Rivers in assessment unit 1 (Gulf of Bothnia, SD 31)

River catches and fishery

In 2012, the catch in Tornionjoki was three times higher than in 2011 and for the first time since the beginning of the time-series with annual catch statistics, it exceeded 100 tonnes (Table 3.1.1.1).

In 2014, the catch increased to 147 tonnes, and in 2016 it reached the present record of 161 tonnes (Table 3.1.1.1). In 2017 and 2018, the catch again declined to around 90 tonnes, but in 2019 and 2020 it increased again, and was 111 and 130 tonnes, respectively. Catch levels similar to those observed in 2012–2020 were observed in the early 20th century (Figure 3.1.1.1). Salmon catches in Simojoki did not rise much in 2012–2013, which is partly due to a low fishing effort. However, in 2014 and 2015 there was a clear increase in the catch and the rising trend continued until 2016, when the catch was 1.8 tonnes (Table 3.1.1.1). As in Tornionjoki, 2017 catches dropped also in Simojoki, and they have been between 0.5–1 tonnes in 2017–2019 but increased again in 2020 to 1.5 tonnes. The catches in Kalixälven have decreased in later years mostly depending on not functional catch reporting system and they do not correspond to the registered number of salmon that have passed the fishway, totally 250 salmon were caught and out of which 100 were retained.

A special kind of fishing from boat (rod fishing by rowing) dominates the salmon fishing in Tor-nionjoki. This type of fishing also occurs in Kalixälven, but there it is not as dominating as in

Tornionjoki. CPUE of this fishery in Tornionjoki has increased tens of times since the late 1980s (Table 3.1.1.1), apparently reflecting the parallel increase in the abundance of spawners in the river. The CPUE has been high (over 1000 grams/fishing day) in 1997, 2008 and 2012–2016, when the total river catches were also peaking. In 2017 CPUE dropped to 860 g/day. In 2018, it in-creased to 1200 g/day and in 2019 and 2020 the CPUE was 970 g/day and 930 g/day, respectively.

Annual changes in CPUE and in total river catch generally follow each other. However, in 2019 and 2020 the CPUE was exceptionally low compared to the total catch.

In Råneälven, the local administration has since 2014 utilized a seasonal catch bag limit regula-tion of maximum of three salmon per person and season. Both obligatory tagging of killed fish (maximum of three tags per person and year) and a digital catch reporting system has been uti-lized to aid in enforcement. Most (80–90%) of the salmon caught with rod are released back; in 2017 a total of 56 salmon were caught, out of which 45 were released, whereas in 2018 only two salmon were caught and tagged (retained). The catch in 2019 was 45 salmon out of which seven were tagged and retained; in 2020 only two salmon were caught and retained.

Spawning runs and their composition

In Kalixälven salmon are counted in the fishway at the waterfall in Jockfall about 100 km from the river mouth. From 2007 to 2012 the mean annual run was 5500 salmon. In 2013, the run in-creased to the highest observed when more than 15 000 salmon passed the fishway. The counted runs in 2014–2019 stayed at a lower level (between 5000–10 000 salmon). In 2020, nearly 19 000 were registered in the fishcounter (Table 3.1.1.2). Yearly very few reared (adipose finclipped) salmon has been registered in the fish counter. Between 2015 to 2018 no reared salmon was reg-istered in the counter. In 2019, six reared salmon was regreg-istered of 9957 salmon, and in 2020 only one salmon with clipped adipose fin was registered of 18 664which results in very low propor-tion of strayers.

A hydroacoustic split-beam technique was employed in 2003–2007 to count the spawning run in Simojoki. It seems evident that these counts covered only a fraction of the total run, as there are irregularities in the river bottom at the counting site, allowing salmon to pass without being recorded. Since 2008, the split-beam technique has been replaced by an echosounder called DID-SON (Dual frequency IDentification DID-SONar) and in 2020 a new generation version of DIDDID-SON (called ARIS) replaced DIDSON. According to monitoring results, the seasonal run size has ranged from less than 1000 up to more than 5000 fish (Table 3.1.1.2). Spawning runs gradually increased from 2004 to 2008–2009, but again dropped in 2010–2011. In 2012, the run increased fourfold from the previous year (to about 3000) and also the runs in 2013–2015 were about as abundant (3000–4000 salmon). The 2016 run was record-high with 5400 salmon counted. In 2017, the run dropped below 2000 salmon but increased in 2018, 2019 and 2020 to about 4000 salmon/year (Table 3.1.1.2). A lot of back-and-forth movement of salmon has been detected in Simojoki, especially in 2018, which erodes the accuracy of the counts. There have also been prob-lems connected to the separation of species.

