Spring salmon on the River South Esk, Scotland: Scottish Marine and Freshwater Science Vol 7 No 10

Part 2: Spawning locations of spring fish

Objective: In which geographic region of the River South Esk do spring salmon spawn?

2.1 Introduction

Salmon returning in spring to the Rivers Spey, Dee and Tay tend to originate from and return to upper catchment tributaries, whereas those returning in the autumn tend to originate from and return to lower catchment tributaries and the river main stem (Laughton & Smith, 1992). However, while this pattern appears to hold true where the relationship between distance upstream and spawning area is relatively simple, it is not universal. In more complex systems, the relationship between distance upstream and spawning areas is less clear, for example in the River Tweed (Campbell, 1998). Our knowledge of the factors affecting where spring fish are likely to spawn in a given catchment is not currently sufficient to predict spawning areas for spring fish for a specific river, and there is no information regarding the locations of spring salmon spawning areas in the River South Esk. It is necessary to identify such areas to establish whether there are low numbers of juveniles associated with an apparent lack of spring returns, and for identifying whether there is local habitat degradation that may have resulted in reduced smolt production.

This issue was approached using two methods. Firstly, spring salmon sourced from either the coastal net fishery adjacent to the River South Esk, or the lower reaches of the River South Esk itself, were radio tagged and tracked to their spawning locations. There is a long history of using radio transmitters placed in the stomach to track in-river migration of salmon (e.g. Hawkins & Smith, 1986; Laughton, 1989; 1991). However, despite being a tried-and-tested method, radio tracking is relatively expensive and relies on fish being subsequently detected post-tagging. Therefore, the potential to identify the natal origin of Atlantic salmon within the River South Esk using genetic analysis was also assessed. Genetic techniques potentially allow returning adults to be assigned to river of origin and / or region of origin within river (i.e. upper catchment vs. lower catchment or individual tributaries). If possible, this technique presents a number of advantages over more traditional techniques such as radio tracking. The sample size is larger using genetic techniques, as all captured fish have the potential to be assigned at relatively low cost. This approach is also not affected by post-tagging losses (e.g. predation, tag regurgitation, fish entering rivers not being monitored for tags), and any variation in these factors among fish destined for different rivers. As long as a comprehensive baseline survey is carried out, this approach is also unaffected by surveying biases, such as differences in tracking effort afforded to different rivers. The genetics approach relies only on samples being collected from returning adults and comparing the genetic signature of these samples to baseline data from the potential rivers of origin. However, a requirement of the genetics approach is that baseline data is sufficiently differentiated to allow separation among areas of interest, and the results of the present study will demonstrate whether such differentiation exists.

2.2 Methods

2.2.1 Radio tracking

Source of fish

In the first year of this three-year investigation (2012), spring salmon were captured from the coastal net fishery near the mouth of the River South Esk ( Figure 2.2.1.1 a) and radio tagged. In the second year of the investigation (2013), radio tagging of fish sourced from the coastal nets continued. Tagging of spring salmon within the River South Esk itself was also initiated in 2013 by establishing a net in the lower reaches of the river ( Figure 2.2.1.1 b). In the third and final year of the investigation (2014), coastal netting was discontinued and fish for tagging were sourced from the in-river net as well as from a rod fishery adjacent to the in-river netting site. A fixed engine net of the type used in the coastal net fishery is shown in Figure 2.2.1.2 a, while the in-river net used is shown in Figure 2.2.1.2 b.

Figure 2.2.1.1 a

Figure 2.2.1.1 b

Figure 2.2.1.2 a: A fixed engine net of the type used in the coastal net fishery.

Figure 2.2.1.2 b: The in-river net.

All nets were operated by personnel from Usan Salmon Fisheries Ltd. Weather permitting, coastal nets were operated from Monday to Friday inclusive between the dates specified in Table 2.2.1.1. River level permitting, the in-river net was operated from Monday to Sunday inclusive between the dates specified in Table 2.2.1.1. The numbers of fish tagged by capture method are indicated in Figures 2.2.1.3 a to 2.2.1.3 c.

Table 2.2.1.1: Summary of the tagging regime and resulting numbers of spring salmon tagged from each source during the three years of the investigation.

Number of fish tagged from each source
Year	Tagging period	Coastal nets	In-river net	Rods
2012	20 February to 31 May	153	N/A	N/A
2013	04 March to 26 April	038	22	N/A
2014	17 February to 27 April	N/A	20	12

Figure 2.2.1.3 a: Number of fish tagged per week during 2012 sourced from the coastal nets. Figure

Figure 2.2.1.3 b

Figure 2.2.1.3 c

In the coastal net fishery, salmon were removed from the net in which they were captured and placed into a holding tank filled with approximately 150 - 200 L of sea water. Fish were then removed from the holding tank one at a time and placed in a bath containing 75 L of anaesthetic solution (23 mg / L eugenol in sea water). Fish caught in the in-river net were transferred directly from the net in which they were captured to the anaesthetic solution one at a time. Once anaesthetised, fitness for tagging was assessed (fish with signs of physical damage were not tagged), the sex of the fish was assigned using external physical characteristics, five scales were removed for later age determination, fork length was measured (to the nearest cm), and a small sample of fin tissue was removed and retained in ethanol for subsequent genetic analysis.

Sex assignment was carried out using a method described by Shearer (1972) based on head shape. These assessments are subjective, and were, given the time of year, carried out on fish before the secondary sexual characteristics had developed. As it was not possible to determine the age of the fish at the time of tagging, criteria were developed to ensure that fish selected for tagging were likely to be MSW salmon. Analysis of sample data from the nearby River North Esk suggested that at the time of year in which fish were being tagged, 1SW salmon were rare and were generally smaller than 60 cm, and thus only fish with a fork length of 60 cm (to the nearest whole cm) or greater were selected for tagging. The ages of tagged fish were subsequently determined by scale reading.

Each fish selected for tagging had a radio tag (Lotek individually coded MCFT2-16-CE tag with integrated antenna, 46 mm in length with a diameter of 16 mm and weighing 16 g in air) gently introduced into the stomach. This was achieved by inserting an acrylic tube containing a radio tag into the oesophagus. A plunger was then used to place the radio tag into the stomach. The tube and plunger were then removed, and the fish was placed in a tank of sea water (coastal) or river water (in-river) and allowed to recover fully. Those fish that were too small and / or were not in prime condition were placed in a tank of sea water (coastal) or a keepnet positioned adjacent to the river bank (in-river) and allowed to recover before being released. Fish captured in the coastal nets were transported to and released at a specific point approximately 0.5 km from the nearest net to reduce the likelihood of recapture ( Figure 2.2.1.1 a). Similarly, all fish captured by the in-river net were released, approximately 200 m upstream of the netting site.

Salmon sourced from rods during the final year of the project were processed in exactly the same way as fish captured using the in-river net. A number of keepnets were positioned at strategic locations in the area around the in-river netting site for use by anglers to allow time for the tagging team to arrive on site. These fish were tagged and released at the site of capture. All procedures carried out on the fish were performed under Home Office licence.

Tracking locations

It was possible to detect radio tagged spring salmon only in fresh water as radio signals are quickly attenuated in salt and brackish water. A combination of fixed and mobile radio receivers was used to track tagged fish in fresh water in three ways - fixed receiver stations, aerial surveys and manual tracking. Fixed receiver stations were deployed on the River South Esk and a number of other rivers in the east of Scotland ( Figure 2.2.1.4). Mobile receivers were deployed, either during helicopter surveys, or by manual tracking, either on foot or by vehicle. A summary of the extent of aerial survey coverage on each river during the three years of the investigation is shown in Table 2.2.1.2. A summary of the tracking methods employed on each river during the three years of the investigation and the numbers of tags detected in rivers by fixed receiver stations (where they existed) and aerial surveys (where they were carried out) is shown in Table 2.2.1.3.

