18 oC) and 195autumn (4-12 oC). 196 Depth and bed roughness were also characterized in detail throughout the reach. 197Topography was surveyed using a robotic total station (Trimble 5600DR) by combining a198systematic transect sampling approximately 1 m apart with the characterization of 199individual roughness elements that protruded approximately 10 cm above the local mean 200bed elevation. This strategy was adopted to optimize sampling effort, as sampling point 201density increased proportionally with bed complexity. From the total of 6250 sampled 202points, a digital elevation model was created using a triangular irregular network 203interpolation with pixel size of 10 cm. Topography was detrended for mean thalweg slope204and water surface at median flow was subtracted to obtain flow depth. Therefore, 205variability of flow depth mainly reflected height variation induced by the riffle-pool 206channel morphology. Depth was not temporally adjusted to flow stage to reflect the use 207of specific habitats rather than specific depth values. This way, across flow stages, high 17 9 18 208depth use could be interpreted as the use of habitat located in a pool rather than be 209confused with habitats located in the riffle at a higher flow stage.210 Bed roughness, expressed as the spatial standard deviation of bed elevation values211of the DEM pixels included in a moving window of 65 cm2, was characterized by 212computing an index based on the estimate of local bed elevation variability. The size of 213the window was determined in order to characterize the roughness of most of the largest 214particles present on the reach. We focused on protuberance from the bed that might be 215more important in creating flow refuges and cover than average particle size. For 216instance, it is common to observe large particles buried in the bed that do not protrude 217higher above the average bed height than smaller particles (Nikora et al. 1998). The 218downstream pool exhibited the largest coherent region of high bed roughness, whereas 219the remainder of the reach showed an apparently random spatial pattern of bed roughness.220For every antenna, mean depth at median flow and bed roughness were estimated by 221averaging all pixel values located in a circle matching the antenna detection range. Fish 222daily habitat use was then estimated by averaging the mean values associated with all 223visited locations weighted by the number of detections per antenna.224225Data analysis226 Fish behaviour was characterized on a daily basis using four variables. The 227number of movements and the distance travelled provided estimates of fish mobility, 228whereas the number of sites visited and the extent of the reach used by fish gave estimates229of home range size. We defined a fish movement as a change of fish location (antenna): 230i.e. every time a fish was detected at two different locations successively, a movement 19 10 20 231was recorded. The number of movements was an indicator of activity that did not take 232into account the distance between locations. In contrast, the daily distance travelled was 233defined as the sum of the distance (m) between each antenna successively visited. The 234number of sites represented the number of different antenna locations where a fish was 235detected in a day. However, despite a high spatial coverage and a high temporal sampling236frequency of the tracking system, fish could sometimes travel between two distant 237locations without being detected by antennas located in between. Hence, the variable 238extent fills this gap by representing a home range length, or the distance between the two 239most distant locations visited by a fish in a day.240 We used a principal component analysis (PCA) on the daily mobility variables to 241describe the variability of every fish. Prior to the PCA, each variable was normalized 242(log10+1) and standardized. Then, based on the mobility variables, fish behaviour was 243classified using a k-means clustering algorithm. The correct number of behavioural types 244(clusters) was determined by comparing silhouette values between three and five 245behavioural types (Kaufman and Rousseeuw, 1990) 246 The frequency of occurrence of behavioural types per individual that spent six 247days or more in the study reach was examined. Then, the components of variance of the 248four mobility variables were estimated using an additive-variance component model, 249using individuals as a random factor, in which yij (mobility of fish I on day j) = µ+αi+εij where µ is 250the mean of the population, i, deviation from the mean of the ith fish (i=1 to 24) εij, is the251residuals containing the intra-individual variation. To meet the model assumptions, 252transformed variables were used (log10+1). However, the descriptive statistics shown in 253the figures and tables are based on non transformed data. 21 11 22 254To examine the temporal variability of fish behaviour, the proportion of behavioural 255types adopted by each individual was plotted on a time series. Then, the frequency of 256occurrence of behavioural types was examined in relation to flow-stage categories and 257season. A