At high latitudes, early life stage survival of fish is often associated with how spawning time relates to the timing of the spring bloom. With ocean warming, basic physiological rates of ectotherms, like fish, will speed up - including gonadal development rates which dictate spawning time. Since warmer water is thought to influence the spring bloom timing differently than that of fish spawning time, the two may fall out of synchrony in the future. The precise mechanisms between temperature and gonadal development and spawning time have, however, been difficult to disentangle. Here, we take advantage of a series of independent laboratory experiments measuring individual oocyte development up to or near spawning for 153 Atlantic cod (Gadus morhua) kept between 3 and 12 °C. From these data, we derive a predictive, mechanistic equation for daily oocyte growth rate as a function of temperature and oocyte developmental status (diameter). The vitellogenic oocyte growth follows an accelerating pattern, and the model predicts that spawning can advance by up to 7 days per 1 °C increase. Within-treatment variation is, however, of comparable magnitude to between-treatment temperature effects. The model was also tested in the field by back-calculating oocyte development of 82 fish (2018-2021) sampled at two locations along the Norwegian coast, using daily ambient temperatures from telemetry tags during vitellogenesis as model input. We find that Atlantic cod are able to initiate vitellogenesis over a period of several months in late summer and autumn, as well as regulate the oocyte development rate across a wide range of temperatures – both leading to significant phenotypic plasticity in spawning phenology.

All methods were carried out in accordance with relevant guidelines and regulations and also followed the ARRIVE guidelines (https://arriveguidelines.org). Permission to conduct this research at the Institute of Marine Research (IMR), Norway, was given by the Norwegian Animal Research Authority.

Temperature-controlled tank experiments of oocyte development

Our study utilises results from two different temperature-controlled experiments, the first (Experiment 1) conducted in 2005-2006 (5 and 9.6 °C) and the second (Experiment 2) in 2018-2019 (3, 6, 9, and 12 °C); further details on the methodology have been published in Kjesbu et al. (2010) and Skjærven et al. (2024), respectively.

Experiment 1 took place at the Institute of Marine Research (IMR) facilities in Bergen between June 1^st, 2005 (as 2-year olds, median total length of 54 cm) and January 26^th 2006 (as 3-year olds). Fish were supplied from a local rearing facility operated by the IMR in the Parisvatnet marine pond west of Bergen (60.6°N, 4.8°E), after which they were kept at IMR Austevoll Research Station for further on-growth. In spring 2005 155 fish, males and females, were divided into two outdoor 30 m² tanks (1.8 m water depth) with natural seawater (inlets at 50 + 120 m depth). Starting on June 1^st, fish were randomly redistributed between the tanks. One tank remained at ambient seawater temperatures (average of 9.6 °C over the study period), while the other tank was cooled overnight to 5 °C, a temperature maintained for the rest of the experiment.

Experiment 2 took place at the IMR Matre Research Station (north of Bergen) between October 1^st 2018 and April 26^th 2019. A total of 200 coastal cod (median total length of 66 cm) were wild caught with eel nets outside the island Bømlo, south of Bergen (59.8 °N, 05.3 °E) during spring and summer of 2018, and kept in sea cages awaiting transport to Matre in early September. In Matre, the fish were randomly and equally distributed into 12 8 m² indoor tanks (14 to 18 fish per tank) with natural, filtered seawater of 8 °C (inlet at 90 m depth). Light was programmed to follow the natural photoperiod at 60.0 °N. Starting on October 1^st, the temperatures in the tanks were adjusted at a maximum of 1 °C per day, until they reached 3, 6, 9, and 12 °C (3 tanks per temperature).

All tanks in both experiments had low stock density (≤ 8 kg m ^-3), and the fish were fed high-quality aquaculture dry pellets (Amber Neptun, produced by Skretting, Norway) ad libitum three times per week. Within-tank temperatures in Experiment 1 varied temporally by about 0.5 °C but were extremely stable in Experiment 2 (± 0.04 °C) due to the advanced, automated control systems. Throughout the experiments, all individuals were regularly collected, anaesthetized (Finquel, 0.6 g. L ^-1), and total length and whole body weight measured. In parallel, oocytes were sampled using gonad catheterisation (Pipelle®, Laboratoire C.C.D). The ovarian tissue (~0.25 g) was preserved in 3.6 % formaldehyde for automatic image analysis.

During Experiment 1, samples were collected every 4 weeks (Oct 6^th, Nov 2^nd, Dec 6^th, and Jan 10^th), from a total of 33 and 38 females, in the 5 °C and 9.6 °C tanks, respectively. In Experiment 2, we sampled every 3 to 5 weeks (Oct 1^st, Nov 19^th, Dec 18^th, Jan 17^th, Feb 2^nd, Feb 18^th, March 3^rd, March 18^th, and April 1^st), with an average of 5 times from a total of 16, 20, 23, and 23 females, in the 3, 6, 9, and 12 °C tanks, respectively. Spawning individuals were not sampled.

