Smolt migration through the River Dee and harbour · Smolt migration through the River Dee and harbour January 2018 . 2 . 3 ... Smolt losses occurred in the middle and the lower river,

1

Smolt migration through the River Dee and harbour

January 2018

2

3

Executive Summary To investigate the extent and cause of in-river and estuarine mortality of salmon smolts in the River

Dee, 101 smolts from the upper and lower Dee catchment were tagged and tracked in 2017. Mortality

was high - 70% - for smolts from the upper catchment, and lower, but still significant - 13% - for smolts

from the lower catchment. This equated to an overall mortality rate of 0.45% per km migrated.

It is thought that mortality was due to predation. Smolt losses occurred in the middle and the lower

river, where predator densities are greatest. The timing and location of smolt losses showed that

tagged fish were surviving for, on average, at least 12 days after they were tagged, suggesting that

tagging/handling was not the direct cause of mortality. However, it is considered that smolts may be

made more vulnerable to predators by being tagged, and therefore it is possible that these levels of

mortality may be higher than that occurring in the untagged smolt population.

Although there were no confirmed losses in the harbour, as all tags were detected exiting the harbour,

the behaviour of six tagged smolts (11%) was unusual and could be due to the tagged fish being eaten

by a predator, and hence it was the movements of the predator that was detected. Total in-river and

estuarine mortality is therefore estimated as 48%.

Smolts from the upper catchment typically spent 20 days migrating through the river to the harbour,

whilst it took smolts from the lower catchment less than one day. Because smolts from the upper river

spent longer in the main stem Dee, they were therefore vulnerable to in-river predation for

substantially longer than smolts produced in the lower catchment.

The Spring of 2017 was exceptionally dry, and it is unclear whether smolt migration was influenced by

the unusual conditions (e.g. by delaying migration), which may have a bearing on susceptibility to

predation. Therefore in 2018 this study will be repeated to determine whether 2017 mortality levels

are standard or not.

From a management perspective, this is the second study (following a pilot study in 2016) that

highlights a potentially high predation pressure on salmon smolts in the Dee. Although not conclusive,

this work has been acknowledged by agencies and discussions are underway on what measures can

be introduced.

This tracking study has encouraged further investment on the Dee, and in 2018, alongside the third

year of this tracking study, there will be a larger study to investigate marine migration pathways of

smolts.

4

Introduction Around the Atlantic, salmon mortality can be high when smolts first enter the marine environment

(Kocik et al 2009, Thorstad et al 2012). This may be due to predation, and is influenced by the

availability of food for post-smolts, as this sets the rate at which they can outgrow predation risk.

With the recent decline in salmon stocks on the Dee (2012-2016), the Dee’s Fisheries Management

Plan (2015 – 2018) took a new focus to investigate the estuarine and coastal environment, where risk

of mortality is thought to be greatest. The plan set out to:

1) Quantify predation impacts on smolts,

2) Identify timings of smolt migration and their presence in the lower river and harbour area,

3) Establish near-shore habitat use of smolts and migration patterns through the estuary.

In 2016, acoustic tagging and tracking of salmon smolts on the Dee was used to investigate migration

and survival of salmon smolts in the lower river and harbour. Surprisingly, it highlighted smolt losses

occurred within the river, but no losses occurred within the inner harbour. In total, 26% of the tracked

smolts failed to reach the harbour in 2016, and this was thought to be due to either the impact of

tagging or in-river predation (Smolt migration through the lower Dee and inner harbour, River Dee

Trust 2016).

To investigate the cause of in-river losses, in 2017, additional fish were tagged in the upper catchment

(at the Baddoch smolt trap, which is operated by Marine Scotland Science), as well as at the Beltie and

Sheeoch smolt traps in the lower catchment. This was to investigate smolts losses throughout the

length of main stem river and determine the likely cause of mortality: losses due to the impact of

tagging would be expected to be high initially after tagging, but decline over time and as the fish

migrated downstream. On the other hand, mortality due to predation would increase later on during

the migration as smolts moved further downstream, where densities of predators are greater.

The tracking of smolts within Aberdeen Harbour was also extended in 2017, following reliable

performance of the acoustic receivers in the harbour in 2016. Aberdeen Harbour is the busiest port in

the UK, and shipping traffic and harbour works could potentially interrupt smolt migrations. In

addition, predators (seals, birds, estuarine fish, dolphins) are abundant in the harbour.

Methods

Acoustic telemetry Acoustic tags and receivers manufactured by Vemco were used for the study. The V5 tags transmit a

sound every 30 seconds (randomly generated at 15 - 45 second intervals). The sound produced by

each tag is a combination of 8 - 10 distinct pulses that give the tag a unique code, so that individual

fish can be identified. The V5 tags are 12.7 x 4.3 x 5.6 mm in size with a weight (in air) of 0.65g.