generalized estimation equation (GEE) approach was used to describe the 258observed and expected occurrence of a behavioural type as a function of flow stage and 259season. GEEs are an extension of generalized linear models that accommodate repeated 260measurement of the same individuals and a categorical response variable (Diggle et al. 2612002). Therefore, the variable days was used as a repeated measure, fish as subjects, flow 262stage and temperature as fixed factors and behavioural types as a dependent categorical 263variable. GEEs were performed by SPSS 17 © (SPSS Inc. Chicago, Illinois) using a 264Poisson distribution with a log link and repeated measurement covariance structure set to 265first order autoregressive to account for temporal dependence between successive days. 266Similarly, to examine differences in habitat use in terms of depth and bed roughness in 267relation to behavioural types, two distinct mixed-effects models were performed using 268fish as subjects, days as repeated measures, behavioural types as a fixed factor and depth 269and roughness as a dependant variable. Again, first order autoregressive covariance of the270repeated measurements was chosen. Best suited models were selected based on lowest 271Akaike’s information criterion (AIC) and quasi AIC (QIC) values (Burnham et al. 2011). 272For both types of statistical models, the effect of fish mass and length was tested. No 273significant effect was observed (p > 0.05), perhaps because of a low statistical power 274resulting from the need to treat the fish from the two tagging periods separately and of the275overall low variability of values observed among fish. Therefore, mass and length were 276not incorporated in the models.23 12 24 277278Results279Fish tracking280 Of the 66 fish that were PIT-tagged and released in the reach, 4 individuals (6%) 281were never detected by the tracking system and 12 individuals (18%) were detected for 282either less than 24 hours following release or less than three hours in a single day. These 283individuals were not included in further analyses. Of the remaining fish, 10 individuals 284(15%) were detected in the reach during a single day, 16 individuals (24%) were detected285for 1 to 5 days and 24 individuals (37%) remained between 6 and 70 days in the reach.286287Behavioural types288 The daily distance travelled, the number of movements, the number of sites visited289and the extent of the reach used by fish were positively correlated, which allowed for data290reduction. Indeed, 90% of the variability was explained by the two first axes of a PCA 291(Table 1). The primary ordination axis (PCA1), which accounted for 70% of the 292variability, was positively correlated with all mobility variables, but was least strongly 293correlated with extent (Fig. 2). Therefore, low values of PCA1 represented lower mobility294in smaller home ranges, whereas higher values represented higher mobility in larger 295home ranges. In contrast, the secondary axis explained 20% of the variability and was 296negatively correlated to the number of movements and positively correlated to the extent 297of fish movement. Data ordination illustrated the high variability of overall mobility 298exhibited by fish during the entire study period (Fig.2.).25 13 26 299 Based on a cluster analysis, fish spatial behaviour was categorized into four types:300Stationary, Sedentary, Floater and Wanderer (Fig. 2). Stationary behaviour was 301characterized by low mobility, being detected most often by a single antenna (Table 2). 302Fish 15 on Day 26 (Aug 18) adopted typical stationary behaviour (Fig. 3). On some 303occasions, stationary behaviour also included the use of more than one location during 304the day. However, these locations were adjacent to each other and no back and forth 305movements were observed.306 When fish were detected at a few locations in a day, their behaviour was 307characterized as sedentary (Table 2). Fish exhibiting sedentary behaviour travelled on 308average 10m daily and moved three to four times between locations for an average extent 309of 5.7 m. For example, on Day 61 (22 Sep), Fish 50 exhibited typical sedentary behaviour310by using four locations located throughout half the channel length and moved only once 311between each location (Fig. 3). 312 Cluster analysis also discriminated two types of higher mobility behaviour (high 313PCA1 scores) along the extent-number of movement gradient (PCA2 axis) (Fig. 2). When 314individuals used a relatively restricted home range (average extent: 5.7m), but made 315many movements between locations (mean =36), their behaviour was defined as floater 316(Table 2). For instance, Fish 37 on Day 32 (24 Aug) was detected at only five nearby 317locations in the downstream pool, but switched 34 times between these locations (Fig. 3). 