The spawning runs into Tornionjoki have also been monitored using the DIDSON technique since 2009, but in 2019 the old DIDSON units were replaced by ARIS sonars. The observed sea-sonal run size has ranged from 17 200 (year 2010) to 100 200 (year 2014) salmon (Table 3.1.1.2).

Grilse account for a minority (7–24%) of the annual spawning runs. The run size in 2016 (98 300 salmon) was almost as high as in the record year 2014 (101 000 salmon), but as in the Simojoki, the run again dropped in 2017 (to about 41 000 salmon). In 2018 the counted amount increased only slightly (to 47 000 salmon), but in 2019 and 2020 the total count increased further, to 65 500 and 69 100 salmon, respectively.

The Tornionjoki counting site is located about 100 km upstream from the river mouth. Therefore, salmon which are either caught below the site or stay to spawn below the site must be assessed

and added into the hydroacoustic count, in order to get an estimate of the total run size into the river (Lilja et al., 2010). Also, according to auxiliary studies, a small fraction of the spawners pass the counting site via the fast-flowing channel without being detected by sonars. The mid-channel seems to be utilised the more by salmon the lower the river water level is (Isometsä et al., 2021). The 2018 and 2019 counts probably represents a smaller-than-normal proportion of the total run size into the river; observations were made of unusually high amounts of salmon stay-ing on the lowermost river until autumn 2018. Moreover, the very low prevailstay-ing water level in 2018 and 2019 probably allowed many spawners to pass the hydroacoustic counter via the deep-est mid-channel where they may have remained undetected.

In 2014–2019, the spawning run in Råneälven has been monitored with an ultra-sound camera (SIMSONAR). The technique is similar to that used in Tornionjoki and Simojoki. The counting site is located about 35 km upstream from the river mouth, and the counts are expected to rep-resent the total run as almost no salmon spawning areas exist downstream. The total counted salmon runs in the period 2014–2019 has varied between 1000–4000 and in 2020 the salmon run was 2461 (Table 3.1.1.3).

Over 13 000 catch samples have been collected from the Tornionjoki salmon fishery since the mid-1970s. Table 3.1.1.3 shows sample size, sea age composition, sex composition and propor-tion of reared fish (identified either by the absence of adipose fin or by scale reading) of the data for the given time periods. Caught fish have generally become older, and the proportion of repeat spawners has increased in parallel with a decreasing sea fishing pressure (see Section 4). The strong spawning runs into Tornionjoki in 2012–2016 were a result of fish from several smolt co-horts. In these years, the proportion of females has been fairly stable, about two thirds of total biomass, but in 2018 and 2019 only about 55% of the total biomass were females. The proportion of repeat spawners has generally been between 5–10% during the last decade. However, a record high proportion of repeat spawners (14%) was observed in 2014, and the proportion was high also in 2018 and 2020 (12% and 11%, respectively). On the contrary, in 2017 and 2019 the propor-tion of repeat spawners was only 3%, indicating large interannual variapropor-tion. Very few salmon of reared origin (<1%) have been observed in the Tornionjoki catch samples in the last decade (Table 3.1.1.3).

Parr densities and smolt trapping

The lowest parr densities in AU 1 rivers were observed in the mid-1980s (Table 3.1.1.4, Figures 3.1.1.4 and 3.1.1.5). During the 1990s, densities increased in a cyclic pattern with two ‘jumps’.

The second, higher jump started in 1996–1997. Between these increases there was a collapse in densities around the mid-1990s, when also the highest M74 mortality was observed (see below).

Average parr densities are nowadays 5–60 times higher than in the mid-1980s. Since the turn of the millennium, annual parr densities have varied 2–6 fold. In Simojoki, some years with higher-than-earlier densities of 0+ parr have been observed recently, but annual variation has been large and densities of older parr have often not increased in this river after years with high 0+ densities.

In the other AU 1 rivers, however, parr densities of all ages have continued to increase rather steadily until in the mid-2010s.

In some years, like in 2003, high densities of parr hatched in Simojoki, Tornionjoki and Kalixäl-ven despite relatively low preceding river catches (indicating low spawner abundance). Simi-larly, high densities of 0+ parr were observed in Tornionjoki in 2008 and 2011, although river catches and spawners counts in the preceding years were not among the highest. Possible rea-sons for this inconsistency include exceptionally warm and low summer-time river water, which might have affected fishing success in the river and even measurements of parr densities. In years 2006, 2013, 2014, 2018 and 2019 conditions for electrofishing were favourable because of very low river water levels, whereas they were the opposite in 2004 and 2005. These kinds of

changes in electrofishing conditions may have affected the results, and one must therefore be somewhat cautious when interpreting the data obtained.