Telemetry guidelines suggest that tags should be no greater than 5 - 6.5% of the fish’s weight to avoid

adverse effects of tagging (Prentice et al 1990, Adams et al 1998, Anglea et al 2004). The smolts tagged

in this study were 16 – 23.5 g (average 20 g) and tag weight represented 2.8 – 4.1% (average 3.2%) of

smolt body weight.

V5 tags have a 95% battery life of 77 days and power output of 143 dB, presenting a maximum

detection range of approximately 300 m (albeit very dependent on background noise levels). The V5

5

tag is the second smallest tag currently available, with the smallest tag considered to have an

insufficient power output to ensure tag detection against the background noise in Aberdeen Harbour.

The VR2W acoustic receivers used in this study detect the sounds from the acoustic tags on the 180

kHz frequency. The receivers are placed underwater and make an automatic record each time a tag is

detected, recording the tag identification number, date and time, which can then be downloaded from

the receiver via Bluetooth once the receiver is retrieved from the water.

The receivers were placed underwater in the river and harbour, weighted onto the river bed with

anchor weights. In the river, receivers were attached to a metal rod and a 40-kg anchor weight, then

roped off to the bank to aid retrieval (Fig. 1). In the harbour, each receiver was attached to rope and

80 – 200 kg of anchor weight. The rope was held vertical by a sub-trawl float so that the receiver would

face upwards in the water column. A second rope held a surface float so that the position of the

receiver was known to boat traffic. The anchor weight was roped back to the shore/quayside to ensure

it was not lost in heavy storms and to aid retrieval (Fig. 2).

Figure 1. Receiver set up for in-river monitoring, ready for underwater deployment.

6

Figure 2. Deployment of receiver and mooring in Aberdeen Harbour.

Study area Smolts were captured in the upper (Baddoch burn) and lower (Beltie and Sheeoch burns) Dee

catchment. Rotary screw traps were used on the Beltie and Sheeoch burns, whilst a fixed trap was

used on the Baddoch burn (Fig. 3). The latter trap is operated by Marine Scotland Science (MSS), and

MSS personnel tagged the smolts at the Baddoch trap site1. The Baddoch, Beltie, and Sheeoch traps

were 122, 37 and 26.5 km (73, 22 and 16 miles) from the final receiver gate in the harbour,

respectively. The Baddoch burn is a significant tributary of the River Clunie, with the trap site being 11

km from the main stem Dee, whilst the traps on the Beltie and Sheeoch burns were located just 280

and 960 m, respectively, above the confluence with the Dee.

A total of 19 receivers were used to detect tagged fish. Nine of these were in the river and ten within

the harbour (Figs 3 and 4). Because of the channel width and high background noise levels in the

harbour, the receivers were paired up to form ‘gates’, to increase the likelihood of detecting tagged

smolts (Fig. 4).

The harbour was monitored as far seaward as the Old South Breakwater, so that the length of the

harbour over which smolts were monitored was 1.5 km (i.e. gate 1 to gate 5). This was an extension

to the harbour monitoring of 0.5 km since the 2016 pilot study.

1 http://www.gov.scot/Topics/marine/Salmon-Trout-Coarse/Freshwater/Monitoring/Traps

7

Figure 3. Map of River Dee, showing locations of smolt traps and acoustic receivers. The harbour area highlighted by the box is shown in Fig. 4.

8

Figure 4. Map of Dee estuary and Aberdeen Harbour, with locations of acoustic receivers (●).

9

Smolts In total, 101 smolts were tagged: 40 smolts from the Baddoch burn, 30 from the Beltie burn and 31

from the Sheeoch burn. Fish were chosen for tagging randomly, if they were in the length range of

120-125 mm (average 123 mm). This size range was the average smolt length across all three trapping

sites in 2016. There was no significant difference in smolt lengths between the three tagging sites in

2017 (ANOVA statistical test, P = 0.164). The body weight of smolts ranged from 16 to 23.5 g (average

20 g): the weight of Baddoch smolts was significantly greater (20.6 g) than lower catchment smolts

(19.5 g; ANOVA, P = 0.002). All tagged fish showed the physical attributes of smolt development (e.g.

Fig. 5). The condition of these smolts (Fulton Condition Factor; a measure of an individual fish’s health

based on weight) was 0.86 – 1.29 (1.065 ± 0.08; mean ± SD) and was significantly better for Baddoch

smolts (average 1.10) than lower catchment smolts (1.04; ANOVA, P = 0.0004).

Smolts were tagged between 4 and 27 April. Due to the dry spring in 2017, tagging was restricted to a

few days when flows were higher and most smolts entered the traps (smolts tended not to move in

the tributaries during low-flow conditions). Therefore 80 out of the 101 smolts were tagged in one

high-flow event between 25 and 27 April.