Figure 2.2.1.4: Rivers in the east of Scotland on which fixed receiver stations were deployed.

Table 2.2.1.2: Summary of the extent of aerial survey coverage on each river during the three years of the investigation. For clarity of presentation, rivers which were either not monitored and / or where no aerial survey was carried out in a given year have been excluded.

( a) Summary of aerial survey coverage in 2012

					Downstream limit of aerial survey			Upstream limit of aerial survey
Catchment	Date surveyed	Survey number	River name	Description	Easting	Northing	Description	Easting	Northing	Description
Don	15/11/2012	1 (of 1)	River Don	Mainstem of the River Don	393600	809025	Weir upstream of A956 roadbridge, Aberdeen	322200	808800	Torran Deallaig
			River Urie	Tributary of the River Don	378325	820050	Confluence of the River Urie with the mainstem of the River Don	364850	831000	Mill Croft
			Lochter Burn	Tributary of the River Urie (itself a tributary of the River Don)	377325	823000	Confluence of the Lochter Burn with the River Urie	377000	824500	Lethenty Mill
Dee	02/11/2012	1 (of 3)	River Dee	Mainstem of the River Dee	392500	803050	Upstream of Bridge of Dee, Aberdeen	307100	789650	Confluence of the Lui Water with the mainstem of the River Dee
			Water of Tanar	Tributary of the River Dee	351350	797550	Confluence of the Water of Tanar with the mainstem of the River Dee	346825	794250	In forest, Glen Tanar
			River Muick	Tributary of the River Dee	336700	794900	Confluence of the River Muick with the mainstem of the River Dee	329950	784150	Outflow from Loch Muick
			River Gairn	Tributary of the River Dee	335300	796600	Confluence of the GairnRiver with the mainstem of the River Dee	328500	800625	Balnaan
	06/11/2012	2 (of 3)	River Dee	Mainstem of the River Dee	314900	792400	Confluence of the Clunie Water with the mainstem of the River Dee	299550	790800	Glen Dee, approximately 3.5 km upstream of White Bridge
			Clunie Water	Tributary of the River Dee	314900	792400	Confluence of the Clunie Water with the mainstem of the River Dee	313700	783350	Confluence of the Baddoch Burn with the Clunie Water
			Geldie Burn	Tributary of the River Dee	302150	788450	Confluence of the Geldie Burn with the mainstem of the River Dee	297750	787500	Open hillside
	15/11/2012	3 (of 3)	River Dee	Mainstem of the River Dee	314900	792400	Confluence of the Clunie Water with the mainstem of the River Dee	298100	793750	Confluence of the Geusachan Burn with the mainstem of the River Dee
			Ey Burn	Tributary of the River Dee	308700	789700	Confluence of the Ey Burn with the mainstem of the River Dee	310050	784950	Track bridge over the Ey Burn
			Lui Water	Tributary of the River Dee	307100	789700	Confluence of the Lui Water with the mainstem of the River Dee	304000	793300	Confluence of the Luibeg Burn with the Lui Water at Derry Lodge
Bervie Water	30/10/2012	1 (of 1)	Bervie Water	Mainstem of the Bervie Water	383600	772500	Sea at Bervie Bay	376825	780300	Glenbervie
North Esk	16/11/2012	1 (of 2)	Survey unsuccessful for operational reasons
	21/11/2012	2 (of 2)	River North Esk	Mainstem of the River North Esk	374000	762550	Estuary, Montrose Bay	344900	780250	Confluence of the Water of Lee and the Water of Mark with the mainstem of the River North Esk
			Luther Water	Tributary of the River North Esk	365975	766325	Confluence of the Luther Water with the mainstem of the River North Esk	368075	769400	Wakefield
			Cruick Water	Tributary of the River North Esk	362750	765850	Confluence of the Cruick Water with the mainstem of the River North Esk	351200	762550	Milton of Balhall
			West Water	Tributary of the River North Esk	362400	766000	Confluence of the West Water with the mainstem of the River North Esk	341350	774800	Open hillside
			Water of Tarf	Tributary of the River North Esk	349450	779075	Confluence of the Water of Tarf with the mainstem of the River North Esk	348900	782150	Footbridge
			Water of Lee	Tributary of the River North Esk	344900	780250	Confluence of the Water of Lee with the mainstem of the River North Esk	340000	779775	Upstream of Loch Lee
			Water of Mark	Tributary of the River North Esk	344900	780250	Confluence of the River Mark with the mainstem of the River North Esk	340325	783600	Upstream of confluence with the Burn of Doune
South Esk	30/10/2012	1 (of 2)	River South Esk	Mainstem of the River South Esk	373225	757000	Sea at Scurdie Ness	328600	775600	Braedownie
			White Water	Tributary of the River South Esk	328575	775625	Confluence of the White Water with the mainstem of the River South Esk	328100	775950	Acharn
			Prosen Water	Tributary of the River South Esk	340850	758300	Confluence of the Prosen Water with the mainstem of the River South Esk	324900	770700	Kilbo
			Quharity Burn	Tributary of the River South Esk	342150	757925	Confluence of the Quharity Burn with the mainstem of the River South Esk	335375	758425	Bridgend of Balloch
	15/11/2012	2 (of 2)	Rottal Burn	Tributary of the River South Esk	336450	768750	Confluence of the Rottal Burn with the mainstem of the River South Esk	337625	770150	Confluence of the Burn of Heughs and the Kennel Burn with the Rottal Burn
Lunan Water	30/10/2012	1 (of 1)	Lunan Water	Mainstem of the Lunan Water	369000	751075	Sea at Lunan Bay	359150	749850	Friockheim
Tay	23/11/2012	1 (of 1)	River Tay	Mainstem of the River Tay	310400	726550	Scone Park, Perth	277150	745550	Kenmore
			River Tummel	Tributary of the River Tay	297800	751100	Confluence of the River Tummel with the mainstem of the River Tay	293550	757800	Pitlochry Dam

( b) Summary of aerial survey coverage in 2013

					Downstream limit of aerial survey			Upstream limit of aerial survey
Catchment	Date surveyed	Survey number	River name	Description	Easting	Northing	Description	Easting	Northing	Description
Dee	04/11/2013	1 (of 2)	River Dee	Mainstem of the River Dee	392500	803050	Upstream of Bridge of Dee, Aberdeen	326200	794900	B976 road bridge at Crathie, just downstream of Balmoral Castle 1
			Water of Tanar	Tributary of the River Dee	351350	797550	Confluence of the Water of Tanar with the mainstem of the River Dee	346825	794250	In forest, Glen Tanar
	07/11/2013	2 (of 2)	River Dee	Mainstem of the River Dee	324800	794650	Just upstream of Balmoral Castle 1	298100	793650	Confluence of the Geusachan Burn with the mainstem of the River Dee
			River Muick	Tributary of the River Dee	336700	794900	Confluence of the River Muick with the mainstem of the River Dee	329950	784150	Outflow from Loch Muick
			Girnock Burn	Tributary of the River Dee	333150	796300	Confluence of the Girnock Burn with the mainstem of the River Dee	331850	793250	Edge of forest, Glen Girnock
			Clunie Water	Tributary of the River Dee	314900	792400	Confluence of the Clunie Water with the mainstem of the River Dee	313700	783350	Confluence of the Baddoch Burn with the Clunie Water
			Ey Burn	Tributary of the River Dee	308700	789700	Confluence of the Ey Burn with the mainstem of the River Dee	310050	784950	Track bridge over the Ey Burn
			Luibeg Burn	Tributary of the Lui Water (itself a tributary of the River Dee)	304000	793300	Confluence of the Luibeg Burn with the Lui Water at Derry Lodge	301550	793875	Glen Luibeg
			Lui Water	Tributary of the River Dee	307100	789700	Confluence of the Lui Water with the mainstem of the River Dee	304000	793300	Confluence of the Luibeg Burn with the Lui Water at Derry Lodge
			River Gairn	Tributary of the River Dee	335300	796600	Confluence of the River Gairn with the mainstem of the River Dee	328500	800625	Balnaan
South Esk	04/11/2013	1 (of 1)	River South Esk	Mainstem of the River South Esk	373225	757000	Sea at Scurdie Ness	328500	776000	Upstream of Braedownie
			White Water	Tributary of the River South Esk	328575	775625	Confluence of the White Water with the mainstem of the River South Esk	327000	776000	In forest, Glen Doll
			Prosen Water	Tributary of the River South Esk	340850	758300	Confluence of the Prosen Water with the mainstem of the River South Esk	324900	770700	Kilbo
			Quharity Burn	Tributary of the River South Esk	342150	757925	Confluence of the Quharity Burn with the mainstem of the River South Esk	335375	758425	Bridgend of Balloch