318During the study period, the most extreme floater made 525 movements, resulting in a 319daily travelled distance of 2449 m on an extent of 5.3 m.320 In contrast, when a fish exhibited a high distance travelled (avg: 115 m), but over 321a larger extent (avg: 25.3 m), their behaviour was defined as wanderer (Table 2). Typical 27 14 28 322wanderer behaviour involved travelling across the entire reach, from one pool to the 323other. Wanderer behaviour was characterized by a similar number of sites visited as the 324floater. However, the number of movements between locations was generally lower and 325the locations visited were farther away. The number of sites visited by fish adopting 326wandering behaviour was not higher than for floaters, likely because individuals moving 327long distances travelled rapidly and were thus difficult to detect. For example, on Day 61 328(22 Sep), Fish 66 travelled from the downstream pool almost to the upstream pool, then 329back again, but was detected at only 11 sites for a total of 16 movements (Fig. 3).330331Individual variability in behaviour332 Individuals exhibited a variety of types of mobility behaviour rather than 333‘specializing’ on one type over the study period. Among the 26 individuals that were 334detected on the site for less than six days, their behaviour was sedentary, stationary and 335wanderer on average for 36%, 30% and 30% of their time, respectively (Fig. 4a). Floater 336behaviour was only observed in five fish, which represented on average 8% of their time 337(Fig. 4a).338 Among the individuals that stayed more than six days on the study site, high intra-339individual variability of behaviour was observed (Fig. 4b). Out of 24 individuals, 15 340showed all four types of behaviour during the study period. Low-mobility behaviour was 341most frequently observed, as individuals were sedentary and stationary for 33% and 28% 342of the days, respectively, during which they were detected in the reach. Floater and 343wanderer behaviour were less frequent, with an average of 19% and 20% of the days, 344respectively. However, six fish were more mobile than the others, exhibiting floater or 29 15 30 345wanderer behaviour for more than 50% of the days. For fish that stayed more than six 346days, there was no significant trend between the duration in the reach and the proportion 347of days each behaviour type was adopted by an individual (p > 0.1).348 All fish were sedentary most of the time, but many individuals exhibited 349occasional bouts of high mobility. Indeed, all fish that stayed more than 6 days in the 350reach showed a low median distance travelled, but most moved more than 90 m. The 351daily number of movements displayed a similar pattern, with a relatively low median 352number of movements and numerous extreme values. Although the number of sites and 353the extent did not show as many extreme values, there was high intra-individual 354variability. Decomposing the components of variation of the four mobility variables 355indicated that intra-individual variation accounted for 83 to 88% of the total variation, 356compared to 12 to 17% for the inter-individual variation (Table 3). Similarly, average 357ordination scores for individuals that stayed more than six days suggested that a 358considerable number of individuals had relatively similar average mobility (Fig. 2). For 359instance, 17 individuals (70%) had their average ordination scores categorized as 360sedentary while the remaining fish was categorized as floater or wanderer. 361362Temporal variability363 The frequency of behaviour exhibited over the course of the season suggested that on 364most days, a mixture of behavioural types was observed (Fig. 5). Following the tagging 365of 42 parr on 24 July, 14 individuals were present on the reach. The number of 366individuals dropped drastically on 4 Aug. following a major flood event, then fluctuated 367between 4 and 8 before the second tagging session on 28 Aug., after which the number 31 16 32 368peaked at 22 and then constantly decreased until the end of the observation period. 369Despite the variability of behaviour observed on a daily basis, some periods were 370dominated by specific behaviour types. For instance, between 14 Aug. and 28 Aug., 371wandering behaviour was observed only four times, whereas most fish exhibited 372wandering behaviour from Day 16 Oct. to 20 Oct. Similarly, from day 22 Aug. to 29 373Aug., floater behaviour was most frequent.374 Examining the frequencies of occurrence of behavioural types in relation to flow 375stage and season using GEE showed a general decrease in mobility with increasing flow 376stage (Wald χ2=7.974, df=3, p=0.047). Pooled frequencies of occurrence illustrated an 377increase in the proportions of sedentary behaviour from 30% to 45% with an increase in 378flow stage (Fig. 6). Conversely, wandering behaviour decreased from 22% of occurrence 379to 7% from low flow to a very high flow. Over the season, parr used slightly different 380habitats in terms of depth when adopting different behavioural types (F = 5.46, df = 3813,514, p = 0.001). Daily average depth used was 0.2 m higher for the fish exhibiting 382floater behaviour than the other behavioural types (Fig. 7a). However, this trend was 383observed for only five individuals (Fig. 7b). In fact, a mixed effect model accounting for 384individual and temporal dependence indicated that wanderer behaviour was associated 385with lower depths used than the three other behavioural types (confidence interval on 386depth difference, df = 222-249, p < 0.03 for all comparisons, Bonferroni adjustments). In387contrast, no difference in bed roughness used was observed for the different behaviour 388types (F = 1.235, df = 3, 549, p = 0.296).389390Discussion33 17 34 391 In this study, most individuals exhibited low mobility (stationary and sedentary 392behaviour) on most days, but most individuals also showed occasional bouts of high 393mobility, either by carrying out frequent movements in a relatively restricted area 394(floater) or by travelling across the reach, from one pool to the other (wanderer). Our 395results suggest that most fish in the reach switched behaviour on a daily basis.396 Daily behaviour switching contrasts with the common view that fish population 397are composed of fractions of sedentary and mobile individuals (Rodriguez 2002, 398Morrissey and Ferguson 2011). However, comparisons of behavioural categories across 399study designs and spatial scale is complicated. Telemetry and recapture techniques can 400provide a very large spatial scale, which is useful to characterize habitat switches 401sometimes involving migrations of several kilometres to lacustrine (Hutchings 1986, 402Ryan 1986) or estuarine environments (Cunjak et al. 1989). Our relatively restricted 403study area (65 m reach) most likely underestimated the mobility of fish exhibiting 404wandering behaviour. Nevertheless, reach extent likely had a minor effect on the majority405of fish, which adopted sedentary behaviour most of the time and therefore should not 406affect our conclusion about intra-individual variability in behaviour. Hence, our proposed407classification of mobility behaviour is dependent on the spatial scale of the study design, 408which focused on the fine scale movements of parr in a single reach. 409 Most studies confirming the presence of mobile and sedentary fractions of a 410population have used recapture techniques with relatively low temporal sampling 411frequency (e.g. Heggenes et al. 1991; Roghair 2005). Such techniques require sampling 412over a long duration to obtain individual variability without underestimating fish 413mobility. For instance, if an individual moves 40 meters upstream over a short period of 35 18 36 414time and then back to its original location, the following recapture could lead to the 415biased conclusion of sedentary behaviour Rather than being strictly sedentary or mobile, 416individual brown trout switched behaviour over the course of a two-year study (Harcup et417al. 1984). Our data support a similar flexibility of the mobile or sedentary fractions, but 418over shorter time scales.419 Generally, there was less variation in behaviour when environmental conditions 420were more homogeneous. Despite a baseline amount of variation, some types of 421behaviour tended to dominate during particular periods of environmental stasis (i.e. more 422sedentary at high flow, more floating at low flow). Furthermore, the most drastic 423environmental fluctuations might have triggered changes in behaviour. For instance, the 424steady declines in water temperature from 16-20 October were accompanied by 425wandering behaviour by most fish. Perhaps, this was the trigger that winter or spawning 426is imminent, leading to a search for overwintering habitat by females and spawning 427opportunities for males. Similarly, during the major floods the few fish that were detected428were sedentary rather than mobile, and the most sedentary fish were probably even not 429detected.430 Salmon parr in our study exhibited a decrease in mobility with an increase in flow431stage. Similar results have been observed in previous studies (Kemp et al. 2006), whereas432others found no effect of flow stage on mobility (Robertson et al. 2004; Heggenes et al. 4332007). Because salmon parr are well adapted to using flow refuges to maintain station on 434the bed, most of the increased swimming costs that accompany higher flows (Hill and 435Grossman 1993) are likely to be associated with foraging movements or longer range 436movements from one foraging location to another (Liao 2007). Moreover, when 37 19 38 437velocities are higher, parr tend to reduce their foraging territory size in response to the 438increased swimming costs and prey density (Hughes and Dill 1990; Piccolo et al. 2008).439 We found no difference in movement behaviour between the summer and autumn,440despite the decrease in temperature. A decrease in mobility was expected due to a 441decrease in metabolic rate (Jonsson et al. 2001) and the expected decrease of diurnal 442activity (Fraser et al. 1993). However, parr can remain active even when water 443temperature is close to zero (Bremset 2000). Indeed, a radio-telemetry study showed that 444parr home ranges were as large during the autumn as during the summer (Okland et al. 4452004). Furthermore, a recent study undertaken under similar temperature ranges reported 446a higher mobility of parr in the winter (6.6-10.8 oC) than during the autumn (10.7-14.3oC)447(Riley et al. 2006). The authors suggested that this behaviour might be unique to 448groundwater fed systems. Taken together with previous