GILL RAKER COUNTING FOR APPROXIMATING THE RATIO OF RIVER- AND SEA- SPAWNING WHITEFISH, COREGONUS LAVARETUS (ACTINOPTERYGII: SALMONIFOR- MES: SALMONIDAE) IN THE GULF OF BOTHNIA, BALTIC SEA

Background. The ability to distinguish between stocks in mixed fi sheries is a prerequisite for a sustainable fi sheries management. In the Gulf of Bothnia the relative contribution of endangered river-spawning and sea-spawning whitefi sh, Coregonus lavaretus (Linnaeus, 1758), to fi sheries catches are currently not well known. This also applies to the southern Åland Islands, a major feeding ground for river-spawning whitefi sh from northern rivers. Riverand sea-spawning whitefi sh are mixed while away from the breeding grounds and off the spawning season, and cannot be distinguished based on external features. Materials and methods. Analysis on gill raker numbers of river-spawning (n = 480) and sea-spawning (n = 456) whitefi sh from twelve locations at the Finnish west coast and the Åland Islands was performed. In whitefi sh sampled from feeding grounds at the Åland Islands the strontium concentration was analysed in otoliths from fi sh (n = 20) with low (27) and high (30) number of gill rakers. Results. A marked difference in the mean gill raker number of the riverand sea-spawning whitefi sh stocks was observed. The weighted mean of gill rakers of whitefi sh caught at spawning locations showed that the number of gill rakers of fi sh from rivers and the sea were 29.9 ± 2.14 (n = 480) and 26.7 ± 2.21 (n = 456), respectively. The difference between the two groups was highly signifi cant (t = 22.50, df = 934, P < 0.0001). The means differed by 3.20 (2.92–3.48, 95% CL) indicating the groups are well separated. In whitefi sh sampled at feeding grounds at the Åland Islands, otolith strontium concentration was higher (t = 2.09, df = 18, P = 0.04) in fi sh having 27 gill rakers (3.86 ± 0.30 mg · g–1, n = 10), compared to those having 30 gill rakers (3.54 ± 0.35 mg · g–1, n = 10). Otolith strontium analysis thereby supported the utility of gill raker counting data for estimating the proportion of riverand sea-spawning whitefi sh in mixed populations. As expected, the gill raker counting method successfully indicated temporal alterations in the proportions of riverand sea-spawning whitefi sh on feeding grounds. Conclusion. Gill raker counting is an easy, fast, and inexpensive method that can be used to estimate the spatiotemporal occurrence and migratory patterns of riverand sea-spawning whitefi sh at the southern feeding grounds in the Gulf of Bothnia, and thereby aid in a sustainable management of whitefi sh stocks.

River-and sea-spawning whitefi sh belong to the demersal coastal fi sh fauna, and are frequent benthic feeders (Himberg 1995, Verliin et al. 2011. River-spawning whitefi sh spawn in rivers along the Finnish west coast and can migrate long distances, e.g., between rivers in the north of Finland and Åland Islands in the south (>700 km) (Wikgren 1962, Lind and Kaukoranta 1974, Lehtonen 1981, Lehtonen and Himberg 1992. This migration occurs along the Finnish coast ( Fig. 1). Rich food sources due to benefi cial water temperature, light, and salinity are the probable reasons for the southward migration of river-spawning whitefi sh, while preferential spawning locations are the driving forces for their returning. At the Åland Islands and the adjacent Archipelago Sea the feeding conditions are particularly benefi cial for whitefi sh, since thousands of small islands and reefs offer a vast area of shores and shallows abundant with food (Lehtonen 1981, Himberg 1995. River-spawning whitefi sh usually leaves the spawning-rivers and their estuaries for migration towards the south during their fi rst year (Lehtonen and Himberg 1992, Leskelä et al. 2009, Jokikokko et al. 2012. Mature river-spawning whitefi sh males normally return for spawning at the age of 3-4 years and females at the age of 4-5 years (Lehtonen 1981, Leskelä et al. 2009). The migration from the Åland Island for spawning in the northern rivers starts in June-July, and there may be several migration cycles. A migration speed of up to 15 km per day has been reported (Wikgren 1962). Sea-spawning whitefi sh occurs all along the Finnish coast. Sea-spawners normally migrate less than 200 km and move between deep and shallow water (Dahr 1947, Lehtonen andHimberg 1992). Around the Åland Islands sea-spawning whitefi sh is highly stationary (Leskelä unpublished * ). The spawning locations near the seashore are scattered around the Åland Islands. Both river-and sea-spawning whitefi sh spawn in October-November.