( c) Summary of aerial survey coverage in 2014

					Downstream limit of aerial survey			Upstream limit of aerial survey
Catchment	Date surveyed	Survey number	River name	Description	Easting	Northing	Description	Easting	Northing	Description
South Esk	04/11/2014	1 (of 1)	River South Esk	Mainstem of the River South Esk	373225	757000	Sea at Scurdie Ness	328100	777575	Moulzie
			White Water	Tributary of the River South Esk	328575	775625	Confluence of the White Water with the mainstem of the River South Esk	326450	775850	Spot height 325 m, Glen Doll
			Prosen Water	Tributary of the River South Esk	340850	758300	Confluence of the Prosen Water with the mainstem of the River South Esk	325950	769675	Confluence of the Burn of Louie with the Prosen Water
			Quharity Burn	Tributary of the River South Esk	342150	757925	Confluence of the Quharity Burn with the mainstem of the River South Esk	335375	758425	Bridgend of Balloch
			Rottal Burn	Tributary of the River South Esk	336450	768750	Confluence of the Rottal Burn with the mainstem of the River South Esk	337625	770150	Confluence of the Burn of Heughs and the Kennel Burn with the Rottal Burn

Table 2.2.1.3: Summary of the tracking methods employed on rivers in the east of Scotland during the three years of the investigation and the numbers of tags detected in rivers by fixed receiver stations (where they existed) and aerial surveys (where they were carried out) according to year of study and source of tagged fish.

( a) 2012 (Tagged fish sourced from coastal nets)

		Fixed receiver stations			Aerial surveys
River	Total no. of tags detected in-river	No. deployed	No. of tags available for detection	No. of tags detected (% of no. of tags available for detection)	Date(s)	No. of tags in-river at the time	No. of tags detected (% of no. of tags in-river at the time)	Manual tracking?
Spey	Not Monitored
Don	2	0	N/A	N/A	15/11/2012	2	2 (100%)	Yes
Dee	7	2	7	0 b (0%)	02/11/2012, 06/11/2012 & 15/11/2012	7	7 (100%)	Yes
Bervie Water	0	0	N/A	N/A	30/10/2012	0	N/A	No
North Esk	17 a	2	17	14 c (82%)	16/11/2012 & 21/11/2012	6 d	6 (100%)	Yes
South Esk	18 a	15	18	18 (100%)	30/10/2012 & 15/11/2012	11 e	11 (100%)	Yes
Lunan Water	0	0	N/A	N/A	30/10/2012	0	N/A	No
Tay	5	1	5	5 (100%)	23/11/2012	5	1 f (20%)	No
Tweed	Not monitored

a Includes 1 which was detected initially in the North Esk which subsequently returned to sea and entered the South Esk. b Due to operational reasons.
c The 3 not detected include 2 which were caught by the in-river net and coble fishery and 1 which was detected only by a mobile box.
d Of the 17 detected in the North Esk, 7 returned to sea and 4 were caught by the in-river net and coble fishery prior to aerial surveys taking place.
e Of the 18 detected in the South Esk, 6 returned to sea and 1 regurgitated tag was recovered prior to aerial surveys taking place.
f Thought to be due to a relatively small proportion of the Tay catchment being surveyed.

( b) 2013 (Tagged fish sourced from coastal nets)

		Fixed receiver stations			Aerial surveys
River	Total no. of tags detected in-river	No. deployed	No. of tags available for detection	No. of tags detected (% of no. of tags available for detection)	Date(s)	No. of tags in-river at the time	No. of tags detected (% of no. of tags in-river at the time)	Manual tracking?
Spey	2	1	2	2 (100 %)	No aerial survey carried out			No
Don	0	1	0	N/A	No aerial survey carried out			No
Dee	0	1	0	N/A	04/11/2013 & 07/11/2013	0	N/A	No
Bervie Water	Not monitored
North Esk	7 a,b	1	7	7 (100 %)	No aerial survey carried out			No
South Esk	5 b	18	5	5 (100%)	04/11/2013	3 c	3 (100%)	Yes
Lunan Water	0	1	0	N/A	No aerial survey carried out			No
Tay	2 a	1	2	2 (100%)	No aerial survey carried out			No
Tweed	0	1	0	N/A	No aerial survey carried out			No

a Includes 1 which was detected initially in the North Esk which subsequently returned to sea and entered the Tay.
b Includes 1 which was detected initially in the South Esk which subsequently returned to sea and entered the North Esk. This fish returned to sea again and re-entered the South Esk, before returning to sea again (ultimate fate unknown).
c Of the 5 detected in the South Esk, 1 returned to sea and 1 regurgitated tag was recovered prior to aerial surveys taking place.

( c) 2013 (Tagged fish sourced from in-river net)*
* Since none of the tagged fish sourced from the in-river net in 2013 were detected in any other river, the following table has been collapsed significantly for clarity of presentation.

		Fixed receiver stations			Aerial surveys
River	Total no. of tags detected in-river	No. deployed	No. of tags available for detection	No. of tags detected (% of no. of tags available for detection)	Date(s)	No. of tags in-river at the time	No. of tags detected (% of no. of tags in-river at the time)	Manual tracking?
South Esk	22	18	22	19 a (86 %)	04/11/2013	10 b	9 c (90 %)	Yes

a The 3 not detected include 2 regurgitated tags which were confirmed as being no longer associated with live fish close to the in-river tagging site and 1 which was lost (suspected removal by an angler).
b Of the 22 fish tagged in the South Esk, 6 returned to sea and 6 regurgitated tags were recovered prior to aerial surveys taking place.
c The 1 not detected was lost (suspected removal by an angler).

( d) 2014 (Tagged fish sourced from in-river net and rods)

		Fixed receiver stations			Aerial surveys
River	Total no. of tags detected in-river	No. deployed	No. of tags available for detection	No. of tags detected (% of no. of tags available for detection)	Date(s)	No. of tags in-river at the time	No. of tags detected (% of no. of tags in-river at the time)	Manual tracking?
Spey	Not monitored
Don	Not monitored
Dee	Not monitored
Bervie Waters	Not monitored
North Esk	1a	1	1	1 (100%)	No aerial survey carried out			No
South Esk	32 a	18	32	30 b (94%)	04/11/2014	23c	18 b (78%)	Yes
Lunan Water	Not monitored
Tay	Not monitored
Tweed	Not monitored

a Includes 1 which was tagged and detected initially in the South Esk which subsequently returned to sea and entered the North Esk. This fish returned to sea again and re-entered the South Esk.
b The 2 not detected include 1 which was never detected again after tagging (suspected tag malfunction) and 1 regurgitated tag (confirmed as being no longer associated with a live fish between the fixed receiver stations upstream and downstream of the in-river tagging site).
c Of the 32 fish tagged in the South Esk, 6 returned to sea, 2 regurgitated tags were recovered and 1 was caught by the in-river rod fishery prior to aerial surveys taking place.
d The 5 not detected include 2 which were subsequently confirmed as live fish, 1 regurgitated tag (confirmed as being no longer associated with a live fish between the fixed receiver stations upstream and downstream of the in-river tagging site), 1 which was lost (suspected removal by an angler) and 1 which was never detected again after tagging (suspected tag malfunction).