studies, our results suggest that 449mobility can remain relatively high even when water temperatures are low. In this study, 450the effect of lower metabolism on movement might have been offset by several factors 451including a change from sit and-wait drift foraging to benthic cruise foraging due to a 452decrease in drift abundance (Nislow et al. 1998). Interestingly, all individuals adopted 453wandering behaviour on two days in mid-October close to spawning season when 454temperature was between 4 and 6oC. The presence of spawning adults passing through the455site may have increased the mobility of tagged fish, particularly the precocious parr.456 We hypothesized that individuals would be more sedentary in shallow and 457heterogeneous habitats because high habitat heterogeneity is more likely to provide 458complementary feeding and sheltering habitats close together (Johnston et al. 2010) and 459because territory size tends to decrease with habitat heterogeneity (Kemp et al. 2005; 39 20 40 460Dolinsek et al. 2007; Venter et al. 2008). Our results did not support this prediction. 461However, for five individuals that remained over forty days in the reach, deeper habitats 462were associated with floater behaviour. Finding mobile fish in pools is in agreement with 463the assumption that foraging fish occupy a larger territory in lower velocity areas 464(Hughes and Dill 1990). Furthermore, although the term floater tends to refer to 465individuals deprived of a territory, such behaviour could be associated with multiple 466central place foraging, where fish frequently switch from one foraging territory to an 467adjacent one (Steingrimsson and Grant 2008). Nevertheless, when fish adopted wanderer 468behaviour, they used slightly shallower habitats than when adopting other behavioural 469types. These results contrast with our hypothesis and with previous observations on adult 470grayling (Thymallus thymallus (Nykanen et al. 2004). The presence of mobile fish in 471deeper areas might have implications for the accuracy of abundance surveys. As the 472catchability of juvenile salmonids decrease with mobility (Crozier and Kennedy, 1994), 473electrofishing might underestimate abundance in pools, which could be mistakenly 474considered as low quality or unused habitats (Linnansaari et al. 2010).475 One possible mechanism explaining our observation of mostly sedentary 476behaviour interrupted by frequent bouts of high mobility might be the fish’s need to 477‘sample’ their spatially and temporally variable environment to evaluate drift abundance. 478Being aware of the quality of alternative habitats, an assumption of the ideal free 479distribution, is critical to determine which habitats or strategy will generate the greatest 480energy intake and growth (Power 1984), within the constraints of dominance hierarchies 481and predation risk.41 21 42 482 In summary, the present study showed that Atlantic salmon parr are sedentary on 483most days, but also exhibit infrequent bouts of higher mobility. Movement behaviour 484appears to be plastic, allowing individuals to adapt to changing environmental conditions 485(Gowan et al. 1994). Even though movement behaviour was linked to variation in flow 486stage, a high between- and among intra-individual variation was observed, suggesting 487individuals undertake movements in reaction to other proximate factors operating at 488shorter time scales. Several studies on fish behaviour, movement and habitat use have 489reported a high inter-individual variation (Okland et al. 2004; Ovidio et al. 2007). 490However, studies are often conducted for a shorter duration and at a lower temporal 491frequency than our study. Furthermore, intra-individual variability is often overlooked by 492averaging values to estimate home ranges over the entire study period. Therefore, 493differences in individual behaviour may decrease as study duration increases. Because 494Atlantic salmon parr exhibit relatively high mobility, maintaining connectivity between 495different habitats (i.e. pools and riffles) should be considered a priority in salmon 496conservation practices.497498Acknowledgements499 We would like to thank Francis Bérubé and Marc-André Pouliot for their help in 500the development of the tracking system and for essential logistical and technical 501assistance, and Jordan Rosenfeld for interesting discussions. We are also grateful to 502Patricia Johnston, André Boivin, Claude Gibeault, Marie-Êve Roy and René Roy for their503help in the field and Geoide, NSERC and the Canada Research Chair Program for the 43 22 44 504funding of this research. Three anonymous reviewers provided useful comments that 505substantially improved the paper.506507References508Armstrong, J.D., Huntingford, F.A., and Herbert, N.A. 1999. Individual space use 509 strategies of wild juvenile Atlantic salmon. J. Fish Biol. 55: 1201-1212.510Berland, G., Nickelsen, T., Heggenes, J., Okland, F., Thorstad, E.B., and Halleraker, J. 511 2004. Movements of wild Atlantic salmon parr in relation to peaking flows below 512 a hydropower station. River Res. Appl. 20: 957-966.513Bovee, K.D. 1982. A guide to stream habitat analysis using the instream flow incremental514 methodology. Fish Wildl. Serv. FWS/OBS-82/26. 