Field sampling of oocytes

Oocytes were also sampled from wild-caught Atlantic cod at two locations: Austevoll (2020 and 2021), Smøla (2019). Locations are marked in Figure 1.

The oocyte sampling was conducted in relation to a fish-tagging effort of local Norwegian coastal cod (see also below section on acoustic tags). At both locations, individuals were caught using baited pots and gillnets and kept in sea cages for a maximum of three weeks prior to tagging and measurements, which took place on Jan 27- 29^th (2020 and 2021) at Austevoll and Feb 11-12^th (2019) at Smøla. The fish were anaesthetized and tagged, measured for length and weight, and for females, oocytes were sampled using gonad catheterisation (see above) and preserved in 3.6 % formaldehyde for image analysis. After sampling, individuals were released. In total, 40 (2020) and 19 (2021) oocyte samples were taken at Austevoll, and 78 at Smøla (2019). For further details, consult McQueen et al. (2022) for the Austevoll samples and Skjæraasen et al. (2021) for the Smøla samples.

Image analyses of oocyte diameter

For all samples, the preserved ovarian tissue from both laboratory and field was processed automatically to measure the diameter (µm) of a minimum of 200 oocytes per individual using the auto-diametric method (Thorsen and Kjesbu 2001) and the software Image J (v. 1.52) with the ObjectJ-plugin (Schneider et al. 2012). From these measurements, the mean oocyte diameter (D_MN, µm) and the mean leading cohort (LC) oocyte diameter (D_LC, µm) for each individual were estimated. The LC was defined as either the mean (Experiment 1), or 95% median (Experiment 2), of the 10% largest vitellogenic oocytes. The difference between the two estimations is negligible.

Vitellogenic temperature trajectories

We started by utilising depth and temperature information from 700 mature cod with acoustic tags from Innovasea (formerly Vemco) surgically placed in the abdominal cavity under sedation (see also the above section Field sampling of oocytes). Of these, there were 217 (Austevoll) and 319 (Smøla) V13P-tags transmitting ambient depth (P), and 164 (Smøla) V13TP-tags transmitting both ambient temperature (T) and depth (P). All tags transmitted data on average every 250 seconds within an array of 27 and 51 acoustic receivers at Austevoll (2019-2021) and Smøla (2018-2019), respectively. In addition, permanent depth resolved (1, 5, 10, 15, 20, 25, 30, 40, 50, 60, 70, 80 m) temperature loggers with bi-hourly resolution were deployed at both locations: 2 at Austevoll and 4 at Smøla. Ambient fish temperatures for individuals with V13P-tags were estimated by looking up each transmitted depth (P) at the nearest temperature logger (Ti, interpolated to tag-depth) such that the estimated ambient temperature T_E ~ T_i(P). The accuracy was checked by comparing the tag-transmitted temperature and logger-estimated temperatures from the 164 V13TP-tags in Smøla that transmitted both P and T using a major axis regression linear model. From the 700 individual temperature trajectories (T and T_E), we located a total of 24 individuals with near-continuous (max gap between recordings < 14 days) temperature recordings between September and March (vitellogenic temperature).

References

Kjesbu, O. S., C. T. Marshall, D. Righton, M. Krüger-Johnsen, A. Thorsen, K. Michalsen, M. Fonn, and P. R. Witthames. 2010. Thermal dynamics of ovarian maturation in Atlantic cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic Sciences 67:605-625.

McQueen, K., J. J. Meager, D. Nyqvist, J. E. Skjæraasen, E. M. Olsen, Ø. Karlsen, P. H. Kvadsheim, N. O. Handegard, T. N. Forland, and L. D. Sivle. 2022. Spawning Atlantic cod (Gadus morhua L.) exposed to noise from seismic airguns do not abandon their spawning site. ICES Journal of Marine Science 79:2697-2708.

Schneider, C. A., W. S. Rasband, and K. W. Eliceiri. 2012. NIH Image to ImageJ: 25 years of image analysis. Nature Methods 9:671-675.

Skjærven, K. H., M. Alix, L. Kleppe, J. M. O. Fernandes, P. Whatmore, A. Nedoluzhko, E. Andersson, and O. S. Kjesbu. 2024. Ocean warming shapes embryonic developmental prospects of the next generation in Atlantic cod. ICES Journal of Marine Science 81:733-747.

Skjæraasen, J. E., O. Karlsen, O. Langangen, T. Van der Meeren, N. B. Keeley, M. S. Myksvoll, G. Dahle, E. Moland, R. Nilsen, K. M. E. Schroder, R. J. Bannister, and E. M. Olsen. 2021. Impact of salmon farming on Atlantic cod spatio-temporal reproductive dynamics. Aquaculture Environment Interactions 13:399-412.

Thorsen, A., and O. S. Kjesbu. 2001. A rapid method for estimation of oocyte size and potential fecundity in Atlantic cod using a computer-aided particle analysis system. Journal of Sea Research 46:295-308.