Fixed receiver stations detected, decoded and logged (with date and time stamp) the identity of any tagged fish that moved within range. Each fixed receiver station consisted of a watertight metal container which housed a rechargeable battery (Yuasa M26-80 Marine Battery, 12 V, 80 Ah, 450 A) connected to a combined radio receiver / data logger box (Lotek SRX-DL), in turn connected to an external radio antenna (Lotek rigid 3-element Yagi) ( Figure 2.2.1.5). Deployment locations were chosen based on the need to maximise spatial resolution of the resulting data (for example by not having long stretches of river that were without coverage) while also providing coverage for specific points of interest (for example points of entry into major tributaries on the River South Esk). Practicalities were also considered including positioning boxes as high above the river as feasibly possible to reduce the risk of flooding while also maximising antenna "line of sight". To this end, when in-situ on the river bank, antennae were attached securely to a suitable fixed point (e.g. a tree or stake) which also maximised stability.

Figure 2.2.1.5: A typical fixed receiver station in situ. Note the watertight metal container which housed a rechargeable battery connected to a combined radio receiver / data logger box, in turn connected to an external radio antenna. This particular fixed receiver station was located at Shielhill Bridge (see Figure 2.2.1.6), a disused road bridge across the River South Esk. The installation is positioned high and directly above the river with the antenna pointing directly downstream, reducing the risk of flooding while maximising antenna "line of sight". The antenna was attached securely to the bridge railings to maximise stability.

A network of fixed receiver stations was used to monitor the movements of tagged salmon in the River South Esk in each of the three years of the study ( Figure 2.2.1.6). In 2012, a total of 15 fixed receiver stations were deployed around the South Esk catchment. An additional 3 fixed receiver stations (at Brechin Castle, Gella Bridge and Lednathie, giving a total of 18) were added to the network in 2013 and 2014 in order to break up some relatively long stretches of river that were without coverage in 2012 ( Figure 2.2.1.6).

Figure 2.2.1.6: Map showing the locations of the 18 fixed receiver stations deployed around the South Esk catchment. Note that those at Brechin Castle, Gella Bridge and Lednathie were not deployed in the first year of the study (2012).

Aerial surveys were carried out from a helicopter employing a combined mobile radio receiver / data logger box (Lotek SRX-600) with a skid mounted antenna (Lotek rigid 4-element Yagi) and dashboard mounted GPS antenna. The helicopter was flown at low air speed (up to 60 knots) at an altitude of approximately 150 m over the survey area and the receiver was monitored for the identity of any tagged fish. The positions of tag transmissions and the limits of the survey carried out on each river were also recorded on Ordnance Survey 1:25,000 scale paper maps. Aerial surveys meant that large areas could be covered rapidly, although the topography of each river meant that not all areas, particularly small tributaries, could be accessed. Throughout the study, aerial surveys were generally carried out around the period when spawning activity was expected, in practice from 30 October to 23 November. In particular, aerial surveys of the River South Esk were carried out on 30 October and 15 November in 2012, and on 04 November in both 2013 and 2014.

Manual tracking of tagged fish was carried out on foot or by vehicle using a combined mobile radio receiver / data logger box (Lotek SRX-600) with a hand-held antenna (Lotek rigid 3-element Yagi). Manual tracking was carried out on an ad-hoc basis around the South Esk catchment throughout the year in all three years of the investigation ( Table 2.2.1.3). Additional, and more intensive, manual tracking was carried out after aerial surveys had located the positions of the tags in order to verify whether or not a given tag was still associated with a live fish. This was done for all tags in the River South Esk in all three years of the investigation (except those tags which were still actively moving into and out of range of the fixed receiver stations). Additionally, this process was carried out in the Rivers North Esk, Dee and Don in 2012 ( Table 2.2.1.3).

Manual tracking was, in most cases, used successfully to determine the fate of the associated salmon (e.g. alive, dead, tag regurgitated). The verification process was carried out by first locating the tag to within a few metres, either in the water, or on the bank. Tags located on the bank were retrieved when possible. A tag in an active fish would move when approached. Those tags that did not move were retrieved when possible. Manual tracking was very labour intensive, and did not allow large stretches of river to be covered quickly. This method was used in conjunction with fixed receiver stations and aerial surveys, to determine the fate of tagged fish.

2.2.2 Genetic stock assignment

Obtaining baseline samples for genetic analysis

Fin clips were taken from juvenile Atlantic salmon obtained by electrofishing from 14 sites within the River South Esk ( n = 333 fish) and from 12 sites within the River North Esk ( n = 232 fish) ( Figure 2.2.2.1).

Figure 2.2.2.1: Sampling sites in the Rivers South and North Esk from which juvenile Atlantic salmon were obtained. Colour codes are according to clusters identified by multidimensional scaling analysis based on pairwise D _A (see Figure 2.3.2.1). The three outliers (see text) are shown in black. The line represents the upper and lower regions defined for subsequent assignment analysis.

SNP screening

Genomic DNA was extracted using a Blood and Tissue DNeasy purification kit (Qiagen) following the manufacturers' protocols. Extracted DNA was quantified by fluorometry (Qubit, Life Technologies) and diluted to a concentration of 50 ng / µl in a 10 mM Tris-HCl (pH 8.0), 1 mM EDTA buffer.

To avoid potential bias in the analysis due to family effects, relatedness analysis was carried out within each site to determine the presence of full siblings, based on a pedigree likelihood approach in the programme COLONY (Jones & Wang, 2010). This was done based on SNP data from a panel from Karlsson et al. (2011) and 36 random SNPs selected from GenBank (Benson et al., 2005), obtained using a 96 x 96 Fluidigm EP1 SNP genotyping platform (Fluidigm Europe b.v., Netherlands) following the manufacturers' protocols and workflow. For each identified full sib family, all but one individual from that family was removed from the data set.

Following removal of full sibs, each sample was genotyped at 5,568 SNPs, using a modified version of a custom-designed Illumina® iSelect SNP-array (Lien et al., 2011; Bourret et al., 2013; Johnston et al., 2013) at the Centre for Integrative Genetics ( CIGENE), Norway. The procedures are outlined in Johnston et al. (2013). Only normal diploid polymorphic SNPs and SNPs existing on a single paralogue (MSV3) were retained in the analysis - other variants (Gidskehaug et al., 2011) were removed. SNPs with an overall scoring percentage of less than 95 % were removed.

Genetic structuring analysis

Genetic structuring was examined by calculating pairwise F _ST and D _A (Nei, 1972) in GENALEX (Peakall & Smouse, 2012). Pairwise D _A values were used as a basis for multidimensional scaling ( MDS) analysis in R (R Core Team, 2013). A neighbour - joining tree ( NJ) was produced with the programme MEGA6 (Tamura et al., 2013), again based on pairwise D _A. Assignment regions were defined by an examination of the genetic structuring, together with the geographic position of the sites.