515Breau, C., Weir, L.K., and Grant, J.W.A. 2007. Individual variability in activity patterns 516 of juvenile Atlantic salmon (Salmo salar) in Catamaran Brook, New Brunswick. 517 Can. J. Fish.Aquat. Sci. 64: 486-494.518Bremset, G. 2000. Seasonal and diel changes in behaviour, microhabitat use and 519 preferences by young pool-dwelling Atlantic salmon, Salmo salar, and brown 520 trout, Salmo trutta. Environ. Biol. Fish. 59: 163-179.521Burnham K.P., Anderson, D.R., and Huyvaert, K.P. 2011. AICc model selection in the 522 ecological and behavioral sciences: some background, observations and 523 comparisons. Behav Ecol Sociobiol. 65: 23-35.524Crozier, W.W. and Kennedy, G.J.A. 1994. Application of semi-quantitative electrofishing525 to juvenile salmonid stock surveys. J. Fish Biol., 45: 159–164.45 23 46 526Cunjak, R.A. , Chadwick , E.M.P. , & Shears , M. 1989. Downstream Movements and 527 Estuarine Residence by Atlantic Salmon Parr (Salmo Salar). Can. J. of 528 Fish.Aquat. Sci. 46: 1466 – 1471.529Erkinaro, J., Niemelä, E., Saari, A., Shustov, Yu., and Jørgensen, L. 1998. Timing of 530 habitat shift by Atlantic salmon parr from fluvial to lacustrine habitat: analysis of 531 age distribution, growth and scale characteristics. Can. J. of Fish.Aquat. Sci. 55: 532 2266-2273.533Diggle, P.J., Heagerty, P., Liang, K.-Y., and Zeger, S.L. 2002. Longitudinal data analysis.534 Oxford University Press, U.K.535Dolinsek, I.J., Grant, J.W.A., and Biron, P.M. 2007. The effect of habitat heterogeneity 536 on the population density of juvenile Atlantic salmon salmo salar l. J. Fish Biol. 537 70: 206-214.538Ferguson, M.M., and Noakes, D.L.G. 1983. Movers and stayers - genetic-analysis of 539 mobility and positioning in hybrids of lake charr, Salvelinus namaycush, and 540 brook charr, S. fontinalis (Pisces, Salmonidae). Behav. Genet. 13: 213-222.541Fraser, N.H.C., Heggenes, J., Metcalfe, N.B., and Thorpe, J.E. 1995. Low summer 542 temperatures cause juvenile Atlantic salmon to become nocturnal. Can. J. Zoolog. 543 73: 446-451.544Fraser, N.H.C., Metcalfe, N.B., and Thorpe, J.E. 1993. Temperature-dependent switch 545 between diurnal and nocturnal foraging in salmon. P. Roy. Soc. B-Biol. Sci.252: 546 135-139.47 24 48 547Gowan, C., and K. D. Fausch. 1996. Mobile brook trout in two high-elevation Colorado 548 streams: re-evaluating the concept of restricted movement. Can. J. Fish. Aquat. 549 Sci. 53:1370-1381.550Gowan, C., Young, M.K., Fausch, K.D., and Riley, S.C. 1994. Restricted movement in 551 resident stream salmonids - a paradigm lost. Can. J. Fish. Aquat. Sci. 51: 2626-552 2637.553Grant, J.W.A., and Noakes, D.L.G. 1987. Movers and stayers: foraging tactics of young-554 of-the-year brook charr, Salvelinus fontinalis. J. Anim. Ecol. 56: 1001-1013.555Greenberg, L.A., and Giller, P.S. 2000. The potential of flat-bed passive integrated 556 transponder antennae for studying habitat use by stream fishes. Ecol. Freshw. Fish557 9: 74-80.558Harcup, M.F., Williams, R., and Ellis, D.M. 1984. Movements of brown trout, Salmo 559 trutta L., in the River Gwyddon, South Wales. J. Fish Biol. 24: 415-426.560Heggenes, J., Northcote, T.G., and Peter, A. 1991. Spatial stability of cutthroat trout 561 (Oncorhynchus clarki) in a small, coastal stream. Can. J. Fish. Aquat. Sci. 48: 562 757-762.563Heggenes, J., Omholt, P.K., Kristiansen, J.R., Sageie, J., Okland, F., Dokk, J.G., and 564 Beere, M.C. 2007. Movements by wild brown trout in a boreal river: Response to 565 habitat and flow contrasts. Fisheries Manag. Ecol. 14: 333-342.566Hesthagen, T. 1988. Movements of brown trout, Salmo trutta, and juvenile Atlantic 567 salmon, Salmo salar, in a coastal stream in northern Norway. J. Fish Biol. 32: 568 639-653.49 25 50 569Hill, J., and Grossman, G.D. 1993. An energetic model of microhabitat use for rainbow-570 trout and rosyside dace. Ecology. 74: 685-698.571Hughes, N.F., and Dill, L.M. 1990. Position choice by drift-feeding salmonids: Model 572 and test for Arctic grayling (Thymallus arcticus) in subarctic mountain streams, 573 interior Alaska. Can. J. Fish. Aquat. Sci. 47: 2039-2048.574Hutchings, J.A. 1986. Lakeward migrations by juvenile Atlantic salmon, Salmo salar, 575 Can. J. Fish. Aquat. Sci. 43: 732-741.576Johnston P., and Bergeron N. 2010. Variation of juvenile Atlantic salmon (Salmo salar) 577 body composition along sedimentary links. Ecol. Freshw. Fish. 19: 187-196.578Johnston, P., Bergeron, N.E., and Dodson, J.J. 2004. Diel activity patterns of juvenile 579 Atlantic salmon in rivers with summer water temperature near the temperature-580 dependent suppression of diurnal activity. J. Fish Biol. 65: 1305-1318.581Johnston, P., Bérubé, F., and Bergeron, N.E. 2009. Development of a flatbed passive 582 integrated transponder antenna grid for continuous monitoring of fishes in natural 583 stream. J. Fish Biol. 74: 1651–1661.584Jonsson, B., Forseth, T., Jensen, A.J., and Naesje, T.F. 2001. Thermal performance of 585 juvenile Atlantic salmon, Salmo salar L. Funct. Ecol. 15: 701-711.586Kalleberg, H. 1958. Observations in a stream tank of territoriality and competition in 587 juvenile salmon and trout (Salmo salar L. and S. trutta L. ). Inst. Freshw. Res. 588 Drottningholm Rep. 39: 55-98. 