Figure 5. Salmon smolt showing silver colouration with loss of parr markings, streamlined body and

black edges to fins.

10

Tagging Smolts were tagged close to the river to reduce handling and transport. The surgical procedure was

carried out on a table using sterile equipment that was re-sterilised between each fish. Only staff that

had been trained and demonstrated post-training competence carried out the procedure.

Smolts were anaesthetised using MS-222 until they were heavily sedated. Each smolt was measured,

weighed and photographed prior to the tag being inserted. The tag was inserted into the body cavity

via a cut made into the belly of the fish and then the cut was closed using two sutures. The smolt was

then placed into a recovery unit for a minimum of two hours, until it appeared to be fully recovered

and was showing startle responses. Smolts were then released into the burn, along with other

untagged smolts captured in the trap, approximately 100 m downstream of the trap. The Standard

Operating Protocol worked to was based on guidelines from the Atlantic Salmon Federation.

To minimise handling and stress of these fish, scale samples to age fish were not taken. Based on scale

sampling of smolts caught in the traps in 2016, it would be expected that these smolts, of 120-125

mm fork length, would be two years old.

Data and analysis Much of the subsequent information on the tagged smolts is summarised using the ‘median’ value,

instead of an average or ‘mean’. This simply reflects that the factor being reported on (e.g. time taken

for migration) was heavily skewed and therefore the median (the middle value) better reflected the

‘typical’ smolt than the average value.

Various statistical analyses were done to interpret the data and the type of test used is always

reported on:

T-tests were used to compare differences between two groups of fish - such as between surviving and

non-surviving smolts – to identify what causes differences in behaviour or survival. Similarly, ANOVA

was used to compare differences between three groups of fish (e.g. Baddoch, Beltie and Sheeoch

groups).

To determine what factors influence migration speed and timing of migration, stepwise regression

models were used. For these models, each factor that potentially effects migration speed/timing –

e.g. smolt size, date of tagging - is added to the model, until a ‘best fit’ model is produced. This process

selects the factors that have the greatest influence on migration speed/timing. The model uses data

collected for individual fish, so that the migration speed of any smolt is related to its characteristics

(body length, date of tagging) and environmental conditions during its migration (photoperiod and

river flow it experienced during its migration). River flows were obtained from SEPA’s gauging station

at Park (Fig. 3). This records discharge (cubic metres per second, m3sec-1 or ‘cumecs’) every 15 minutes.

Photoperiod was the number of minutes of daylight each day at Aberdeen, based on daylight starting

30 minutes before sunrise and continuing until 30 minutes after sunset.

The strength of these statistical tests are reflected with the P-value. P-values weigh up the strength of

the evidence from the data: A P-value will be between 0 and 1 and the most crucial point is the 0.05

level - a P-value less than 0.05 indicates that the test has found strong evidence of a real effect or

relationship in the data, with only a 5% chance that this could have occurred randomly.

11

Findings

Fish detection 90 of the tagged smolts (89%) were detected subsequently by receivers. It is assumed that the 11

smolts (11%) that were never detected died because of delayed tagging effects. Delayed mortality

may occur for 24 – 36 hours after tagging (C. Adams, pers. comm.). The 11 fish that died were of similar

size (average 123.2 mm length, 20.1 g weight) to the surviving fish (123.4 mm, 19.8 g; t-test, P > 0.5),

they were from all three tagging sites, tagged during the same average water temperature and tagged

by different personnel. There was no difference in condition factor of these 11 fish (1.077) compared

to survivors (1.056; t-test, P = 0.43).

The remainder of the analysis is based on the 90 detected smolts.

Loss rates Total mortality of upper catchment (Baddoch) smolts during their main stem migration was 70%,

which was much greater than lower catchment smolts: Beltie smolt mortality was 18%, and Sheeoch

smolt mortality was 8% (Fig. 6). These mortality rates were based on main stem migration distances

of 107 km for Baddoch smolts, 29 km for Beltie smolts and 14 km for Sheeoch smolts.

The overall loss or mortality rate in the river, for all 90 smolts tracked, equated to 0.45 % km-1 (per

km), i.e. there was a 0.45 % chance of a smolt dying for each 1 km of river it travelled through. Smolts

from the Baddoch had a mortality rate of 0.57 % km-1, whilst mortality of smolts from the Beltie was

0.48 % km-1 and from the Sheeoch it was 0.30 % km-1.

Losses of Baddoch smolts was highest at 60 - 90 km downstream from their release site (Fig. 6), which

corresponds to between Craigendinnie (above Aboyne) and Lower Crathes (below Banchory). The loss

rate in this area was 1.6% km-1 travelled. Further losses of smolts from all three tagging sites occurred

in the lower river between Culter and Waterside, at a rate of 0.76% km-1.

Figure 6. Total cumulative mortality (%) of smolts from detection on the first main stem receiver

until exiting gate 5.