Individual assignment analysis

Genetic assignment regions were defined as described above, and accuracy of assignments to these regions was then examined using individual assignment ( IA) analysis. To avoid ascertainment bias when assessing assignment accuracy to region (Anderson, 2010; Waples, 2010), four training sets ( TSs) were created whereby one different site from each identified region was removed in both rivers. The removed sites were then classified as blind sites and formed the hold-out set (HOS). Each TS was used to rank the SNPs according to their ability to separate between the regions and assignment accuracy tested using the TS as the genetic baseline and assigning the HOS to this baseline. Ranking of the SNPs was carried out by estimating hierarchical F _ST for each SNP using HIERFSTAT (Goudet, 2005).

Individual assignment of the HOS was carried out in GENECLASS (Piry et al., 2004), using a Monte-Carlo resampling approach (Rannala & Mountain, 1997). Assignments were performed using the top 12, 24, 48, 96, 192, 288 and 480 ranked SNPs and a likelihood cut-off of 80 was applied. The likelihood score reflects the best match of the individual fish to a baseline assignment unit. The most likely site of origin for a fish from the available baseline sites is thus the one with the highest likelihood score. Some fish were difficult to assign in the simulations and have low scores for all baseline sites and so a likelihood cut-off is applied that classifies fish as assigned only if they have a score above this value. This exclusion method is similar to that used by Ikediashi et al. (2012). The value of 80 was chosen to give an acceptable balance between the number of fish assigned and the accuracy of assignment. Fish with an assignment likelihood score below this value were considered to be unassigned.

Where the assignment regions were observed to span across both rivers, further IA analysis was carried out to separate the two rivers within each of these identified assignment regions. Within each assignment region, SNPs were again ranked according to their ability to separate between the two rivers, and again assignment accuracy assessed using an increasing number of SNPs. Accuracy was determined based on the same TS / HOSs as used for the regional assignments.

Finally, a hierarchical assignment analysis was performed, by which fish were first assigned to region and then to river of origin within each region. Accuracy across both levels of assignment was then examined (i.e. assignment to regions within each river), as well as overall estimates of the proportion of fish assigned.

2.3 Results

2.3.1Radio tracking

Radio tagged fish

Characteristics of tagged fish are summarised in Table 2.3.1.1. In total, 245 Atlantic salmon were radio tagged, 160 of which (65 %) were assigned as females according to the criteria of Shearer (1972). Age determination by scale reading was possible for 242 of the 245 fish that were tagged (99 %). Of these, 178 were two-sea-winter (SW2) fish (72.7 %), 63 were three-sea-winter (3SW) fish (25.7 %) and 1 was a four-sea-winter (4SW) fish (0.4 %). No age determination was possible for 3 of the fish that were tagged (1 %). The fork lengths of fish tagged ranged from 59.5 cm to 100.5 cm.

Spawning locations in the River South Esk

Of the 153 salmon tagged in the coastal net fishery during 2012, 51 fish (33.3 %) were subsequently detected. During 2013, 38 salmon were tagged in the coastal net fishery, of which 14 fish (36.8 %) were subsequently detected.

Tracking was not undertaken sufficiently intensively to test whether tagged fish had spawned successfully (e.g. Webb & Hawkins, 1989). Spawning location in this sense is taken to mean the highest upstream location that tagged fish were detected during the period over which spawning activity would be expected to have occurred (i.e. from the end of October to the end of December for the upper part of the South Esk catchment). For this assessment, identification of spawning location is restricted to fish migrating within the River South Esk, where sufficient location information is available. A manual verification was carried out on each tag within the River South Esk to determine whether or not the tag was still associated with a live fish at spawning time, either by manual tracking or by observing extensive downstream movement after spawning time (see "Tracking locations" section above).

In 2012, four tags were still associated with live fish at spawning time. Three of these fish were located within Glen Clova, including one which was detected within the Rottal Burn, and one was located within the main stem of the River South Esk at Tannadice ( Figure 2.3.1.1 a). In 2013, six tags were still associated with live fish at spawning time. Of these six fish, two were sourced from the coastal net fishery and four were sourced from the in-river net. Three of these fish were located within Glen Clova, two were located within Glen Prosen, and one was located on the main stem of the River South Esk at Tannadice ( Figure 2.3.1.1 b). In 2014, fourteen tags were still associated with live fish at spawning time. Of these fourteen fish, eight were sourced from the in-river net and six were sourced from the rods. Seven of these fish were located in the upper reaches of Glen Clova, three were located in the lower reaches of Glen Clova, and two were located within Glen Prosen. Two fish were located on the main stem of the River South Esk, one at Tannadice and one downstream of the Aqueduct ( Figure 2.3.1.1 c).

Thus, over the course of the three-year investigation, a total of 24 spring salmon were tracked to spawning in the River South Esk ( Figure 2.3.1.1 d). The majority (20 fish, 83.3 %) spawned in the upper reaches of the River South Esk (Figure 2.3.1.1 d), with 16 fish (66.7 %) spawning in Glen Clova and 4 fish (16.7 %) spawning in Glen Prosen. The remaining 4 fish (16.7 %) spawned in the middle reaches of the River South Esk ( Figure 2.3.1.1 d), with 3 of these fish (12.5 %) spawning in the main stem of the river at Tannadice and 1 of these fish (4.2 %) spawning in the main stem of the river downstream of the Aqueduct. These data are summarised in Table 2.3.1.2.

Table 2.3.1.1: Biological characteristics of spring salmon tagged according to year of the investigation and method of capture.

( a) 2012 coastal nets
Sex	Category	Male			Female 1
	Number (%)	54 (35.3)			99 (64.7)
Age	Category	2SW	3SW	4SW	2SW	3SW	4SW
	Number (%)	40 (74.1)	14 (25.9)	0 (0.0)	67 (67.7)	30 (30.3)	1 (1.0)
Fork length (cm)	Minimum	60.5	076.0	N/A	59.5	60.5	86.5
	Mean	70.7	084.1		67.8	79.0
	Maximum	82.0	100.5		83.0	88.5
( b) 2013 coastal nets
Sex	Category	Male			Female
	Number (%)	7 (18.4)			31 (81.6)
Age	Category	2SW	3SW	4SW	2SW	3SW	4SW
	Number (%)	6 (85.7)	1 (14.3)	0 (0.0)	25 (80.6)	6 (19.4)	0 (0.0)
Fork length (cm)	Minimum	66.0	81.5	N/A	64.5	75.0	N/A
	Mean	72.9			70.5	79.1
	Maximum	79.5			77.0	88.5
( c) 2013 in-river net
Sex	Category	Male			Female 2
	Number (%)	8 (36.4)			14 (63.6)
Age	Category	2SW	3SW	4SW	2SW	3SW	4SW
	Number (%)	6 (75.0)	2 (25.0)	0 (0.0)	13 (92.9)	0 (0.0)	0 (0.0)
Fork length (cm)	Minimum	63.0	81.5	N/A	62.0	N/A	N/A
	Mean	68.6	82.8		66.3
	Maximum	76.5	84.0		72.0
( d) 2014 in-river net
Sex	Category	Male			Female 3
	Number (%)	12 (60.0)			8 (40.0)
Age	Category	2SW	3SW	4SW	2SW	3SW	4SW
	Number (%)	8 (66.7)	4 (33.3)	0 (0.0)	6 (75.0)	1 (12.5)	0 (0.0)
Fork length (cm)	Minimum	60.5	75.5	N/A	66.5	77.5	N/A
	Mean	68.9	79.8		69.3
	Maximum	82.0	84.0		74.0
( e) 2014 Rods
Sex	Category	Male			Female
	Number (%)	4 (33.3)			8 (66.7)
Age	Category	2SW	3SW	4SW	2SW	3SW	4SW
	Number (%)	2 (50.0)	2 (50.0)	0 (0.0)	5 (62.5)	3 (37.5)	0 (0.0)
Fork length (cm)	Minimum	64.0	73.0	N/A	60.0	66.3	N/A
	Mean	67.5	73.8		67.1	73.1
	Maximum	71.0	74.5		74.0	79.0

^{1, 2, 3} Age determination by scale reading was not possible for 1 fish in each of these 3 categories.

