**2.2 Field survey: Radio-tagging and tracking procedures**

Fish activity and movements were monitored using a sequential scanning receiver- Lotek Eng. Inc. SRX\_400 and a hand-held directional Yagi antenna (flexible elements) (Figure 2). Two different microprocessor coded radio transmitters (Lotek Engineering Ltd.) were used for the experiments carried out during 2002 and 2005. In the first and exploratory experiment, from 15 to 28 October 2002, one non-resident (from Sabor stream belonging to the contiguous watershed) native (330 mm total length, *L*T) and one stocked brown trout (270 mm *L*T) were tagged using a MCFT-3KM model (18 mm long x 7.3 mm diameter, 1.4 g in water) with 14 warranty life days and 5.00 sec. of signal burst rate. The transmitters operated with two codes (10 and 11) at the same frequency (149.420 MHz) and were attached (adding 0.5g, in air), alongside the base of the dorsal fin (Figure 3). The fish were previously anesthetized with 2-phenoxy-ethanol solution (0.25 mg.l-1) and the radio-tags externally attached with nylon cords, which passed through the body muscles (inside of a hypodermic needle) to plastic plates cushioned with foam on the two sides of the fish to minimise scale damage.

Fig. 2. Radiotelemetry monitoring session in the Baceiro river (summer 2005).

This previous study allowed to set the methodology for the 2nd experiment, which was conducted from 16 September to 18 November 2005 and a MCFT-3D model used with the following characteristics: 61 warranty life days, 5.00 sec. of signal burst rate, 29 mm long x 10.3 mm diameter and weight of 2.1g in water. They operated with six different codes (001 to 006) at the same frequency (149.460 MHz) and were externally attached on six hatchedreared brown trout (size range 255-277 mm in total length, *L*T, mean 265 ± 0.745 S.D. mm). Stocked trout were tagged according to the methodology defined, and maintained during one day in the hatchery to recover from the surgical procedures (Figure 3). After this period, fish were conditioned and transported in aerated tanks and, subsequently, released in the stream.

Fish activity and movements were monitored using a sequential scanning receiver- Lotek Eng. Inc. SRX\_400 and a hand-held directional Yagi antenna (flexible elements) (Figure 2). Two different microprocessor coded radio transmitters (Lotek Engineering Ltd.) were used for the experiments carried out during 2002 and 2005. In the first and exploratory experiment, from 15 to 28 October 2002, one non-resident (from Sabor stream belonging to the contiguous watershed) native (330 mm total length, *L*T) and one stocked brown trout (270 mm *L*T) were tagged using a MCFT-3KM model (18 mm long x 7.3 mm diameter, 1.4 g in water) with 14 warranty life days and 5.00 sec. of signal burst rate. The transmitters operated with two codes (10 and 11) at the same frequency (149.420 MHz) and were attached (adding 0.5g, in air), alongside the base of the dorsal fin (Figure 3). The fish were previously anesthetized with 2-phenoxy-ethanol solution (0.25 mg.l-1) and the radio-tags externally attached with nylon cords, which passed through the body muscles (inside of a hypodermic needle) to plastic plates cushioned with foam on the two sides of the fish to

Fig. 2. Radiotelemetry monitoring session in the Baceiro river (summer 2005).

This previous study allowed to set the methodology for the 2nd experiment, which was conducted from 16 September to 18 November 2005 and a MCFT-3D model used with the following characteristics: 61 warranty life days, 5.00 sec. of signal burst rate, 29 mm long x 10.3 mm diameter and weight of 2.1g in water. They operated with six different codes (001 to 006) at the same frequency (149.460 MHz) and were externally attached on six hatchedreared brown trout (size range 255-277 mm in total length, *L*T, mean 265 ± 0.745 S.D. mm). Stocked trout were tagged according to the methodology defined, and maintained during one day in the hatchery to recover from the surgical procedures (Figure 3). After this period, fish were conditioned and transported in aerated tanks and, subsequently, released in the

**2.2 Field survey: Radio-tagging and tracking procedures** 

minimise scale damage.

stream.