589Kaufman L., and Rousseeuw, P.J. 1990. Finding Groups in Data: An Introduction to 590 Cluster Analysis. John Wiley & Sons, Inc. Hoboken, NJ. USA.Keenleyside, 591 M.H.A. 1962. Skin diving observations of Atlantic salmon and brook trout51 26 52 592 in the Miramichi River, New Brunswick. J. Fish. Res. Board Can. 19: 625-634.593Kemp, P.S., Armstrong, J.D., and Gilvear, D.J. 2005. Behavioural responses of juvenile 594 Atlantic salmon (Salmo salar) to presence of boulders. River Res. Appl. 21: 1053-595 1060.596Kemp, P.S., Gilvear, D.J., and Armstrong, J.D. 2006. Variation in performance reveals 597 discharge-related energy costs for foraging Atlantic salmon (Salmo salar) parr. 598 Ecol. Freshw. Fish 15: 565-571.599Liao, J.C. 2007. A review of fish swimming mechanics and behaviour in altered flows. 600 Philos. T. Roy. Soc. B. 362: 1973-1993.601Linnansaari, T., Keskinen, A., Romakkaniemi, A., Erkinaro, J., and Orell, P. 2010. Deep 602 habitats are important for juvenile Atlantic salmon Ssalmo salar L. In large rivers.603 Ecol. of Freshw. Fish 19: 618-626.604Lucas, M.C., and Baras, E. 2000. Methods for studying spatial behaviour of freshwater 605 fishes in the natural environment. Fish Fish. 1: 283-316.606Martel, G. 1996. Growth rate and influence of predation territoriality in juvenile coho 607 salmon (Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci. 53: 660-669.608McLaughlin, R.L., Grant, J.W.A., and Kramer, D.L. 1994. Foraging movements in 609 relation to morphology, water-column use, and diet for recently emerged brook 610 trout (Salvelinus fontinalis) in still-water pools. Can. J. Fish. Aquat. Sci 51: 268-611 279.612Morrissey, M.B., and Ferguson, M.M. 2011. Individual variation in movement 613 throughout the life cycle of a stream-dwelling salmonid fish. Mol. Ecol. 20: 235-614 248.53 27 54 615Nikora, V.I., Goring, D.G., and Biggs, B.J.F. 1998. On gravel-bed roughness 616 characterization. Water Resour. Res. 34: 517-527.617Nislow, K.H., Folt, C., and Seandel, M. 1998. Food and foraging behavior in relation to 618 microhabitat use and survival of age-0 Atlantic salmon. Can. J. Fish. Aquat. Sci. 619 55: 116-127.620Nykanen, M., Huusko, A., and Lahti, M. 2004. Changes in movement, range and habitat 621 preferences of adult grayling from late summer to early winter. J. Fish Biol. 64: 622 1386-1398.623Okland, F., Thorstad, E.B., and Naesje, T.F. 2004. Is Atlantic salmon production limited 624 by number of territories? J. Fish Biol. 65: 1047-1055.625Ovidio, M., Enders, E.C., Hallot, E.J., Roy, M.L., Philippart, J.C., Petit, F., and Roy, 626 A.G. 2007. Mobility and home-range use of Atlantic salmon parr over short time 627 scales. Aquat. Living Resour. 20: 95-101.628Piccolo, J.J., Hughes, N.F., and Bryant, M.D. 2008. Water velocity influences prey 629 detection and capture by drift-feeding juvenile coho salmon (Oncorhynchus 630 kisutch) and steelhead (Oncorhynchus mykiss irideus). Can. J. Fish. Aquat. Sci. 631 65: 266-275.632Power, M.E. 1984. Habitat quality and the distribution of algae-grazing catfish in a 633 Panamanian stream. J. Anim. Ecol. 53: 357-374. 634Riley, W.D., Eagle, M.O., Ives, M.J., Rycroft, P., and Wilkinson, A. 2003. A portable 635 passive integrated transponder multi-point decoder system for monitoring habitat 636 use and behaviour of freshwater fish in small streams. Fisheries Manag. Ecol. 10: 637 265-268.55 28 56 638Riley, W.D., Ives, M.J., Pawson, M.G., and Maxwell, D.L. 2006. Seasonal variation in 639 habitat use by salmon, Salmo salar, trout, Salmo trutta and grayling, Thymallus 640 thymallus, in a chalk stream. Fisheries Manag. Ecol. 13: 221-236.641Riley, W.D., Maxwell, D.L., Pawson, M.G., and Ives, M.J. 2009. The effects of low 642 summer flow on wild salmon (Salmo salar), trout (Salmo trutta) and grayling 643 (Thymallus thymallus) in a small stream. Freshwater Biol. 54: 2581-2599.644Robertson, M.J., Pennell, C.J., Scruton, D.A., Robertson, G.J., and Brown, J.A. 2004. 645 Effect of increased flow on the behaviour of Atlantic salmon parr in winter. J. 646 Fish Biol. 65: 1070-1079.647Rodriguez, M.A. 2002. Restricted movement in stream fish: The paradigm is incomplete, 648 not lost. Ecology 83: 1-13.649Roghair, C.N. 2005. Brook trout movement during and after recolonization of a naturally 650 defaunated stream reach. N. Am. J. Fish. Manage. 25: 777-784.651Ryan, P.M. 1986. Lake Use by Wild Anadromous Atlantic Salmon, Salmo salar, as an 652 Index of Subsequent Adult Abundance. Can. J. Aquat. Sci. 43: 2-11.653Scruton, D.A., Ollerhead, L.M.N., Clarke, K.D., Pennell, C., Alfredsen, K., Harby, A., 654 and Kelly, D. 2003. The behavioural response of juvenile Atlantic salmon (Salmo 655 salar) and brook trout (Salvelinus fontinalis) to experimental hydropeaking on a 656 Newfoundland (Canada) river. River Res. Appl. 19: 577-587.657Sigourney, D.B., Horton, G.E., Dubreuil, T.L., Varaday, A.M., and Letcher, B.H. 2005. 658 Electroshocking and PIT Tagging of Juvenile Atlantic Salmon: Are There 659 Interactive Effects on Growth and Survival? N. Am. J. Fish. Manage. 