Studies of life-history variation across a species range are crucial for ecological understanding and successful conservation. Here, we examined the growth and age of Wels catfish (Silurus glanis) in Sweden, which represent the northernmost populations in Europe. A total of 1183 individuals were captured, marked and released between 2006 and 2020. Mark-recapture data from 162 individuals (size range: 13–195 cm) were used to estimate von Bertalanffy growth curve parameters which revealed very slow growth rates compared to catfish within the core distribution area (central Europe). The fitted von Bertalanffy growth curve predicted a 150 cm catfish to be around 40 years old, while the largest recaptured individual (length 195 cm) was estimated to be 70 (95% CI 50–112) years old. This was substantially older than the previously documented maximum age of a catfish. The weight at length relationships in these northern peripheral populations were similar to those documented for catfish in central Europe indicating that resources did not constrain growth. This indicates that the slow growth and exceptional high age in the northern catfish populations are the result of lower temperatures and/or local adaptations.

Sampling of catfish was conducted from 2014 to 2020 in Lake Möckeln (56° 39′ 48.9′′N 14° 8′ 57.7′′ E) and between 2006 and 2020 in River Emån (57° 7′ 46.7′′ N 16° 30′ 10.8′′ E) (Fig. 1). With the aim to capture the complete size structure of the focal populations we sampled individuals in both systems (Lake Möckeln and River Emån) between May and September using two methods, longlining for targeting adults (> 70 cm) and fyke-nets for juveniles and subadults. Longlines consisted of a floating mainline (100–400 m depending on the sampled area) to which monofilament (1.2 mm) leaders (1–1.5 m, looped to Scotty release clips) were fitted with single treble hooks and baited with native cyprinids were attached every 10–20 m. Fyke-nets varied in design/size between systems, in Lake Möckeln we employed a large two-armed fyke-net (20–30 m arms, 1.5 m diameter entrance) whereas in River Emån we used a smaller single-arm fyke net (8 m arm, 0.5 m diameter entrance). Both fyke-nets and longlines were initially set randomly throughout the systems but over time placed in hot-spot locations to obtain higher n-values. All gear were checked daily.

All fish were measured (total length) and weighted immediately after capture and date of capture was recorded. Using a measuring tape, small individuals (< 40 cm) was measured to the nearest mm whereas the length of larger fish was determined to the closest cm. The weight of small individuals was determined to nearest gram (Berkley 50 lbs/22 kg), larger individuals to the nearest 0.1 kg (Steinberg Systems SBS-KW-300/100-O). All individuals were marked with a PIT-tag (Passive Integrated Transponder from Biomark, 23 mm HDX), injected in the pelvic girdle or the abdominal cavity. Recaptured individuals, as identified by the PIT-id, was measured and weighted to be able to compare growth since the initial day of tagging before released. The sample procedure of a catfish took < 5 min.

The mark-recapture data was used to construct a von Bertalanffy growth curve (VBGC) of age at length by adopting the method by Fabens 1965 (sensu Hamel 2014). A reformulated VBGC-equation (ΔL = (L∞ − Lm)(1 − e(−K × ΔT)) was fitted to the mark-recapture data of growth between mark and recapture (ΔL), number of years between mark and recapture (ΔT, fractional values) and length at tagging (Lm). The parameters K and L∞ for the VBGC-equation were estimated iteratively by non-linear regression using the nls-function in the package stats (version 4.1.2, part of base R) in R (version 4.1.2; R Core Team). Only individuals recaptured 10 months or more after tagging were included (N = 162) to ensure that all individuals had experienced at least one growth season. The last recapture with the longest time span between mark and recapture was used for individuals with repeated recaptures such that every individual contributed with one observation of growth. Length at hatching (0.7 cm) was used as L0 in the VBGC-equation. A reformulated VBGC-equation (Age = log_e ((Length − L∞)/(L0 − L∞))/− K) with the L0 and the estimated parameters K and L∞ was used to predict age at length. The length (L) at weight (W) relationship was analysed by fitting a classic length–weight model with a power function (W = a × Lb). The parameters a and b were estimated by fitting a linear regression of log_e(W) as a function of log_e(L).