Figure 2.3.1.1 a

Figure 2.3.1.1 b

Figure 2.3.1.1 c

Figure 2.3.1.1 d

Table 2.3.1.2: Summary of assumed spawning locations for tagged fish within the River South Esk during the three years of the investigation.

		Upper catchment			Lower catchment
Year	Number of tagged fish tracked to spawning	Number of fish in Glen Clova (%)	Number of fish in Glen Prosen (%)	Total (%)	Number of fish in the main stem (%)	Total (%)
2012	04	03 (75.0)	00 (0.0)	0 3 (75.0)	1 (25.0)	1 (25.0)
2013	06	03 (50.0)	2 (33.3)	0 5 (83.3)	1 (16.7)	1 (16.7)
2014	14	10 (71.4)	2 (14.3)	12 (85.7)	2 (14.3)	2 (14.3)
Total	24	16 (66.7)	4 (16.7)	20 (83.3)	4 (16.7)	4 (16.7)

2.3.2 Genetic stock assignment

Four full sib families were identified, the largest consisting of three individuals. As such, five fish were removed from further analysis, reducing the total number of fish from 565 to 560.

From the 5,568 SNPs screened, 1,143 were removed as they were not classed as a SNP or MSV3. Furthermore, for 22 SNPs, scoring percentage was lower than 95 %. This resulted in a final panel of 4,403 SNPs for use in subsequent analysis.

Genetic structuring

The pairwise F _ST values ranged from 0.011 (S_Pow Burn and S_Prosen_Br) to 0.027 (N_LowEdz and N_Lee), with a mean (± S.D.) of 0.015 ± 0.003. Similar values were observed for D _A (mean ± S.D. = 0.015 ± 0.003), with a minimum of 0.010 between S_Pow Burn and S_Prosen_Br and a maximum value of 0.026 between N_LowEdz and N_Lee.

The MDS analysis showed three outlier sites (N_Stracathro, N_Lee and N_LowEdz) and a central cluster ( Figure 2.3.2.1). When examining this central cluster in more detail, a split into two groups could be seen, which, with the exception of three sites (S_Braedownie, S_Cortachy and S_Quharity), corresponded to upper and lower regions in both rivers ( Figure 2.3.2.1). Similarly, the NJ tree also showed a split, with the majority of sites from the upper reaches of the rivers clustered into one branch of the tree, with the exception of N_Westmouth, whilst all the lower sites formed the other branches, apart from S_Prosen_L ( Figure 2.3.2.2). Based on a combination of both analyses, two assignment regions (upper and lower) were defined, and all sites classed as belonging to one of them, according to a dividing line shown in Figure 2.2.2.1.

Figure 2.3.2.1: MDS plot of sample sites from the rivers North and South Esk. Circles represent the two identified regions, upper (blue) and lower (green).

Figure 2.3.2.2: Neighbour - Joining Tree of sample sites. The line indicates a separation between mostly upper and lower sites.
Individual assignment analysis

Four TS / HOSs were created whereby two upper and two lower sites were removed, one from each river (i.e. blind sites) ( Table 2.3.2.1).

Table 2.3.2.1: Sites removed from the four training sets (TS1 - TS4), including one site from each region within each river.

Training set	Upper North Esk	Lower North Esk	Upper South Esk	Lower South Esk
TS1	N_upper	N_Cruick	S_Clova	S_Brechin
TS2	N_Tillybardine	N_LowMar	S_ProsenL	S_Justin
TS3	N_Mark	N_LowEdz	S_Braedownie	S_Pow
TS4	N_Lee	N_Stracathro	S_Craig	S_Cortachy

Assignment of fish from the HOS to region was assessed. Figure 2.3.2.3 shows the accuracy of the assignments to either upper or lower region in the four HOSs, for an increasing number of SNPs, as well as the number of fish assigned at a cut-off of 80 %. With the exception of the accuracy of the blind sites to the lower region (dashed green line in Figure 2.3.2.3), accuracy and numbers of fish plateaued at the top ranked 192 SNPs and remained relatively stable for higher numbers of SNPs. At 192 SNPs, on average 79.2 % of fish could be assigned to the lower region with 57.2 % accuracy, whilst 79.1 % of fish could be correctly assigned (with an accuracy of 74.4%) to the upper region.

Figure 2.3.2.3: Average assignment accuracies (solid lines) to the upper (blue) or lower (green) region and average proportion of fish assigned (dashed lines) for an increasing number of SNPs, after a cut-off of 80 % was applied. Bars represent standard errors calculated over all replicate TS / HOS sets.

Between river differentiation

Within the upper region

Assignment success to the different rivers within the upper region is detailed in Figure 2.3.2.4. Average accuracy was highest for the top 12 SNPs and then gradually decreased with increasing number of SNPs, though it plateaued at the top 192 SNPs. The proportion of fish assigned increased gradually and started to plateau at the top 288 SNPs. Based on the top 12 SNPs, 55.9 % of fish were assigned to the South Esk with an accuracy of 76.3 %.

Figure 2.3.2.4: Average assignment accuracies (solid lines) to the River North Esk (grey) or River South Esk (red), and average proportion of fish assigned (dashed lines), for an increasing number of SNPs, after a cut-off of 80 % was applied. Bars represent standard errors calculated over all replicate TS / HOS sets.

Within the lower region

The assignment accuracy to river of origin within the lower region increased with increasing number of SNPs, and levelled off at the top 192 SNPs, with the exception of the accuracy to the River North Esk based on the top 480 SNPs ( Figure 2.3.2.5). The proportion of fish assigned increased gradually and stabilised after the top 192 SNPs. Using the top 480 SNPs, 83.1 % of fish could be assigned to the River South Esk with an accuracy of 65.5 %.

Figure 2.3.2.5: Average assignment accuracies (solid lines) to the River North Esk (grey) or River South Esk (red), and average proportion of fish assigned (dashed lines) for an increasing number of SNPs, after a cut-off of 80 % was applied. Bars represent standard errors calculated over all replicate TS / HOS sets.

Hierarchical assignment

Fish from the blind sites were firstly assigned to region of origin (upper / lower) using the top 192 SNPs identified in the upper / lower analysis, and subsequently to river of origin within each region, based on the top 12 SNPs and top 480 SNPs identified within the upper and lower region analysis, respectively. After applying an assignment likelihood cut-off of 80, accuracy to the upper and lower South Esk was 22.7 % and 44.9 %, respectively, and an average of 15.7 % of fish were assigned ( Table 2.3.2.2).

Table 2.3.2.2: Summary of the hierarchical assignment analysis (cut-off of 80 applied), where fish were first assigned to region and then to river of origin within the upper and lower region, resulting in three assignment classes. Proportion assigned (%) to original sample size ( n = 329) and accuracy (%) to each of the three classes are shown in italics. Figures in bold are correctly assigned fish on which the accuracy calculation is based.

Fish origin	Assigned region / river
	Upper South Esk	Lower South Esk	North Esk	Proportion assigned (%)
Upper South Esk	5	7	34	14.0
Lower South Esk	10	31	16	17.3
North Esk	7	31	44	24.9
Accuracy (%)	22.7	44.9	46.8
Average accuracy (%) to South Esk regions	39.6

2.4 Discussion

Levels of genetic differentiation between sites and areas (upper / lower, main stem / tributaries) were examined, as well as assignment success to site and / or areas of the system. Levels of differentiation between sites and areas were found to be relatively low within the system, and accurate assignments to site / area were thus not possible. The genetic approach employed was therefore not a useful approach in this situation and interpretation of spawning locations is based exclusively on the results of radio tracking work.