25: 1016-660 1021.57 29 58 661Steingrimsson, S.O., and Grant, J.W.A. 2003. Patterns and correlates of movement and 662 site fidelity in individually tagged young-of-the-year Atlantic salmon (Salmo 663 salar). Can. J. Fish. Aquat. Sci 60: 193-202.664Steingrimsson, S.O., and Grant, J.W.A. 2008. Multiple central-place territories in wild 665 young-of-the-year Atlantic salmon Salmo salar. J. Anim. Ecol. 77: 448-457.666Valdimarsson, S.K., Metcalfe, N.B., Thorpe, J.E., and Huntingford, F.A. 1997. Seasonal 667 changes in sheltering: effect of light and temperature on diel activity in juvenile 668 salmon. Anim. Behav. 54: 1405-1412.669Venter, O., Grant, J.W.A., Noel, M.V., and Kim, J.-W. 2008. Mechanisms underlying the670 increase in young-of-the-year Atlantic salmon (Salmo salar) density with habitat 671 complexity. Can. J. Fish. Aquat. Sci 65: 1956-1964.672Wolman, M.G. 1954. A method of sampling coarse river-bed material. Trans. Am. 673 Geophys. Union. 35: 951-956.67467559 30 60 676 677678Table 1. Pearson correlation coefficients of mobility variables versus axis scores from an 679ordination of daily fish spatial behaviour in the study reach and proportion of total 680variance expressed by the two first ordination axes (n=681).PCA1 PCA2 Distance travelled 0.56 - 0.52 Number of movements 0.51 - 0.14 Number of sites 0.52 0.04 Extent 0.38 0.83 Proportion of variance 0.70 0.20 681682683684685686687688689690691692693694Table 2. Frequency of occurrence (n) and mean (range) of the four mobility variables for 695each behavioural type pooled for all individuals. Stationary Sedentary Floater Wanderer N 161 213 134 111 Distance traveled (m) 2.2(0-12) 10.6(0.7-41) 70(7.6-2454) 115(15-2249) Number of movements 1.0(0-4) 3.4(1-11) 36(4-558) 10.9(3-154)61 31 62 Number of sites 1.8(1-4) 3.3(2-6) 5.7(2-12) 5.8(3-17) Extent (m) 2.5(0-6) 7.23(1.2-35) 5.7(1.3-15) 28.2(16-43) 696697698699700701702703704705706707708709710Table 3. Geometric mean, total sum of squares, and within- and between-individual 711variation in four mobility variables and principle component 1 for 24 juvenile Atlantic 712salmon parr monitored for 6-97 days (619 observations). 713714Mean Total Intra Inter(%) (%) PCA1 -0.1 1655.7 87 13 Distance travelled 0.93 167.7 88 12 Number of movements 0.68 92.5 87 13 Number of sites 0.69 18.9 85 1563 32 64 Extent (m) 0.76 57.6 83 17 715716717718719720721722723724725726727728Figure captions729Fig. 1. Water temperature (upper curve) and stage (lower curve) recorded from 24 July to 73030 October 2008. The vertical dashed line divides the study into summer and autumn 731periods based on a threshold of 12oC. The horizontal dashed line shows the flow stage 732matching bankfull discharge. *indicates the two fish tagging sessions. 733734Fig. 2. Principal component analysis (PCA) on 50 Atlantic salmon parr daily mobility 735variables (N=681) during 97 days. Each dot represents the mobility of an individual on a 736particular day. Open circles show individual average values for the 24 fish that remained 737in the reach for six days or more. Polygons delineate behavioural types (stationary, 738sedentary, floater and wanderer) discriminated by a cluster analysis (K-means) on the 65 33 66 739daily mobility data: Number of sites, Distance travelled (m), Number of movements and 740Extent (m).741742Fig.3. Typical daily mobility corresponding to four behavioural types. Examples were 743selected based on the closest average PCA1 and PCA2 scores for each type: 1- (star) 744Stationary: 0 movement (18 Aug., Fish 15); 2- (black) Sedentary: 3 movements, 4 sites 745(22 Sept., Fish 50), 3- (white) Floater: 37 movements, 5 sites (24 Aug., Fish 37), 4- 746(dashed) Wanderer 14 movements, 11 sites (22 Sept, Fish 66). Contour shows depth at an747estimated discharge of 0.4 m3/s (flow stage: 15 cm).748749750751Fig. 4. a) Number of days Atlantic salmon parr stayed in the reach subdivided by 752behavioural type. Dashed line indicates the fish that stayed more than 6 days. b) 753Proportion of days fish showed each of the behavioural types. Most individuals exhibited 754all types of behaviour during the study period.755756Fig. 5. Time series of the number of individuals tracked on the study site decomposed by 757behavioural types.42 individuals were PIT tagged on 24 July and 24 more were added on 75828 Aug. 2008. 759760Fig.6. Proportion of fish behaviour exhibited on a daily basis by all individuals in relation761to flow stage: low (0-10 cm, n= 34days), median (10-15, n=34 days), high (15-20 cm, 762n=13 days), very high (>20 cm, n=16 days).67 34 68 763764Fig.7. a) Daily averaged flow depth used per behaviour types pooled for all fish (unequal 765number of days per fish). b) Daily averaged flow depth used, each line representing 766averages per behaviour types per individual. Five individuals (bold lines) exhibited a 767relatively higher depth used while adopting floater behaviour. 768769 770 771 772 773 774 775 776 777 778 779 780Figures 1 to 778169 35 70 78278378471 36 72 78578673 37 74 787 78875 38 76

Individual Variability in the Movement Behaviour of Juvenile Atlantic Salmon

Details

Download

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

Support