In Simojoki, the mean density of one-summer old parr increased by about 50% from 2015 to 2016 and it continued to increase in 2017 (Table 3.1.1.4). The 2019 density of 0+ parr (40.9 ind./100 sqm) is record high in the time-series, although most of the uppermost sites still lack 0+ parr. In 2020 the 0+ parr density dropped to about half (21.3 ind./100 sqm) of that of 2019, although the number of spawners giving rise to these parr densities was almost identical (Table 3.1.1.2). The density of older parr increased rapidly from 2015 (6.5 ind./100 sqm) to a record high level in 2018 (42 ind./100 sqm). In 2019, however, the density dropped to 14.4 ind./100 sqm and in 2020 the density increased to 19.9 ind./100 sqm. In Tornionjoki the densities of 0+ parr in 2014 and 2015 were clearly higher than in any earlier year in the time-series. In 2016, the average density of 0+

parr on the sampled sites was somewhat lower than in 2015. Several flood peaks due to heavy rains prevented electrofishing on the lower and on some of the middle and upper sections of the river system. In 2017, the average density of 0+ parr increased and was the third highest in the time-series (28.5 ind./100 sqm). In 2018 the mean 0+ parr density again dropped to only 18.3 ind./100 sqm, however in 2019 and 2020 the densities were higher: 25.5 and 20.5 ind./100 sqm, respectively. The average density of older parr in 2017 (17.2 ind./100 m²) dropped from the two earlier years and in 2019 a further decrease (to 15.2 ind./100 sqm) was observed, but there was again an increase in 2020 (19.8 ind./100 sqm). Thus, in Tornionjoki parr production dropped after the record years in the mid-2010s, but again a slight increase is observed during the last 1–2 years.

In Kalixälven, the mean density of 0+ stayed at same level in 2020 compared to the average for the five latest years. The density of older parr has been relative stable, varying between 12-26 ind./100 sqm during the five latest year. (Table 3.1.1.4). In Råneälven the density of 0+ parr decreased with half compared to densities 2019. The density of older parr increased and was the highest observed so far.

Smolt production has been monitored in Simojoki and Tornionjoki by annual partial smolt trapping and mark–recapture experiments (see Annex 2 for methodology) since 1977 and 1987, respectively (Table 3.1.1.5). A so-called river model (also referred to as “hierarchical linear re-gression analysis”) has been applied to combine information from electrofishing and smolt trap-ping results, to obtain updated estimates of wild smolt production in years when high water flow has prevented complete trapping, including also rivers without smolt trapping (Annex 2).

With a 1–3 year time-lag (needed for parr to transform to smolts) wild smolt runs have followed changes in wild parr densities. In the late 1980s, the annual estimated wild smolt run was only some thousands in Simojoki and less than 100 000 in Tornionjoki (Table 3.1.1.5). The first in-crease in the production occurred in the early 1990s, and a second, higher jump occurred in the turn of the millennium. Since then, smolt runs have not increased in Simojoki, while in Torni-onjoki the runs have continued to increase until the late 2010s Since the turn of the millennium, annual estimated runs of wild smolt have exceeded 20 000 and 500 000 smolts with high certainty in Simojoki and Tornionjoki, respectively. Since 2008, estimates of wild smolt runs have ex-ceeded one million smolts in the Tornionjoki.

Smolt trapping in 2020 was unsuccessful in Tornionjoki, due to too high and late spring flood, which prevented setting up the trap early enough. The river model updated with the latest parr density and smolt trapping data estimated the 2020 smolt run to be approximately 1.4 million smolts (median value, 90% PI’s 1.2–1.8 million). The river model predicts about 1.5–1.8 million smolts for 2021–2022.

Smolt trapping in Simojoki was conducted successfully in 2020, although the trap was set up relatively late in comparison to the water temperature. This together with daily catches being record high soon after the starting date indicate that some smolts had already migrated to the sea before trapping started. The trapping with mark–recapture experiments resulted in an

estimate of about 30 000 smolts (median value, 95% PI’s 19 400–49 300). The river model with electrofishing and smolt trapping data up to 2020 updated the smolt run estimate to about 38 000 smolts for 2020 (median value, 90% PI’s 27 100–53 800 inds.). Moreover, the river model predicts an increase to approx. 50 000 smolts/year for the years 2021–2022.