0

20

40

60

80

100

0 20 40 60 80 100 120

Tota

l mo

rtal

ity

(%)

Distance downstream (km)

Baddoch

Beltie

Sheeoch

12

Counts of goosanders are done monthly during Winter and Spring on the Dee. The river is canoed and

counted in four sections:

1. Feardar burn – Dinnet burn

2. Dinnet burn – Banchory Bridge

3. Banchory Bridge – Waterside

4. Waterside – Harbour

Counts since 2009 show that during the Spring (April and May), the number of goosanders is greatest

in the middle and lower river (sections 2 and 3), and low in the upper river and tidal waters, which

corresponds to the overall locations of smolt losses (Fig. 7).

Figure 7. Smolt mortality (Baddoch, Sheeoch and Beltie combined) at each receiver (●) and average

number of goosanders counted in April and May 2017 in each river section (●).

What influenced survival of smolts to the harbour? Overall, 57 (63%) smolts made the journey to the harbour whilst 33 (37%) smolts failed to reach the

harbour. There was no obvious factor influencing whether a smolt survived or not:

There was no significant difference in weight of survivors (19.8 g) and non-survivors (20.1 g; t-test,

P=0.43) or length (both 123 mm; t-test, P = 0.47).

There was no significant difference in the Fulton condition factor of survivors (1.056) and non-

survivors (1.076; t-test, P = 0.22).

Tagging date was similar for both survivors (21 April) and non-survivors (22 April; t-test, P = 0.45).

There was no difference in the time taken to reach the first receiver after being tagged, for surviving

and non-surviving smolts from the Baddoch (5 days, 14 hours and 5 days 11 hours, respectively; P =

0.96) or for smolts from all sites together (7 days 13 hrs and 5 d, 12 hours; P=0.24) (note there were

too few non-survivors from the Beltie or Sheeoch to test these sites individually).

0

10

20

30

40

50

60

0

5

10

15

20

25

30

35

0 20 40 60 80 100 120

Go

osa

nd

er n

um

ber

s

Mo

rtal

ity

(%)

Distance downstream (km)

Section 4

Section 3

Section 2

Section 1

13

The only significant difference between survivors and non-survivors out of all the factors recorded was

water temperature on the day of tagging, which was warmer for survivors (5.2°C) than non-survivors

(3.5°C; t-test, P < 0.0001). This may be due to most of the non-survivors being from the Baddoch,

which had lower water temperatures than the other two sites, and not for any biological reason.

Journey times As smolts are expected to take time to recover from tagging before returning to normal migration, the

time taken for smolts to reach the first receiver (between 9 and 15 km below the tagging sites) was

discounted from the rest of their in-river journey. Journey times are therefore measured once the fish

is recorded on the first receiver.

The Baddoch smolts typically spent 20 days (median value) in the river, from the first receiver to the

tidal waters in the lower river (Waterside receiver). This ranged between individual fish from 7 to 28

days. In contrast, the Beltie smolts typically took 8 hours (range from 4 hours - 12 days) to reach the

tidal waters, whilst the Sheeoch fish took 22 hours (range from 1.5 - 6 days). This variation is despite

most smolts being tagged within a three-day period.

Smolts from both the upper and lower catchment spent little time in the tidal part of the river (a 4.5

km stretch), typically moving through it within 3-6 hours.

Migration through the harbour The time smolts spent in the harbour was brief: from arriving at gate 1 to leaving at gate 5 (a distance

of 1.5 km), median journey time was 1 hour and 17 minutes, equivalent to a migration speed of 1.2

km hr-1; km per hour), with the quickest smolt taking just 37 minutes (2.4 km hr-1).

However, there were a few fish that spent a lot longer in the harbour and showed unexpected

movements. Five smolts spent between one and six days in the inner harbour, being detected on

different receivers which suggested that they were swimming back and forth. All five fish did

subsequently exit through the final gate and were not recorded again. A sixth smolt moved through

the habour initially but then spent 28 hours being recorded at the final gate. Furthermore, this was

during the high flow period of 28-29 April, making it seem unlikely that a smolt would choose to hold

station here. It is thought likely that the vastly different behaviour of these six fish (median time in

harbour was 4 days, 19 hours) compared to the other tagged smolts (median time 1 hour 9 min) was

because the smolts had been consumed by a predator and the tags were showing the predator’s

movements.

Beltie smolts reached the harbour significantly earlier (median 28 April) than Sheeoch (6 May) and

Baddoch (17 May) smolts. The date that a smolt arrived at the harbour was influenced by river flow

(mean daily discharge during the fish’s migration; Fig. 8) and the day they started their main stem

migration (date of arrival at the first receiver). Together, both factors explained 62% of the variation

in arrival dates at the harbour (Stepwise regression, P < 0.001). No other factors (body weight,

condition factor, migration speed, tagging date) were helpful in explaining harbour arrival date.

