Tracking oocyte development and the timing of skipped spawning for north‐east Arctic haddock ( Melanogrammus aeglefinus )

Abstract The present study tracked oocyte development over 9 months and noted incidences of ‘skipping’, i.e., adults terminating their upcoming reproductive cycle, in field‐caught north‐east Arctic (NEA) haddock (Melanogrammus aeglefinus), currently the largest stock of this species. Applications of advanced image and histological techniques revealed the presence of cortical alveoli oocytes (CAO), which prevailed as the most advanced oocyte phase for 4–5 months. This new finding of an extended and early appearance of CAOs in this gadoid was supported by that vitellogenesis first started to appear 3 months later. The subsequent oocyte growth trajectories indicated that larger individuals [total length (TL) = 70 cm] typically spawn in the order of 3 weeks earlier than the smaller ones (TL = 40 cm). The spawning season appeared stretched over about 3 months. The majority of skipping females arrested oocyte growth at the CAO phase followed by atretic reabsorption. Compared to those individuals maturing for the spawning season, ‘skippers’ generally exhibited lower body condition, characterized also by relatively lower liver sizes at the time of the main spawning season. This study demonstrated well‐developed skipping dynamics, but also that the CAO period, i.e., when skipping takes place, may be exceedingly long in this commercially valuable gadoid and that its reproductive cycle in many ways deviates from that of the data‐rich, sympatric NEA cod (Gadus morhua).

north Atlantic. The largest stock of this species-north-east Arctic (NEA) hereafter referred to as M. aeglefinus (NEA stock)-resides in the Barents Sea (Johannesen et al., 2019) whereas its management unit area extends south to 62 N, although a recent genetic study suggests a genetic break at 68 N (Berg et al., 2021) (Figure 1). As for M. aeglefinus in general, the oocyte development and maturity dynamics of this stock have so far been addressed within restricted time windows, i.e., at or near the spawning season (Skjaeraasen et al., 2013(Skjaeraasen et al., , 2015. Hence, central aspects of its reproductive cycle still require closer attention, as addressed below. As an iteroparous species, M. aeglefinus is assumed to follow an annual reproductive cycle. A growing body of evidences, however, suggests that in some years more than half of the sexually mature portion of M. aeglefinus (NEA stock) does not follow this pattern (Skjaeraasen et al., 2015). This phenomenon-known as skipped spawning, i.e., adults terminating their upcoming reproductive cycle-has the potential to bias the spawning stock biomass (SSB) if not accounted for in assessment models (Rideout & Tomkiewicz, 2011). Although skipping individuals of M. aeglefinus (NEA stock) are excluded from annual SSB estimates, uncovering under what conditions insufficient energy reserves or associated trade-offs (Skjaeraasen et al., 2020) are controlling interruption of the reproductive cycle would help to improve the prediction of the stock's response to environmental stressors.
It is generally accepted that the presence of cortical alveoli oocytes (CAO), or advanced previtellogenic oocytes (PVO)   (Figure 2), is the first marker that the female under scrutiny is going to spawn in the forthcoming spawning season (Skjaeraasen et al., 2010). The reproductive cycle of some females may, however, get arrested at this advanced PVO phase (resting skipper) (Rideout et al., 2005). Alternatively, oocytes might be reabsorbed at the early CAO phase, a process which has been associated with insufficient energy reserves (reabsorbing-CAO skipper) (Rideout & Tomkiewicz, 2011;Skjaeraasen et al., 2009Skjaeraasen et al., , 2015. Conversely, maturing M. aeglefinus (NEA stock) which continue oocyte development will migrate southwards to spawning grounds ($68 to 72 N) along the shelf break from the Lofoten archipelago and northwards (Sivle & Johnsen, 2016) in January -February (Bergstad et al., 1987;Saetersdal, 1952). However, some of these individuals may still interrupt their reproductive cycle by reabsorbing all vitellogenic oocytes (VO) (Figure 2) through follicular atresia (reabsorbing-VO skippers) ( Figure 3) or even retain their fully ripened eggs (retaining skippers) (Rideout & Tomkiewicz, 2011).
Although various aspects of the reproductive dynamics of M. aeglefinus do exist in the literature, no studies have dedicatedly tracked oocyte development following photoperiod cues in summer (Davie et al., 2007) and autumn , i.e., when processes associated with the fish's energetic state may be important for the reproductive decision (see above). We took advantage of modern image analysis techniques in combination with histology to pin-point the time course of reproductive events in field-caught M. aeglefinus (NEA stock) through the initiation of the reproductive cycle until the onset of spawning to (i) examine oocyte development and timing of reproductive commitment, (ii) investigate the F I G U R E 1 Map of sampling stations in the Barents Sea. The size of the bullets is proportional to the number of individuals sampled at each location. The colours of the bullets correspond to the surveys involved in the sampling collection. Melanogrammus aeglefinus (NEA stock) distribution (hatched area) follows the shelf break and the extent of Atlantic water flowing into the Barents Sea from the south-west. The main spawning grounds are represented in black, modified from the official M. aeglefinus (NEA stock) map from the Institute of Marine Research, Norway (Sivle & Johnsen, 2016) dominant mode and fundamental mechanisms determining skipping, and, finally, (iii) discuss and compare our findings to related studies on the sympatric Gadus morhua (L.) (NEA stock), a groupsynchronous and determinate batch spawner like M. aeglefinus (Murua & Saborido-Rey, 2003).  (Table 1). Any gear selectivity was left unconsidered in this individual-based study. In accordance with standard fishing practices, all animals were deceased at landing. As such, animal ethics approval for this project was not required.

