Material and Methods

Study location

The river Imsa in southwestern Norway (Figure 1) is an unregulated oligotrophic system. The catchment covers an area of 12800 ha, of which 1536 ha (12%) is lake surface (major lakes are Imsvatnet, 40 ha, and Storavatnet, 819 ha). A trap catching descending silver eels as well as ascending juveniles is located about 100 m from the river outlet in the sea. The distance from the trap, which has been in operation since 1975, to the upper end of eel habitat in the catchment is set at 20 km, and the eels ascend the system up to an altitude of approximately 215 m above sea level (Vøllestad & Jonsson 1988). Descending, predominantly silver, eels are caught in a Wolf trap (apertures 10 mm, inclination 1:10). The juveniles entering this watershed are small yellow eels (elvers or recruits) that are typically 70-90 mm long and weighing less than 1.0 gram, although a few individuals may be larger. The distance from the nearest lake along the free-flowing river to the fish trap is 970 m. There has been no stocking of eels in this watershed. Before 2006, there was a restricted seasonal yellow and silver eel fishery upstream of the trapping station. The number of ascending recruits and descending silver eels is given per calendar year.
Wolf traps with aperture 10 mm generally catch all eels larger than approximately 25 cm in length, which includes all silver eels in the river Imsa (Vøllestad & Jonsson, 1986). The elver trap leads all ascending recruits into a capture chamber where their numbers are recorded, and sub-sampling of size is performed, before they are released to continue upstream. The traps are checked twice every day (at circa 08:00 and 15:00 h).

Otolith data and age determination

Otoliths from the historical collection (collected in the 1980s and 90s) were initially read in toto by clearing them in 96% ethanol for 18-24 h before counting the annuli directly using a stereo microscope and 96% ethanol as refraction medium (Vøllestad, 1985). These otoliths were, since then, stored dry in an envelope, each labeled with length, sex and stage (yellow or silver). For the reanalysis, a subsample was selected from years with the highest sampling effort (1982, 1986, 1991 and 1992).
The more recent otoliths (from the 2010s) were sampled from eels caught during their downstream run in the river Imsa. Twenty-five silver eels were sacrificed per year (61 eels in 2016). Length, weight, fin and eye diameters were measured for stage determination (Durif, Dufour, & Elie, 2005; Durif, Guibert, & Elie 2009). The eels were dissected for sex determination and removal of otoliths. Otoliths from the 2010s were not analyzed using the old in toto method.
A total of 798 fish were processed. All otoliths (historical and new) were prepared by grinding, etching and staining and read according to the protocol described in ICES (2009, 2011) (Figure 2). The year 0 band was assigned as the first winter after the oceanic migration,i.e . it marked the beginning of the continental life stage. The last year was defined as the outer edge of the otolith since eels were caught during the fall season. Some otoliths had clear and regularly spaced annuli (Figure 2A). Others presented numerous tight rings, unevenly spaced, which sometimes joined in a “bundle” or fused into one large annulus on the other side of the otolith (Figure 2B, 2C). Whether these bundles represent one or several years is unknown. Here, we assumed that some of the marks forming a bundle represented false checks and thus one bundle represented one year. Otoliths were read by 2 or 3 observers, or for some samples by the same observer twice, but the second time several months after the first reading. As expected, some otoliths were easier to interpret than others, and the age estimates did not always agree between observers. To characterize the uncertainty in the readings, we assigned each age estimate with an Otolith Uncertainty Index (OUI) corresponding to how much the observations differed between observers/observations:

Calculations and statistics

Indicative growth rate of each individual was calculated by dividing the body length LT (mm) of the eel minus 70 mm, which is the mean size of glass eels when they recruit to European coasts (Elie, 1979; Svedäng et al., 1996), by the continental age.
Differences in mean age and mean length between decades and between OUI levels were tested using linear regression models. Differences in proportion of OUI levels between decades were tested using a Chi-square test. Differences between length-weight relationships were tested with ANCOVA after the variables were log-transformed to investigate changes in condition over the decades. Statistics were carried out using the statistical software R (R Core Team, 2019, v. 3.6.0).