Spawning locations in the River South Esk

Over the course of the three-year investigation, a total of 24 spring salmon were tracked to spawning in the River South Esk. The majority (83.3 %) spawned in the upper reaches of the River South Esk, with 66.7 % spawning in Glen Clova and 16.7 % spawning in Glen Prosen. It is therefore concluded that River South Esk spring salmon spawn predominantly in the upper reaches of the catchment. This finding concurs with that observed previously on the Rivers Dee (Hawkins & Smith, 1986; Laughton & Smith, 1992) and Spey (Laughton, 1989; 1991; Laughton & Smith, 1992). It was interesting to note, however, that as on the River Tweed (Campbell, 1998), spring salmon in the present study did not spawn exclusively in the upper catchment, as defined by changes in geography (see Section 1.2). Indeed, 16.7 % spawned in the middle reaches of the River South Esk on the main stem. Therefore, distance upstream is unlikely to be the only feature defining the spawning location of spring fish, and other factors such as the relative availability of spawning habitat, are also likely to be important.

Of the tagged fish that stayed in the River South Esk, the proportions of fish surviving and retaining their tags until spawning time ranged from 25.0 % to 75.0 % among years and sources of tagged fish. These proportions are broadly similar to the range seen in previous in-river tagging studies. In a radio tracking study of 10 Atlantic salmon tagged near the mouth of the Aberdeenshire Dee, 2 fish (20 %) were monitored for most of the time they spent in fresh water (Hawkins & Smith, 1986). In a similar study carried out on the River Spey over two consecutive years (1988 and 1989), Laughton (1991) reported that 20 % and 29 % of tagged fish were subsequently tracked through to spawning in 1988 and 1989 respectively. In another study (Laughton, 1989), 10 fish out of a total of 24 that were tagged (42 %) were successfully tracked to spawning. The fish in that study (a mixture of salmon and grilse) were tagged between April and August, and it appears that the higher rate of successful tracking to spawning than in the present study reflects the shorter time period between tagging and spawning experienced by fish tagged later in the study period. Of the three seasonal groups of fish considered in the study by Laughton (1989), only the first group consisting of 10 salmon released in late April and early May are directly comparable with the present study. Of these, just 1 fish (10 %) was successfully tracked to spawning. In the second seasonal group of fish (three salmon and one grilse released during low water in June), 2 fish (50 %) were successfully tracked to spawning (Laughton, 1989). In the third seasonal group of fish (ten grilse released in August), 7 fish (70 %) were successfully tracked to spawning (Laughton, 1989). Webb (1989) reported that of 44 salmon radio tagged in the River Tay estuary during June and July, 5 tags (11.4 %) were detected during an aerial survey carried out at the end of September. Tag retention could be improved by surgically implanting tags into the body cavities of salmon. However, there is no information with which to assess how such intervention would itself affect mortality risk. It was decided in this study to minimise risk to the fish by using intragastric tagging and accept a possible higher cost due to regurgitation of tags.

Detection rates of fish post-tagging

In the present study, approximately one third of the radio tagged salmon sourced from the coastal net fishery were detected again after tagging (51 out of 153 tagged in 2012, 33.3 %; 14 out of 38 tagged in 2013, 36.8 %). Of the 153 fish tagged in 2012, 3 fish (2.0 %) were re-caught by the coastal net fishery. Of the 48 fish that were detected in the monitored rivers, 4 fish (2.6 % of all fish tagged) were caught by the net and coble fishery in the River North Esk. No tagged fish were reported as being taken by the rod and line fishery in 2012, while in 2013, no tagged fish sourced from the coastal net fishery were recaptured at sea or taken by fisheries.

The overall detection rate of tagged fish sourced from the coastal net fishery after release is at the lower end of the ranges reported by previous studies. In a study of Atlantic salmon radio tagged at sea south of the Aberdeenshire Dee between May and August, Smith & Johnstone (1996) detected 51 % of the tagged fish after release. Monthly detection rates varied from 30 % to 67 %. However, in contrast to the current study, the proportion of detections made by fisheries in Smith & Johnstone (1996) was greater; 4.9 % were caught by coastal fisheries and 11.5 % were caught by rod and line (1.6 % returned). In the first two years of the present study, no external marking of tagged fish to indicate the presence of an internal radio tag was carried out. Thus, only tags extracted from killed, gutted fish are likely to be reported by anglers. The lack of reported tags from rod caught fish in the current study compared to Smith & Johnstone (1996) may be due to the increased rates of catch and release in the rod fishery. In 2014, just one tag was returned by an angler. Given the high incidence of catch and release for salmon in Scotland [80.0 % in 2013 ( ICES, 2014)], detection rates for tagged fish sourced from the coastal net fishery by anglers is likely to have been low in the present study.

A further difference between the present study and previous studies which may have affected the detection rate of tagged fish after release is the time of year at which fish were tagged. Those fish tagged in the present study were spring fish, tagged at sea between February and May. In the study by Smith et al. (1994), the tagging period was longer, running from February to August, while those tagged by Smith & Johnstone (1996) were tagged later in the year (May: 52.46 %; June: 14.75 %; and August: 32.79%). Differences in fish behaviour and angling practices among different stock components (spring salmon, summer salmon and grilse) and varying environmental factors in the period following release may have affected the detection rate of tagged fish after release.

The fate of radio tagged fish

For the study as a whole, 24 fish out of 245 that were tagged (9.8 %) were tracked to spawning. For tagged fish sourced from the coastal net fishery, 6 fish out of 191 that were tagged over two years (3.1 %) were tracked to spawning. In contrast, for tagged fish sourced from within the River South Esk itself, 18 fish out of 54 that were tagged over two years (33.3 %) were tracked to spawning. Thus, the coastal net fishery accounted for one quarter of all tagged fish that were tracked to spawning, with the remaining three quarters all sourced in-river. Furthermore, "survival" rates from tagging to spawning were an order of magnitude greater for fish tagged in-river than for fish sourced from the coastal nets as significant numbers of tagged fish were either lost in the marine phase after tagging or returned to rivers that were not monitored with receivers.

There are several reasons for failing to subsequently detect fish tagged and released from coastal fisheries. Firstly, it is likely that a proportion of the tagged fish would have regurgitated their tag. In a study of Atlantic salmon radio tagged in the River Tweed estuary between April and September, Smith et al. (1998) recorded an in-river regurgitation rate (where fish are not expected to be feeding) ranging from 12.5 % to 16.7 %. At sea, where the fish in the present study were tagged, regurgitation rate is likely to be higher as feeding may still be occurring to some extent. Support for this comes from the observation that one of the tagged fish caught in nets during the present study showed evidence of post-tagging feeding in the gut. Furthermore, Mills (1989) cites data from Grey & Tosh (1894) which shows that the proportion of salmon caught in the estuary of the River Tweed with food in their stomachs is highest during spring (February: 14.3 %; March: 43.5 %; April: 39.8 %; May: 16.7 %; June: 13.1 %; July: 1.8 %; August: 3.8 %; and September: 1.9 %). Thus, Grey & Tosh (1894) showed that the greatest proportion of fish with food in their stomachs occurred in the months during which tagging was carried out in the present study.

Secondly, it is likely that predation accounted for the loss of some tagged fish sourced from the coastal net fishery. Middlemas et al. (2003) assessed the influence of marine mammal predators on salmon. Four species known to be predators of Atlantic salmon inhabit the waters off the east coast of Scotland - grey seals ( Halichoerus grypus), harbour seals ( Phoca vitulina), harbour porpoises ( Phocoena phocoena) and bottlenose dolphins ( Tursiops truncatus). All four species are present in the area in which the present study was carried out. Harbour seals are regularly seen in the Montrose Basin, and grey seals are occasionally seen (Scottish Wildlife Trust, 2011). Bottlenose dolphins have also been observed foraging in the Montrose Bay area (Dempsey, 2009), as have harbour porpoises (Marine Life Angus, 2013).