Fig. 3. Trout radio-tagging procedures and recover period of stocked trout in fishfarms, located near of Baceiro River.

The habitat unit selected for the release of stocked trout was 210 m long by 9.0 m mean width by 2.5 m of maximum depth, comprising all representative microhabitats of stream segment. Temperature (thermometer, accuracy of 0.1 ºC) and water column velocity (Valeport flowmeter, accuracy of 0.01 m.s-1) were daily measured (Figure 4) and stream discharge determined near the stocking site. Velocity at 0.6 of total depth was considered as the mean water column velocity when the depth was less than 0.75 m. At deeper points the readings were averaged at 0.2 and 0.8 of total depth.

Fig. 4. Measuring temperature (oC) and water column velocity (m.s-1) in the Baceiro River (summer 2005).

The fish were monitored and located at least once a day until the end of their transmitter's battery life during the whole study period (14 days in 2002 and 64 days in 2005). Net daily journeys were registered, which were defined as the distance between locations at two consecutive days. During 2005, the fish were also monitored hourly for a partial diel cycle (from 06.00 a.m. to 24.00 p.m.) for eight days (week periodicity). Such registrations took place on 23 and 30 September, on 7, 14, 21 and 28 October and on 4 and 12 November. All tracks were conducted along the stream banks and the potential disturbance of fish activity minimized. To measure the trout movements, yellow fluorescent marks were sprayed on the

Combining Radio and PIT-Telemetry to Study the Large and Fine-Scale

with 12 mm PIT tags and the adipose fin clipped.

unique identification codes obtained for all tagged fish.

procedures.

the incision.

Movements of Stocked and Wild Brown Trout (*Salmo trutta* L.) in a Northeastern Stream, Portugal 335

bedrock. Cover types were divided into five categories: 1) objects > 15 cm (substrate emerging from the streambed); 2) overhanging vegetation; 3) roots, undercut banks and submerged woody debris; 4) surface turbulence and 5) no cover. Total depth was directly measured with a stick meter and the velocities were measured with a Valeport electronic flowmeter. The following characteristics were determined for the available habitat: mean total depth of 40 cm (maximum depth= 90 cm); maximum water column velocity detected near the riffle zone of 0.90 m.s-1; substrate composition dominated by sand, cobbles and boulders; main cover for fish provided by undercut banks and boulders. Water temperature ranged from 12 to 19 ºC. Between 12 August and 30 September 2005, the entire stream reach section selected was closed with stop nets. Previously to the beginning of the experiments, the study area was depleted of fish through several electrofishing sweeps (Hans Grassl ELT60 DC, 1.5W, 300/600 volts) and biometric data of local trout population recorded. Twenty-five resident native trout, distributed into three size classes (Table 1), were marked

Fig. 5. PIT equipment (battery-pack and multi-point decoder- MPD) unit and PIT tagging

After a recovery period of two hours, the wild trout population was released into the blocked stream reach. At the same time, a sympatric condition was promoted in the confined area adding a total of fifty PIT tagged stocked trout using transponders Type I and II (Table 1). Before tagging, individual fish were anesthetized with a solution of 2 phenoxy-ethanol (0.25 ml.l-1) and the abdominal region disinfected (Betadine®). A sterilised needle linked to a special tagging gun was used for surgical implantation of the Type I tag in the fish peritoneal cavity (Figure 5). The Type II tag was manually implanted through an incision of approximately 4 mm made in the midventral line, without suturing