| Laboratory analysis
2.2.1 | Image analysis (whole mount) All ovarian samples presently considered (n = 590) were subjected to digital image analysis according to the auto-diametric method (Thorsen & Kjesbu, 2001). The initial step in this protocol included the use of an ultrasonic pen on a pipetted subsample to separate oocytes, which were stained with 2% toluidine blue and 1% sodium tetraborate, and, finally, photographed under a dissecting microscope, i.e., using similar laboratory equipment and protocols as detailed earlier (Anderson et al., 2020). The diameter of at least 200 oocytes (to the nearest μm) were measured using open-source ImageJ software (v. 1.52, https://imagej.nih.gov/ij/) with the plugin ObjectJ (https:// sils.fnwi.uva.nl/bcb/objectj/). The mean size of the 20 largest oocytes, hereafter labelled as oocyte leading cohort diameter (LC, in μm), reflected the female maturity stage (Thorsen & Kjesbu, 2001).

| Histology
Ovarian subsamples (n = 298) were analysed histologically to detail microscopic structures-oocyte development phases as well as postovulatory and atretic follicles (Figures 2 and 3)-to aid highly precise F I G U R E 2 Histological sections of the studied oocyte development phases of Melanogrammus aeglefinus (NEA stock) stained with toluidine blue. (a) First developmental phases, oogonia and PVO phases 1, 2 and 3, cytoplasm is uniform and homogenous. (b) PVO phase 4A, a circumnuclear ring (CNR; arrow) appears in the cytoplasm as an indistinct feature located centrally. (c) PVO phase 4B, the CNR has become distinct and has migrated towards the periphery of the cytoplasm. (d) PVO phase 4C, the CNR is gradually disappearing. (e) ECAO phase, commences when cortical alveoli (CA) appear at the periphery of the cytoplasm. (f) LCAO phase, CA increases in size and quantity, and the chorion becomes more pronounced. (g) EVO phase, small yolk granules (YG) appear on the periphery of the cytoplasm. (h) LVO phase, an increase in number, size and distribution of the yolk granules, which occupy virtually all the cytoplasm maturity staging (see below). These pieces of ovarian tissue were processed using standard protocols for resin embedding (Kulzer, Technovit 7100, Wehrheim, Germany), producing two sets of 4 μm sections stained with 2% toluidine blue and 1% sodium tetraborate or, alternatively, periodic acid-Schiff (PAS) and Mallory trichrome stain to further document any presence of post-ovulatory follicles (POF, Figure 3a) . During the subsequent screening of sections, PVO (late primary growth oocytes) were divided according to Shirokova (1977)

| Statistical analysis
Prior to performing any test on changes in LC over time, the date variable was converted into the number of days, where the earliest  test was used to examine differences detected by ANOVA.