The spatiotemporal occurrence of river-spawning whitefi sh and its ratio in whitefi sh catches at the Åland Islands and the Finnish west coast are not well known, as river-spawning whitefi sh cannot be distinguished from sea-spawning whitefi sh based on the external features. However, it has been suggested that on the population level identifi cation is possible by gill raker counting (Him-berg 1978, Lehtonen and Himberg 1979, Lehtonen 1981, Lehtonen and Böhling 1988. The gill raker number is a temporally stable (Svärdson 1957, Amundsen et al. 2004, Bernatchez et al. 2010, Siwertsson et al. 2012 (Table 1) and highly heritable (Kirpichnikov 1981) character. The river-spawning whitefi sh is characterised by the weighted mean number of 29-30 gill rakers (Järvi 1928, Svärdson 1957, Himberg 1970, Lehtonen 1981. The sea spawning whitefi sh again is characterised by a lower weighted mean number of gill rakers: 26-27 at the Åland Islands including the Archipelago Sea (Himberg 1970, Lehtonen 1981.
The strontium concentration in seawater increases with salinity (Campana 1999, Elsdon et al. 2008. Since strontium accumulation in otoliths is positively correlated with salinity of the ambient water (Campana 1999, Elsdon et al. 2008, Doubleday et al. 2013, river-and sea-spawning whitefi sh should be distinguishable from each other based * Leskelä A. 2008  River, J = Kokemäki River, K = Aura River, L = Kisko River; A schematic migration route for river-spawning whitefi sh along the Finnish west coast is marked by black line and arrows on their otolith strontium concentrations. During early lifetime, at least, river-spawning whitefi sh spend time in water of low salinity and is therefore expected to have lower otolith strontium levels. Proper specimen identifi cation is necessary for collecting data on the spatiotemporal occurrence and ratio in catches of river-and sea-spawning whitefi sh. This study aimed at validating the method of gill raker counting for discriminating between river-and sea-spawning whitefi sh at the Åland Islands. For this, the gill raker numbers of river-and sea-spawning whitefi sh stocks were compared, and otolith strontium concentration analysed in a subset of whitefi sh with low (27) and high (30) number of gill rakers. Our results may help in the administration of whitefi sh stocks preservation.

MATERIALS AND METHODS
Study area. The Gulf of Bothnia reaches ~750 km from the Åland Islands in the south, to the Tornio region in the north ( Fig. 1) (Anonymous 2010). It is a shallow sea with an average depth of ~60 m. The whole sea area is usually covered by ice during the winter. Average insolation, water temperature, and salinity (2‰-6‰) increase from the north to the south. Stock identifi cation analysis. To establish the basis for stock identifi cation analysis mature river-and sea-spawning whitefi sh were collected from the Åland Islands and the Finnish west coast (Table 1), and the gill rakers counted as described below. Additional data on gill raker numbers of river-and sea-spawning whitefi sh were collected from literature or extracted from our own previously obtained data (Table 1). Mixed whitefi sh sampling. To sample mixed whitefi sh populations, the fi sh were caught at reefs close to Kobba Klintar Island (60º1.8′N, 19º53′E) located south of the city of Mariehamn at the Åland Islands (Fig. 1). Fish were captured in June, July, and August 2012 with stationary gill nets (1.8 m deep, 45 mm mesh size (square measure; knotted square mesh; 0.17 mm nylon tread diameter) at 2-5 m depths. Four to six samplings were