The MPD unit, the antennae installation and the data acquisition were made following a similar design described in Riley et al. (2003) and Teixeira & Cortes (2007), using a random distribution of antennae, changing their position every two days (Figure 6). During the study period the dry weather conditions verified and the values of microhabitat measurements were assumed constant for every two days. Biometric data of both sympatric populations were obtained five weeks after stocking through an electrofishing survey and

stream bank (alder branches or rocks were selected) at regular intervals of 25 meters. The identification of a fish position was registered after the detection of the maximum signal strength for at least 1 min. The positions of each fish were used to determine the dispersal (defined as the distance travelled by individual fish from the stocking site), the daily home range (D.H.R., the difference between the most upstream and most downstream positions), and the total distance moved (T.D.M., the sum of all displacements detected). Nonparametric Mann-Whitney *U*-tests were performed to detect statistical differences between native and stocked fish dispersal in 2002 and between stocked trout for dispersal, D.H.R. and T.D.M. throughout 2005. Spearman rank order correlations (*r*S) were made to assess the significant relationship between the dispersal of stocked fish and two relevant environmental variables: water temperature and discharge. All statistical analyses were performed using STATISTICA 7.0 package (Statsoft, 2004).

#### **2.3 Field survey: PIT-tagging and monitoring design**

The Passive Integrated Transponder (PIT) technology is composed of PIT tags, which are internally implanted in the fish, and one or several antennae connected to a transceiver. The PIT tag is detected and their individual code recorded when a tagged fish passed within the read range of the antenna. The fish detection is recorded when the transceiver energizes the tag by sending an electric current through the antenna, which emits an electromagnetic signal captured by the circuit board of the PIT tag that sends their individual code back to the transceiver (Riley et al., 2003; Gibbons & Andrews, 2004). The PIT technology used was based on a multi-point decoder (MPD) unit (UKID Systems Ltd, Preston, U.K.). This unit consists of DC integrated MPD/antenna multiplexer (8-channel) powered by a 24 V (18 Ah) rechargeable lead-acid battery pack, which provided more than 24 hours of continuous use, and eight black circular panel antennae connected to the PIT–tag reader by cable lengths of 10 m. Each panel antenna (22 mm deep and 300 mm in diameter) operates at a frequency of 134 kHz. Two distinct PIT tags (UKID Systems) were used in this study: 1) 12.0 mm long x 2.1 mm in diameter (122IJ) (defined as Type I) and 2) 34 mm (L) x 4 mm (D) (Type II) (344GL), with detection ranges of approximately 90 mm and 300 mm, respectively. This system enables logging up to 1000 time-stamped events from an onboard Real Time Clock and the Battery Backed-up Memory. In order to reduce the number of repetitive events, resulting from a fish that remained over the same antenna, a data repeated filter precluded the repeat reading of the same tag code within each 25 seconds period. The identification data (ID) output was further downloaded from MPD (via RS232) to a personal computer. The battery pack and the MPD was safeguard by a special enclosure (Peli-Plastic case) (Figure 5). A Casper Handheld reader was used when fish were captured and a unique identification required.

A stream segment (30 m long by a width ranging from 3 to 10 m), with riffle and pool habitats, was selected in the Baceiro stream. Before PIT telemetry experiment, the aquatic habitat was assessed based on transects (starting point randomly chosen), made perpendicular to the stream, with intervals of 5 m throughout each stream segment. Point measurements were done at 0.5 m intervals across each transect for the variables of total depth, surface velocity (measured 10 cm below the surface), bottom velocity (10 cm above the streambed) and mean water column velocity (0.6 of total depth), substrate composition and cover. Substrate composition was classified according to a modified Wentworth scale, adopting the following categories: 1) organic detritus; 2) silt and sand (< 2 mm); 3) gravel (2- 16 mm); 4) pebble (17- 64 mm); 5) cobble (65- 256 mm); 6) boulder (> 256 mm) and 7)

#### Combining Radio and PIT-Telemetry to Study the Large and Fine-Scale Movements of Stocked and Wild Brown Trout (*Salmo trutta* L.) in a Northeastern Stream, Portugal 335