| Skipping modes (histology)
The first skipping female was identified (Figure 3a) at the end of October by the presence of massive atresia (>50%) of ECAO (Figure 7).
From November onwards, the occurrence of skippers became much more frequent and was observed in all subsequent months. In total, 65 skipping females were identified during the present sampling period.
Of these, 42 were identified as reabsorbing-CAO skippers, exhibiting reabsorption of nearly all CAO, and 22 were classified as resting skippers by the presence of advanced PVO, with very low atretic intensity (<5%) (Figure 7). Only one female was identified as a reabsorbing-VO

| Biometric and age differences by maturity category (histology)
Within the material examined by histology, significant differences in age and TL existed among maturity categories (Figure 8), where maturing females were significantly older (two-way ANOVA, F 4,287 = 25.92, P < 0.001) and larger (F 4,293 = 46.67, P < 0.001) (6.1 ± 2.2 years, 55.9 ± 9.5 cm) compared to immature (3.8 ± 0.8 years, 40.4 ± 7.1 cm) and skipping (4.9 ± 1.0 years, 48.9 ± 6.1 cm) females.  (Figure 9a). In April, both skipping and spent females had a significantly lower HSI S than spawning and maturing females (two-way ANOVA, F 26,260 = 8.13, P < 0.001; Figure 9a). For K, the interaction term maturity category Â month was insignificant (P > 0.05), but the main effect, i.e., maturity category + month, was evidently in place (two-way ANOVA, F 12,285 = 5.17, P < 0.001; Figure 9b), where skipping females (November-April) showed a significantly lower K compared to maturing females (P = 0.004). Consequently, overall, HSI S was a more sensitive parameter in these respects than K; maturity category + month explained 38.9% of the variability in HSI S but only 14.4% in K.

| DISCUSSION
The In the months close to the start of the spawning season, the considerable variability in oocyte development across M. aeglefinus (NEA stock) individuals was evident in that some females, at a given day, showed up to 300 μm larger LC sizes compared to others. The reason for the observed and extrapolated variability in spawning time was partly related to size-specific effects (Tobin et al., 2010;Wright & Trippel, 2009), with the present simulation suggesting the start of spawning at TL = 70 cm about 3 weeks earlier than at TL = 40 cm.
Such a size effect on LC could potentially be related to the fact that immature females developing for their first spawning season (recruit spawners) may start the formation of CAO later in the autumn compared to already sexually mature females (repeat spawners), as observed in both M. aeglefinus in the North Sea (Tobin et al., 2010) and G. morhua . Size-specific differences in LC are generally found in G. morhua (Kjesbu, 1994;Kjesbu et al., 2010;Skjaeraasen et al., 2010) but have not previously been documented for M. aeglefinus (NEA stock) (Skjaeraasen et al., 2013), and may thus be a contributing factor to the long CAO phase observed in the present study.
In this tracking study we were for the first time able to demonstrate that the ultimate cause behind reproductive interruption of M. aeglefinus is massive atresia of CAOs. Such extensive reabsorption has earlier been reported both in field-caught and experimentally monitored G. morhua (Rideout, 2000;Skjaeraasen et al., 2009 (Kjesbu et al., 1998;Skjaeraasen et al., 2013), but also typically prey on Echinodermata (Jiang & Jørgensen, 1996;Tam et al., 2016)  being made much earlier, following a critical post-spawning feeding period (Burton, 1994;Lambert & Dutil, 1997;Rideout et al., 2005).
Skipping females in the present study were on average $5 years old due to increased gross primary and secondary production and less sea ice abundance in the Barents Sea . Increasing bottom temperature is largely counteracted by behavioural overcompensation, i.e., moving into ambient colder waters even though the environmental temperature as such increases (Landa et al., 2014), a phenomenon seen in many species able to move polewards or into a colder ocean current branch .
To conclude, the present study has highlighted important new cycle (e.g., immediately after the spawning season) to assess the nutritional status of females at a time when the physiological decision to spawn might be predetermined.

ACKNOWLEDGEMENTS
The authors thank technicians and coordinators onboard IMR research vessels and the Norwegian Reference Fleet for sampling of ovarian tissue. We are grateful to Anders Thorsen for expert supervision within digital methods and Grethe Thorsheim for assisting in the processing of images as well as helping with the histological work. Vemund Mangerud is thanked for various types of technical assistance.

AUTHOR CONTRIBUTIONS
This contribution is a revision of an earlier Master of Science thesis in