made every month (1-12 June, 11-20 July, and 7-16 August) when six to twelve fi sh were caught. The fi sh weight was between 0.45 and 1.2 kg. The sex and degree of maturation of the fi sh were determined based on the development stage of the gonads (Nikolsky 1963) into juvenile whitefi sh (immature, sex unknown), mature whitefi sh at development stages 1-2 (immature/resting, not to spawn the present year), and 3+ (mature, to spawn the present year). The ratios of female to male and immature to mature fi sh ranged from 0.9 (57/62) to 2.7 (94/35), respectively. Gill raker counting. The gills were isolated from the head and the number of gill rakers on the fi rst (outer) gill arch on the left side subsequently counted under a stereomicroscope. The gill rakers were counted independently by two researchers (HH and MH), and when necessary recounted until consensus was reached. Otolith analysis. Strontium concentration was measured in whole sagittal otoliths from an equal number of randomly selected males and females with 27 gill rakers (n = 10) and with 30 gill rakers (n = 10). Otoliths were analysed with Varian VISTA-MPX inductively coupled plasma-optical (atomic) emission spectrometer (ICP-OES). Whole otoliths were dissolved in 2 mL HNO 3 in the microwave accelerating system MARS 5 (CEM), 1200 W, 50% power, 15 min, 414 kPa pressure, max temperature 210ºC, 10 min hold time. After mineralization, 8 mL water was added. For calibration certifi ed ICP reference standards (Fluka) were used. Data analysis. Student's t-test (unpaired two-tailed) was used to compare gill raker number data from river-and sea-spawning whitefi sh, and otolith strontium concentrations, while ANOVA with Tukey post-hoc test was used to study alteration of gill raker number data over time on whitefi sh sampled from feeding grounds.  Fig. 1; MPS = mixed population, sea (summer ), SPS = spawning population, sea (autumn), SPR = spawning population, river (autumn).

RESULTS
Gill raker number data of eleven river-and sea-spawning whitefi sh populations at the Finnish west coast and the Åland Islands are presented in Table 1. The weighted mean number of gill rakers of river-spawners is 29.87 (SD = 2.14, n = 480) and the corresponding number of gill rakers of sea-spawners is 26.67 (SD = 2.21, n = 456). The difference between the two groups was highly signifi cant (t = 22.50, df = 934, P < 0.0001). The means differed by 3.20 (2.92-3.48, 95% CL) indicating the groups are well separated.
Based on the weighted mean values of spawning reference populations river-spawning whitefi sh was defi ned to have ≥ 29 gill rakers and sea-spawning whitefi sh ≤ 28 gill rakers, respectively.
To study mixed whitefi sh populations, a total of 129 whitefi sh were sampled at feeding grounds at the Åland Islands during a three-month period. The gill raker number ranged from 23 to 34 (mean: 28.7) ( Table 2) The gill raker number distribution showed two modes, at 27 and 30 rakers (Table 2). Altogether, 54% of the whitefi sh had ≥ 29 rakers ( Table 2). The distribution of gill raker counts in the catch from June showed a major peak at 30, in July two peaks at 27 and 29, and in August a major peak at 27 (Table 2). ANOVA indicated differences between gill raker number means in samples from June, July, and August (F 2,126 = 5.73, P = 0.004). Pairwise comparison using Tukey post-hoc test showed that the weighted mean number of gill rakers in whitefi sh caught in June (29.7) was signifi cantly higher (P = 0.01 and 0.008, respectively) than the number in July (28.4) and in August (28.2) ( Table 2).