14

Figure 8. Arrival date at the harbour (gate 1), compared to river flows (line).

Smolts exited the final gate in the inner harbour between 28 April and 26 May. The main smolt

migration through the harbour, when between 25% and 75% of the smolts moved (standard for

migration time described by Malcolm et al 2015), was 28 April - 8 May.

Migration speeds In-river migration speeds depended on where the smolts had originated from: Smolts from the Beltie

burn moved very rapidly, at around 2.8 km hr-1, whereas smolts from the Baddoch and Sheeoch moved

much slower, at 0.22 km hr-1 and 0.39 km hr-1, respectively (Fig. 9).

In both the tidal river (downstream of Waterside) and the harbour, the swimming speeds of the

Baddoch and the Sheeoch smolts increased (0.8 – 1.6 km hr-1), whilst the Beltie smolts slowed down

(0.9 – 1.6 km hr-1). Overall, speeds through the tidal river and harbour were 0.8 and 1.2 km hr-1,

respectively.

Migration speed of the smolts in the river (from first receiver to gate 1) was most influenced by river

flow (mean daily discharge during each fish’s migration) and explained 41% of the variation in

migration speeds (Stepwise regression, P < 0.0001). None of the other tested factors had a significant

influence on migration speed (tagging date, body weight, condition factor, arrival date at first receiver,

photoperiod).

0

10

20

30

40

50

60

70

80

90

0

5

10

15

20

25

30

27

-Ap

r2

8-A

pr

29

-Ap

r3

0-A

pr

01

-May

02

-May

03

-May

04

-May

05

-May

06

-May

07

-May

08

-May

09

-May

10

-May

11

-May

12

-May

13

-May

14

-May

15

-May

16

-May

17

-May

18

-May

19

-May

20

-May

21

-May

22

-May

23

-May

24

-May

25

-May

26

-May

Mea

n f

low

at

Par

k (c

um

ecs)

Nu

mb

er o

f fi

sh a

rriv

ing

at h

arb

ou

r

Total fish Park flows

15

Figure 9. Speed of migrating smolts (km per hour).

0

1

2

3

4

5

6

7

Trap site to first receiver 1st receiver to Waterside Waterside to Gate 1 Gate 1 to Gate 5

Mig

rati

on

sp

eed

(km

per

ho

ur)

Baddoch River Tidal river Harbour

0

1

2

3

4

5

6

7

Trap site to first receiver 1st receiver to Waterside Waterside to Gate 1 Gate 1 to Gate 5

Mig

rati

on

sp

eed

(km

per

ho

ur)

Beltie

0

1

2

3

4

5

6

7

Trap site to first receiver 1st receiver to waterside Waterside to Gate 1 Gate 1 to Gate 5

Mig

rati

on

sp

eed

(km

per

ho

ur)

Sheeoch

River Tidal river Harbour

River Tidal river Harbour

16

However, whilst most smolts at all tagging sites were tagged prior to a rise in water, only smolts in the

Beltie migrated during the subsequent high flow. In contrast, Baddoch smolts reached the first

receiver during the high flow, but subsequently moved slowly, whilst Sheeoch smolts did not respond

to the rise in water and suffered a delay of several days before leaving the tributary and arriving at

the first receiver.

River flows 2017 was an exceptionally dry spring, with average daily flows in April - May being approximately 40%

of long-term average flow levels (SEPA, http://apps.sepa.org.uk/waterlevels/). Average daily flows at

Park in April and May 2017 were 18.0 cumecs, compared to 53.4 cumecs in 2016 (Fig. 10).

During April and May 2017 there was only a single rise in water, from 27-29 April. The three gauging

stations (Mar Lodge, Woodend, Park) corresponding best to the three tagging locations all showed a

similar flow response (Fig. 10) and therefore to simplify, the flow rates at Park were used as an

indicator of flow for the movements of all tagged fish.

Figure 10. River flows (in cubic metres per second; cumecs) during the 2017 smolt run.

The low flows before 27 April meant that there were few fish moving out of the tributaries, and

therefore available to tag. The single high flow event in the spring seemed to trigger nearly all fish to

move, allowing us to tag fish.

Further evidence that the unusually low river flows in spring 2017 delayed early migration of smolts

was seen in the Baddoch burn, where it was possible to tag a few fish before the high flow event: The

nine smolts that were tagged early in April typically took 25 days (median value) to reach the first

receiver at Lower Invercauld. In contrast, the 28 smolts that were tagged at the end of April took 2

days and 5 hours (median) to reach Lower Invercauld (t-test, P < 0.001).

The relationship between flow and main stem smolt migration varied between location (Fig. 11):

Baddoch smolts showed movement patterns that followed flow, whilst Beltie smolts used the high

flow event to move out of the river quickly, therefore providing only a narrow window for monitoring

their movements. In contrast, Sheeoch smolts did not move in the high flows after they were tagged,

but delayed migration for five days and then migrated in low flow conditions.