334 Modern Telemetry

stream bank (alder branches or rocks were selected) at regular intervals of 25 meters. The identification of a fish position was registered after the detection of the maximum signal strength for at least 1 min. The positions of each fish were used to determine the dispersal (defined as the distance travelled by individual fish from the stocking site), the daily home range (D.H.R., the difference between the most upstream and most downstream positions), and the total distance moved (T.D.M., the sum of all displacements detected). Nonparametric Mann-Whitney *U*-tests were performed to detect statistical differences between native and stocked fish dispersal in 2002 and between stocked trout for dispersal, D.H.R. and T.D.M. throughout 2005. Spearman rank order correlations (*r*S) were made to assess the significant relationship between the dispersal of stocked fish and two relevant environmental variables: water temperature and discharge. All statistical analyses were

The Passive Integrated Transponder (PIT) technology is composed of PIT tags, which are internally implanted in the fish, and one or several antennae connected to a transceiver. The PIT tag is detected and their individual code recorded when a tagged fish passed within the read range of the antenna. The fish detection is recorded when the transceiver energizes the tag by sending an electric current through the antenna, which emits an electromagnetic signal captured by the circuit board of the PIT tag that sends their individual code back to the transceiver (Riley et al., 2003; Gibbons & Andrews, 2004). The PIT technology used was based on a multi-point decoder (MPD) unit (UKID Systems Ltd, Preston, U.K.). This unit consists of DC integrated MPD/antenna multiplexer (8-channel) powered by a 24 V (18 Ah) rechargeable lead-acid battery pack, which provided more than 24 hours of continuous use, and eight black circular panel antennae connected to the PIT–tag reader by cable lengths of 10 m. Each panel antenna (22 mm deep and 300 mm in diameter) operates at a frequency of 134 kHz. Two distinct PIT tags (UKID Systems) were used in this study: 1) 12.0 mm long x 2.1 mm in diameter (122IJ) (defined as Type I) and 2) 34 mm (L) x 4 mm (D) (Type II) (344GL), with detection ranges of approximately 90 mm and 300 mm, respectively. This system enables logging up to 1000 time-stamped events from an onboard Real Time Clock and the Battery Backed-up Memory. In order to reduce the number of repetitive events, resulting from a fish that remained over the same antenna, a data repeated filter precluded the repeat reading of the same tag code within each 25 seconds period. The identification data (ID) output was further downloaded from MPD (via RS232) to a personal computer. The battery pack and the MPD was safeguard by a special enclosure (Peli-Plastic case) (Figure 5). A Casper Handheld reader was used when fish were captured and a unique

A stream segment (30 m long by a width ranging from 3 to 10 m), with riffle and pool habitats, was selected in the Baceiro stream. Before PIT telemetry experiment, the aquatic habitat was assessed based on transects (starting point randomly chosen), made perpendicular to the stream, with intervals of 5 m throughout each stream segment. Point measurements were done at 0.5 m intervals across each transect for the variables of total depth, surface velocity (measured 10 cm below the surface), bottom velocity (10 cm above the streambed) and mean water column velocity (0.6 of total depth), substrate composition and cover. Substrate composition was classified according to a modified Wentworth scale, adopting the following categories: 1) organic detritus; 2) silt and sand (< 2 mm); 3) gravel (2- 16 mm); 4) pebble (17- 64 mm); 5) cobble (65- 256 mm); 6) boulder (> 256 mm) and 7)

performed using STATISTICA 7.0 package (Statsoft, 2004).

**2.3 Field survey: PIT-tagging and monitoring design** 

identification required.

bedrock. Cover types were divided into five categories: 1) objects > 15 cm (substrate emerging from the streambed); 2) overhanging vegetation; 3) roots, undercut banks and submerged woody debris; 4) surface turbulence and 5) no cover. Total depth was directly measured with a stick meter and the velocities were measured with a Valeport electronic flowmeter. The following characteristics were determined for the available habitat: mean total depth of 40 cm (maximum depth= 90 cm); maximum water column velocity detected near the riffle zone of 0.90 m.s-1; substrate composition dominated by sand, cobbles and boulders; main cover for fish provided by undercut banks and boulders. Water temperature ranged from 12 to 19 ºC. Between 12 August and 30 September 2005, the entire stream reach section selected was closed with stop nets. Previously to the beginning of the experiments, the study area was depleted of fish through several electrofishing sweeps (Hans Grassl ELT60 DC, 1.5W, 300/600 volts) and biometric data of local trout population recorded. Twenty-five resident native trout, distributed into three size classes (Table 1), were marked with 12 mm PIT tags and the adipose fin clipped.