DISCUSSION
The presented analysis on gill raker number data from river-and sea-spawning whitefi sh stocks shows that gill raker counting can be used to estimate the proportion of river-and sea-spawning whitefi sh, as these whitefi sh forms cannot be distinguished based on outer features. Our results verify previous indications that the mean number of gill rakers is higher in river-spawning whitefi sh populations than in sea-spawning ones at the Åland Islands including the Archipelago Sea (Himberg 1970, 1978, Lehtonen and Himberg 1979, Lehtonen 1981, Lehtonen and Böhling 1988, Ozerov et al. 2015. Otolith microchemistry analysis supported the utility of the gill raker counting method. In the material collected at reefs close to Kobba Klintar, Åland Islands (Fig. 1), the whitefi sh with 30 gill rakers had signifi cantly lower otolith strontium concentration than the specimens with 27 gill rakers. The lower strontium concentration in otoliths of fi sh with high gill raker number is apparently due to the time of this river-spawning whitefi sh spent in water with low salinity as fry and juvenile, and due to possible spawning migrations as mature. Strontium concentration of seawater increases with increasing salinity (Campana 1999, Elsdon et al. 2008, and this salinity related strontium level is refl ected in otolith strontium concentrations (Secor and Rooker 2000, Zimmermann 2005, Engstedt et al. 2010, Macdonald and Crook 2010, Engstedt et al. 2012. In accordance with our results, the otolith strontium concentration in sea spawning whitefi sh (Mariehamn) caught at spawning grounds is signifi cantly higher than in otoliths of river spawning whitefi sh (Kokemäki and Tornio rivers) (Hägerstrand, unpublished results).
Gill raker number data indicated temporal alteration of river-and sea-spawning whitefi sh at the feeding grounds close to Kobba Klintar, Åland Islands (Fig. 1). The mean number of gill rakers of the sampled whitefi sh was 28.7 (Table 1 and 2). This value was intermediate compared with the mean gill raker number of sea-and river-spawning whitefi sh stocks (Table 1), i.e., typical for a mixed sample (Himberg 1978, Lehtonen 1981, Himberg 1995. When the sampled whitefi sh was separated into three groups; fi sh caught in June, July, and August, a marked decrease in gill raker number over the period was revealed. The mean number of gill rakers of whitefi sh caught in June was signifi cantly higher (29.7) compared to fi sh caught in July (28.4) and August (28.2) ( Table 2). The results indicate that both river-and sea-spawning whitefi sh with a different mean number of gill rakers occurred on the reefs at different times during the summer. Apparently, river-spawning whitefi sh (≥ 29 gill rakers) dominated (~79%) in the beginning of the summer (June), but their frequency decreased through July until August. Sea-spawning whitefi sh (≤ 28 gill rakers) dominated (~65%) the sampling sites by the end of the summer (August). The reduction in the ratio of river-spawning whitefi sh from June to July and August on the sampling site could be due to migration towards spawning rivers. The migration of river-spawning whitefi sh from the feeding areas to the northern rivers starts in June-July (Lindroth 1957, Wikgren 1962, Lind and Kaukoranta 1974, Lehtonen 1981. The time of departure is probably dependent on the distance to the spawning rivers (Lehtonen 1981). Immature fi sh may not migrate. Some immature high gill raker number whitefi sh remained at the sampling site throughout the summer (Table 3). This is consistent with previous fi ndings indicating that river-spawning whitefi sh may stay in the south for feeding for several (3-5), years until maturation (Lehtonen 1981). The presence of mature males with high gill raker number in the August catches raises the question whether river-spawning males may spawn in nearby rivers like the Aura and Kisko rivers (Fig. 1), or in the sea. While homing may be less pronounced in males than in females (Huusko and Grotnes 1988), the destruction of traditional spawning locations and potential homogenization of river-and sea-spawning whitefi sh gene pools as a result of stocking activities may have affected the homing instinct of river-spawning whitefi sh. Sea-spawning whitefi sh dominated the catches in August. Sea-spawning whitefi sh has short distance to their scattered spawning locations at the seashores around the Åland Islands. It is unclear whether water temperature preference, search for a particular prey or competition for food infl uence the spatiotemporal occurrence of the two main whitefi sh groups at the feeding reefs. Abundant annual stocking of mainly river but also sea-spawning whitefi sh has for several decades been undertaken along the Finnish west coast, including the Archipelago Sea (Anonymous 2013). At the Åland Islands only sea-spawning whitefi sh has been stocked (Leskelä 2008). Whitefi sh to be stocked is typically raised in freshwater. It is not known how the raising conditions affect otolith elemental concentrations of whitefi sh. Low mean values of strontium concentrations in otoliths in high gill raker number whitefi sh may occur both in wild-born and stocked fi sh of the river spawning type. To distinguish between stocked and wild-born whitefi sh is an important issue that remains to be clarifi ed.