Thirdly, it is possible that some of the salmon tagged in the coastal net fishery were destined for other rivers that were not being monitored for tags. Fisheries operating along migration routes of salmon as they return through coastal waters to spawn may exploit fish from a number of different rivers (Gilbey et al., 2012). Therefore, salmon caught and tagged at a particular coastal location may be returning to a variety of rivers, possibly a considerable distance from the point of tagging (e.g. Shearer, 1986; 1992; Smith & Johnstone, 1996; Malcolm et al., 2010). In the present study, a proportion of the tagged fish sourced from the coastal net fishery were detected in rivers other than the South Esk that were also being monitored for tags (see Section 5), and it is possible that some tagged fish returned to rivers elsewhere.

Efficacy of radio tracking

Due to operational reasons, the fixed receiver stations deployed on the River Dee in 2012 did not detect any tags, although 7 tags were subsequently detected during aerial surveys. Elsewhere, and in other years, fixed receiver stations had a high tag detection efficacy, ranging from 82 % to 100 %. When fixed receiver station efficacy was less than 100 %, there were plausible explanations as to why some tags were seemingly not detected, and these are discussed briefly below. In light of this additional information, efficacy figures presented should be viewed as minimum estimates.

On the River North Esk in 2012, 14 of the 17 tags available for detection (82 %) were detected by the fixed receiver stations. The 3 tags not detected include 2 tags in fish which were caught and removed by the in-river net and coble fishery. It is entirely possible that these fish were not "missed" by the fixed receiver stations as the net and coble fishery is located in the estuary, downstream of the downstream-most fixed receiver station. It seems likely that these 2 fish may have been intercepted shortly after entering the River North Esk and before they had moved sufficiently far upstream to be detected by the downstream-most fixed receiver station. However, it is also possible that these 2 fish were "missed" by the fixed receiver stations and were intercepted by the net and coble fishery as they returned to sea, having initially progressed some distance up-river beyond the fixed receiver stations. Although we have no way of verifying which scenario is correct, it is worth noting that a further 2 fish caught and removed by the net and coble fishery in the River North Esk in 2012 were detected successfully by a fixed receiver station, and were subsequently captured by the net and coble fishery, presumably whilst attempting to return to sea. The other tag that was not detected by a fixed receiver station in the River North Esk in 2012 was detected only by a mobile receiver, and is not thought to have progressed sufficiently far up-river to have reached the downstream-most fixed receiver station. The efficacy figure of 82 % on the River North Esk in 2012 should therefore be seen as a minimum estimate as it is entirely plausible that all fish progressing sufficiently far up-river to be detected by at least one of the two fixed receiver stations were infact detected.

On the River South Esk in 2013, for those tagged fish sourced from the in-river net, 19 of the 22 tags available for detection (86 %) were detected by the fixed receiver stations. This efficacy figure should also be viewed as a minimum estimate as the 3 tags not detected are thought not to have passed any fixed receiver stations. The 3 tags not detected include 2 regurgitated tags which were confirmed as being no longer associated with live fish and which were located close to the in-river tagging site. It seems likely that these tags were regurgitated soon after the fish in question were tagged, and did not therefore have the opportunity to be detected by fixed receiver stations situated either upstream or downstream of the netting site. The other tag that was not detected by a fixed receiver station is thought to have been in a fish removed by an angler in the vicinity of the in-river tagging site.

On the River South Esk in 2014, 30 of the 32 tags available for detection (94 %) were detected by the fixed receiver stations. Similar reasons to those noted above were suspected for these "missed" tags. 1 tag that was not detected by a fixed receiver station was never detected again after tagging, and tag malfunction was suspected. 1 regurgitated tag was confirmed as being no longer associated with a live fish between the fixed receiver stations upstream and downstream of the in-river tagging site.

Aerial surveys were also generally highly efficient at detecting tags known to be in a given river at the time of the survey, i.e. after taking into account tags no longer available for detection such as those in fish known to have returned to sea, or regurgitated tags recovered from a river prior to aerial surveys taking place. The exception to this was on the River Tay in 2012, when only 1 of the 5 tags thought to be in the river at the time (20 %) were detected by the aerial survey. It is thought this was due to the relatively small proportion of the Tay catchment that was surveyed. On the River South Esk in 2013, for those tagged fish sourced from the in-river net, 9 of the 10 tags available for detection (90 %) were detected by the aerial survey. The 1 tag that was not detected was lost, and it was suspected that this tag was in a fish removed by an angler. On the River South Esk in 2014, 18 of the 23 tags available for detection (78 %) were detected by the aerial survey. Of the 5 tags that were "missed" by the aerial survey, 1 tag was lost (suspected removal by an angler) and 1 tag was never detected again after tagging (suspected tag malfunction). However, the other 3 tags were genuinely "missed" by the aerial survey, and included 1 regurgitated tag (confirmed as being no longer associated with a live fish between the fixed receiver stations upstream and downstream of the in-river tagging site) which was subsequently found (lodged deeply in the river bed) by manual tracking, and 2 tags in fish confirmed as being alive at spawning time, having proceeded upstream of the upstream-most fixed receiver station in Glen Prosen. These 2 tags were presumably outside the area covered by the aerial survey in the very highest reaches of Glen Prosen, although it is also possible that the aerial survey passed over them, but failed to detect them. (One of these fish was subsequently detected moving downstream by fixed receiver stations after the aerial survey took place).

Fixed receiver stations and aerial surveys were generally highly efficient at detecting tags. In combination, and supplemented by manual tracking, the radio tracking methodology was sufficiently robust to enable conclusions regarding the fate of individual fish to be made in the vast majority of cases for the River South Esk. This was particularly the case when dealing with fish that were tagged in-river as there was a known number of fish within the river to start with. Two fixed receiver stations were situated downstream of the in-river tagging site on the River South Esk, enabling all fish returning to sea to be identified successfully. Efficacy values for both fixed receiver stations and aerial surveys were high for tagged fish sourced in-river, being at least 86 % and 94 % for fixed receiver stations on the River South Esk in 2013 and 2014 respectively, and at least 90 % and 78 % for aerial surveys on the River South Esk in 2013 and 2014 respectively.

In the case of tagged fish sourced from the coastal net fishery, the number of fish actually entering each river was ultimately unknown, since fish tagged at sea may have regurgitated their tags prior to entering a river. It is not therefore possible to truly validate the efficacy of fixed receiver stations versus aerial surveys since the number of tags available to be detected by either method was only known from the other method. Nonetheless, there was a high degree of congruence between the number of tags detected by both methods, suggesting that the numbers detected by the two methods were infact representative of the numbers of tagged fish entering the different rivers. It is also worth considering that the mechanism of detecting tags by the two methods differs somewhat. Fixed receiver stations detect fish that are actively moving whereas aerial surveying detects tags (not necessarily associated with live fish). Manual tracking is required in combination with these approaches to provide an overall picture of fish movements over time.

2.5 Appendices

Appendix 2.5 a: The fate of all tagged fish sourced from the coastal net fishery in 2012.

^a Coastal net fishery; ^b In-river net and coble fishery.

Appendix 2.5 b: The fate of all tagged fish sourced from the coastal net fishery in 2013.

Appendix 2.5 c: The fate of all tagged fish sourced from the in-river net in 2013.

Appendix 2.5 d: The fate of all tagged fish sourced from the in-river net in 2014.

Appendix 2.5 e: The fate of all tagged fish sourced from the rods in 2014.

^a In-river rod fishery.