Fig. 5. PIT equipment (battery-pack and multi-point decoder- MPD) unit and PIT tagging procedures.

After a recovery period of two hours, the wild trout population was released into the blocked stream reach. At the same time, a sympatric condition was promoted in the confined area adding a total of fifty PIT tagged stocked trout using transponders Type I and II (Table 1). Before tagging, individual fish were anesthetized with a solution of 2 phenoxy-ethanol (0.25 ml.l-1) and the abdominal region disinfected (Betadine®). A sterilised needle linked to a special tagging gun was used for surgical implantation of the Type I tag in the fish peritoneal cavity (Figure 5). The Type II tag was manually implanted through an incision of approximately 4 mm made in the midventral line, without suturing the incision.

The MPD unit, the antennae installation and the data acquisition were made following a similar design described in Riley et al. (2003) and Teixeira & Cortes (2007), using a random distribution of antennae, changing their position every two days (Figure 6). During the study period the dry weather conditions verified and the values of microhabitat measurements were assumed constant for every two days. Biometric data of both sympatric populations were obtained five weeks after stocking through an electrofishing survey and unique identification codes obtained for all tagged fish.

Combining Radio and PIT-Telemetry to Study the Large and Fine-Scale

0.05 was accepted.

riffle/run habitat.

**3.1 Radio-telemetry analysis** 

**3. Results** 

Movements of Stocked and Wild Brown Trout (*Salmo trutta* L.) in a Northeastern Stream, Portugal 337

the sampling period. Non-metric multidimensional scaling (NMDS) analysis, an ordination method based on a rank order of Bray-Curtis similarities, was applied to nonrepeated frequency to detect the behaviour differences between native and stocked trout. The NMDS was computed using the log transformed [log (x+1)] data. A multivariate analysis of similarities- one-way ANOSIM test, as a nonparametric randomization approach, was then applied to the Bray-Curtis similarity matrix to test the statistical differences between the two considered groups. These analyses were performed through the package PRIMER 5 (Clarke & Gorley, 2001). The relationship between the microhabitat variables and stocked and wild trout were assessed through a canonical correspondence analysis (CCA), a method of direct gradient analysis, where the ordination of objects (stocked and native fish) is based on species data (fish positions) and on environmental information associated (Jongman et al., 1987). Two CCA's were performed for two distinct periods: 1) first week- the adaptation period of stocked fish to wild environment; 2) from the 2nd to the 5th week, considering the post-adaptation period. This analysis was performed using the CANOCO software package (ter Braak & Smilauer, 1998). Data were standardized for microhabitat variables and log transformed [log (x+1)] for the non-repeated frequency data of the detected fish in all antennae positions. Only those variables with a variation inflation factor (VIF) of less than 20 were included to avoid multicollinearity (ter Braak, 1986). In addition, a Monte Carlo permutation test (199 permutations) was performed to test the significance of the axes. The fish activity was analysed based on the antennae non-repeated frequency data for the tagged trout populations, considering the following classes for *1)* native trout: A< 15.0; B-15.0 to 20.0; C> 20.0 cm using Type I PIT tags and *2)* stocked brown trout: Type I and Type II PIT tags. Polynomial regressions were fitted to the data. Trout activity pattern was analyzed over 24-hours cycles, but discriminated for dawn, day, dusk and night periods. The influence of distinct detection range of the two types of PIT tags (only for stocked fish) and the ontogenetic variation of native trout was assessed. The differences between the trout classes for the defined periods were assessed using non-parametric Mann-Whitney *U*-tests (data did not fit to the assumptions of normality- Bartlet test). These *U*tests were also performed for the comparisons between size (total length, *L*T), mass (*M*) and the Fulton's condition factor (*K*) of stocked and native trout. A significant level of P<