0

20

40

60

80

100

120

140

01

-Ap

r

05

-Ap

r

09

-Ap

r

13

-Ap

r

17

-Ap

r

21

-Ap

r

25

-Ap

r

29

-Ap

r

03

-May

07

-May

11

-May

15

-May

19

-May

23

-May

27

-May

31

-May

Mea

n d

aily

flo

w (

cum

ecs)

Mean daily flow at Park Mean daily flow at Woodend

mean daily flow at Mar Lodge mean daily flow at Park 2016

17

Figure 11. Smolt movements (initial detections at each receiver; bars) compared to river flows (line).

0

20

40

60

80

100

0

10

20

30

40

50

01

-Ap

r

05

-Ap

r

09

-Ap

r

13

-Ap

r

17

-Ap

r

21

-Ap

r

25

-Ap

r

29

-Ap

r

03

-May

07

-May

11

-May

15

-May

19

-May

23

-May

27

-May

31

-May

Mea

n d

aily

flo

w a

t P

ark

(cu

mec

s)

Nu

mb

er o

f fi

sh a

rriv

ing

at r

ecei

ver

Baddoch

L Invercauld Monaltrie Craigendinnie

L Woodend Culter Waterside

Boating Club Gate 1 Mean daily flow at Park

0

10

20

30

40

50

60

70

80

90

0

10

20

30

40

50

60

70

80

90

01

-Ap

r

05

-Ap

r

09

-Ap

r

13

-Ap

r

17

-Ap

r

21

-Ap

r

25

-Ap

r

29

-Ap

r

03

-May

07

-May

11

-May

15

-May

19

-May

23

-May

27

-May

31

-May

Mea

n d

aily

flo

w a

t P

ark

(cu

mec

s)

Nu

mb

er o

f fi

sh a

rriv

ing

at r

ecei

ver

Beltie

Culter Waterside Boating Club Gate 1 Mean daily flow at Park

0

10

20

30

40

50

60

70

80

90

0

2

4

6

8

10

12

14

01

-Ap

r

05

-Ap

r

09

-Ap

r

13

-Ap

r

17

-Ap

r

21

-Ap

r

25

-Ap

r

29

-Ap

r

03

-May

07

-May

11

-May

15

-May

19

-May

23

-May

27

-May

31

-May

Mea

n d

aily

flo

w a

t P

ark

(cu

mec

s)

Nu

mb

er o

f fi

sh a

rriv

ing

at r

ecei

ver Sheeoch

Culter Waterside Boating Club Gate 1 Mean daily flow at Park

18

Diurnal patterns In-river migration was predominantly nocturnal, and upstream of the tidal zone, between 81 and 100%

of fish detections on each receiver occurred during night time. Diurnal movements occurred later in

the spring (median date 3 May) than nocturnal movements (median date 28 April; t-test, P < 0.001), a

pattern that was also seen in 2016.

Once smolts moved into the tidal zone and harbour, diurnal migration increased (also seen in 2016),

accounting for approximately 50% of all recordings on the receivers. In the tidal zone of the river, the

preference for diurnal migration did not differ over the spring period. In the harbour, there was a

preference for diurnal migration earlier in the spring (median date 28 April) and nocturnal migration

increased later in the spring (4 May; P < 0.0001). This differed to that observed in 2016, when diurnal

migration increased during the spring, and may be due to the overriding influence of the single high-

flow event in 2017.

Tidal influences Smolts entered the harbour (gate 1) throughout the tidal cycle, although more smolts entered during

a falling tide (72%) than a rising tide (28%; Fig. 12). This was not evident in 2016, when slightly more

smolts arrived preceding low tide. It is also possible that tidal influences are a by-product of fish

moving during the high flows on 28 April.

Figure 12. Number of smolts arriving at the harbour (gate 1). Black vertical line represents low tide.

Receiver detection performance Maximum efficiency of each receiver was estimated based on the number of smolts that the receiver

failed to detect (i.e. smolts that were subsequently detected on receivers further downstream),

relative to the total number of smolts that could have been detected.

Based on this, most receivers (16 out of 19) had 95 - 100% maximum detection efficiency. However,

there were two receivers in the river (at Abergeldie and Lower Crathes) that had substantially lower

efficiencies (73 and 83%, respectively). It is thought that the location of these two receivers was

0

2

4

6

8

10

12

14

16

18

20

0-2 2-4 4-6 6-8 8-10 10-12

Nu

mb

er o

f sm

olt

s

Hours before high tide

19

unsuitable, and in the case of the Lower Crathes receiver, may have been exposed above the water

surface due to the low flow conditions.