A distinct movement pattern was detected comparing stocked and native brown trout (Mann-Withney *U*-test, P< 0.001), from the radiotelemetry survey carried out in autumn 2002 (Figure 7). An initial stationary behaviour of stocked trout (for five days remaining in the stocking site) was replaced by their migration in a downstream direction, and at the end of 14 days (transmitter battery life) the fish were located in a small pool, 1,500 m from the stocking site. The magnitude of the displacement was correlated with the increase of stream discharge (Spearman correlation *r*S> 0.85, P< 0.01). Conversely, wild trout remained near the stocking site hiding under a fallen tree, in spite of the non-residency status. It was only detected a downstream movement of 200 m coinciding with a sudden rainstorm, which raised the water level by 1 m. Nevertheless, after this period, the wild trout followed the upstream migration and travelled to feeding zones (90 m from stocking site) near a


*\* K = 100(M. LT-b), where M*- trout mass (g); *LT*- total lenght (cm); *b-* alometric coefficient \*\* 100.(tag mass).(trout mass)-1

Table 1. Mean ± standard deviation (S.D.) of total length (*L*T), mass (*M*), Fulton's condition factor (*K*) and tag ratio of the pit-tagged brown trout in the Baceiro stream, during summer 2005.

Fig. 6. PIT-tagged trout passing over an antenna, during field experiment in the Baceiro River (summer 2005).

The analyses of movement and activity patterns of stocked and native trout populations were based on the non-repeated data (the continuous repeated records of each fish in the same antenna were not considered) recorded by the MPD unit during the five weeks of the sampling period. Non-metric multidimensional scaling (NMDS) analysis, an ordination method based on a rank order of Bray-Curtis similarities, was applied to nonrepeated frequency to detect the behaviour differences between native and stocked trout. The NMDS was computed using the log transformed [log (x+1)] data. A multivariate analysis of similarities- one-way ANOSIM test, as a nonparametric randomization approach, was then applied to the Bray-Curtis similarity matrix to test the statistical differences between the two considered groups. These analyses were performed through the package PRIMER 5 (Clarke & Gorley, 2001). The relationship between the microhabitat variables and stocked and wild trout were assessed through a canonical correspondence analysis (CCA), a method of direct gradient analysis, where the ordination of objects (stocked and native fish) is based on species data (fish positions) and on environmental information associated (Jongman et al., 1987). Two CCA's were performed for two distinct periods: 1) first week- the adaptation period of stocked fish to wild environment; 2) from the 2nd to the 5th week, considering the post-adaptation period. This analysis was performed using the CANOCO software package (ter Braak & Smilauer, 1998). Data were standardized for microhabitat variables and log transformed [log (x+1)] for the non-repeated frequency data of the detected fish in all antennae positions. Only those variables with a variation inflation factor (VIF) of less than 20 were included to avoid multicollinearity (ter Braak, 1986). In addition, a Monte Carlo permutation test (199 permutations) was performed to test the significance of the axes. The fish activity was analysed based on the antennae non-repeated frequency data for the tagged trout populations, considering the following classes for *1)* native trout: A< 15.0; B-15.0 to 20.0; C> 20.0 cm using Type I PIT tags and *2)* stocked brown trout: Type I and Type II PIT tags. Polynomial regressions were fitted to the data. Trout activity pattern was analyzed over 24-hours cycles, but discriminated for dawn, day, dusk and night periods. The influence of distinct detection range of the two types of PIT tags (only for stocked fish) and the ontogenetic variation of native trout was assessed. The differences between the trout classes for the defined periods were assessed using non-parametric Mann-Whitney *U*-tests (data did not fit to the assumptions of normality- Bartlet test). These *U*tests were also performed for the comparisons between size (total length, *L*T), mass (*M*) and the Fulton's condition factor (*K*) of stocked and native trout. A significant level of P< 0.05 was accepted.