One of the receivers in the final gate within the harbour (gate 5) also had a lower efficiency, recording

91% of the smolts. However, given that the width of gate 5 was 160 metres (due to the need to keep

equipment out of the shipping channel), we had not expected the receivers to fully monitor this

distance. Nevertheless, between the two receivers comprising gate 5, every fish was detected - as was

the case with all the other gates in the harbour. Therefore, as in 2016, gate efficiency in the harbour

was 100%.

Conclusions This was the second year of a three year smolt tracking programme on the Dee to investigate the

estuarine and coastal environment, where risk of salmon mortality was thought to be greatest. The

objectives set out in the Fisheries Management Plan are discussed below.

Quantifying predation impacts on smolts 37% of tagged smolts died in the river, equating to a mortality rate of 0.45% km-1 of river migration.

Although no losses occurred in the harbour, it is thought that an additional 11% of smolts were taken

by predators, as the movement patterns of these smolts were unexpected. In total, therefore, 48% of

smolts did not make the journey from their natal stream and out through the harbour.

Due to the efficiencies of receivers in the harbour area, it is assumed that 63% of fish did not reach

the harbour and either died in the river or decided not to migrate. The latter is considered unlikely as

all smolts were well advanced in the smolting process and all had migrated downstream since being

tagged, as they were detected on receivers below the traps. Tag failure rate is less than 2% (for all

VEMCO tags; specific failure rate for V5 tag not available) and so would not be expected to account

for more than two missing fish in the study. It is therefore considered that loss of tags from the study

is due to mortality.

Mortality was substantially higher for smolts from the upper catchment (70%) than the lower

catchment (13%). The only difference identified between tagged smolts from the upper and lower

catchment was that upper smolts were significantly heavier and had higher condition factor, which if

anything, should have been advantageous to survival. However, smolts travelling from the upper

catchment in 2017 experienced much longer exposure to any dangers that are within the river,

typically spending 20 days to reach the tidal limit, whereas lower catchment fish reached the tidal limit

in less than 24 hours.

The 33 fish that did not reach the harbour were tracked for an average of 11.5 days (range 1 – 34).

This demonstrates that fish survived for many days after undergoing the tagging procedure,

suggesting that the tagging was not the direct cause of mortality. It is possible that the tagging

procedure, or presence of the tag, affected smolt behaviour such as swim speed or manoeuvrability,

such that they were more vulnerable to predation. If this was the case, then the tagging could have

indirectly caused mortality and the level of mortality in these tagged smolts may not be representative

of untagged smolts. However, unlike most acoustic tracking studies, a particularly small acoustic tag

was used, which represented about 3% of body weight. Until further work is done on behaviour and

survival of smolts in the wild after tagging, indirect tagging effects cannot be ruled out, however, such

effects are expected to be smaller than in most other studies.

The results suggest that predators are the cause of the smolt mortalities, although the level of

mortality may be inflated because of tagging. Most mortality occurred in the middle river (very

20

approximately, between Aboyne and Crathes), and the main predator in this area would be

goosanders. Goosander densities are also greatest in the middle and lower river, excluding the tidal

waters. Mortality in the lower river (between Culter and Aberdeen) may also have been due to

predation from seals and kelts.

Identifying timings of smolt migration and presence in the lower river and harbour The peak time for smolt arrival at the harbour was 28 April – 8 May, just a few days earlier than in

2016 (2 – 10 May). This is surprising, given the very low flows in April 2017 that appeared to delay

smolts in initiating their migration in the tributaries – as evidenced by the lack of fish captured in the

fish traps. However, after rainfall in late April, smolts from the Beltie then moved rapidly, and Sheeoch

smolts also subsequently moved. Although Baddoch smolts had a late migration time, due to their

high in-river mortality they only represented 19% of survivors, and therefore did not have a significant

impact on the overall picture of migration timing.

The timing of migration and arrival at the harbour differed for smolts from the different tributaries.

Smolts from the upper catchment were much later to reach the harbour (17 May) compared to smolts

from the two tributaries in the lower catchment (28 April and 6 May). It is possible that this is a result

of the unusual flows in 2017 and needs to be looked at further, as the significance is that the timing

of smolt arrival into the marine environment is thought to be critical in determining survival at sea

(Friedland et al 2000). Furthermore, as river flow was found to be the crucial factor influencing the

timing of migration in both 2017 and 2016, the potential for climate-related changes to river flow

could also have a bearing on timing of smolt arrival in the marine environment and hence survival.

Establishing near-shore habitat use and migration patterns through the estuary In both 2016 and 2017, smolts spent very limited time in the estuarine area: approximately 3-6 hours

in the tidal river and 1¼ hours in the harbour. This time probably represents continuous travelling

along the total distance of 5.5 km. Despite the short length of time in estuarine water, mortality was

thought to be 11%, based on unusual tracks of six tagged smolts that were most likely consumed by

predators. The estuarine area would be the area of greatest predation risk, as fish-eating birds

(goosanders, mergansers, cormorants), seals and predatory marine fish would be in this area, and this

could perhaps explain why smolts spend so little time in this area.

Tidal patterns appeared to be related to the timing of smolts entering the harbour in 2017, which was

not seen in 2016. However, this is possibly a coincidence resulting from many fish migrating through

the harbour during high flows on 28 April; indeed, as the smolts moved quickly through the estuarine

area, this suggests that they did not need to wait for the tidal cycle.

2018 programme of work Smolt tracking will continue for a third year in 2018. The unusual flow conditions in 2017 may have

contributed to delayed migration and therefore greater vulnerability to in-river predation, so it is

important to repeat this work to determine if mortality rates remain high in 2018. As in 2017, 100

smolts will be tagged, including 40 smolts from the upper catchment (Baddoch), and 30 smolts from

each of the Beltie and Sheeoch tributaries in the lower catchment. Given the emerging picture of the

importance of in-river mortality, five extra receivers will be deployed to provide greater insight into

where losses occur in the river.

Although it is not possible to rule out that tagging the smolts made them more vulnerable to

predation, i.e. indirectly caused mortality, there is national and international work ongoing to look at

the issue of tagging effects. Through the national Tracking and Telemetry group, this work will be

21

pulled together in 2018 and will help determine whether the mortality found in the Dee smolts is

influenced by the use of tags.

A further smolt tracking project will start on the Dee in 2018, which is being delivered jointly by the

River Dee Trust and Marine Scotland Science, with funding from Aberdeen Offshore Wind Farm Ltd.

The focus of this tracking is to determine the migration route of smolts after they have left the river

in their early marine migration. 100 salmon smolts will be tagged on the Dee in 2018, and will be

detected in semi-circular arrays of acoustic receivers installed at distances of 4 km and 10 km from

the mouth of the Dee. Further smolts will be tagged in 2019 and 2020, including salmon and sea trout

smolts from the Rivers Don and Ythan, and arrays of acoustic receivers will be extended further

offshore to follow migration routes. The information from these tracked smolts will be used by Marine

Scotland Science to develop a model that can predict smolt migration pathways from other Scottish

rivers, to help establish where the sensitive marine areas for salmon exist.

Acknowledgements This work has been possible due to support from various people and groups:

Marine Scotland Science, in particular Rob Main helped with tagging and deploying receivers, Aya

Thorne, Stephen McLaren and Denise Stirling helped with tagging, and Iain Malcolm and John

Armstrong assisted in study design.

Aberdeen Harbour Board provided vessel, crew and maintenance staff to deploy and retrieve receivers

in the harbour and assist with moorings for the receivers.

We benefitted from advice from people with expertise in salmon acoustic telemetry to design this

study. Jon Carr (Atlantic Salmon Federation, Canada) trained staff in tagging procedures and offered

advice on study design, Dr Matt Newton and Professor Colin Adams (University of Glasgow) undertook

range testing and provided advice on study design and equipment.

SEPA provided river flow data from their gauging station at Park.

22

References Adams NS, Rondorf DW, Evans SD, Kelly JE & Perry RW (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. Anglea SM, Geist DR, Brown RS, Deters KA & McDonald RD (2004). Effects of acoustic transmitters on

swimming performance and predator avoidance of juvenile Chinook Salmon. North American Journal

of Fisheries Management 24, 162–170.

Friedland KD, Hansen LP, Dunkley DA & MacLean JC (2000). Linkage between ocean climate, post-

smolt growth, and survival of Atlantic salmon (Salmo salar L.) in the North Sea area. ICES Journal of

Marine Science 57, 419-429.

Kocik JF, Hawkes JP, Sheehan TF, Music PA & Beland KF (2009). Assessing estuarine and coastal

migration and survival of wild Atlantic salmon smolts from the Narraguagus river, Maine using

ultrasonic telemetry. American Fisheries Society Symposium 69, 293-310.

Malcolm IA, Millar CP & Millidine KJ (2015). Spatio-temporal variability in Scottish smolt emigration

times and sizes. Scottish Marine and Freshwater Science 6(2). Marine Scotland Science, Scottish

Government, pp.15.

Prentice EF, Flagg TA & McCutcheon CS (1990). Feasibility of using implantable Passive Integrated

Transponder (PIT) tags in salmonids. American Fisheries Society Symposium 7, 317-3220.

River Dee Trust (2016). Smolt migration through the lower Dee and inner harbour.

Thorstad EB, Whoriskey F, Uglem I, Moore A, Rikardsen AH & Finstad B (2012). A critical life stage of the Atlantic salmon Salmo salar: Behaviour and survival during the smolt and initial post-smolt migration. Journal of Fish Biology 81, 500–542.