First thorough assessment of de novo oocyte recruitment in a teleost serial spawner, the Northeast Atlantic mackerel (Scomber scombrus) case

Oocyte size frequency distribution

The OSFD, based on wholemount analysis (formalin-preserved diameter measurements), did not show any hiatus between the assumingly largest PVOs and the smallest VO (Supplementary, Fig. S1). The corresponding mean threshold value, determined statistically by the Gamma/Gaussian method (see technical details below), was 192 µm (95% CI: 187–196 µm) (Supplementary, Fig. S1). Based on histology, this value was, however, at ~ 230 µm, i.e. the formalin-preserved oocyte diameter of PVO4c (Supplementary, Figs. S2B, S3, Table S1).

Spawning progress

Addressing firstly “the population (wholemount) data set” of 1561 individuals (Table S2), the relative frequency of early-spawning (ORC1), mid-spawning (ORC2), and late-spawning (ORC3) females changed significantly as the spawning season progressed, although with dissimilarity between 2018 and 2019 (Supplementary, Fig. S4). Overall, a significant difference was found among the ORCs frequencies between the two field-sampling years (two-way ANOVA; p = 0.003). In June 2018, over 60% of the females caught were very late spawners or spent (ORC4), this relative frequency increased to almost 90% in July 2018 (Supplementary, Fig. S4A). For 2019, the ORC4 in June was about 50% (Supplementary, Fig. S4B). Combining these 2018 and 2019 data sets, the subsequent comparison showed that July 2018 clearly differed in terms of ORC (a posteriori Tukey test; Supplementary, Fig. S5). More females in mid-spawning were recorded in May and June 2019 compared to the same months in 2018, though this noted difference was statistically insignificant (Supplementary, Fig. S5). Altogether, these outlined variations in ORC (Fig. 1) may be related to survey coverage, i.e. in 2018 these samples were collected in Nordic waters, while in 2019 exclusively within the main spawning area (Fig. 2).

Population-level ORC and biometrics appeared linked, the latter represented either by total length (TL)-based gonadosomatic index (GSI_TL) or relative condition (K_n) (Fig. 3). The 2018 results showed that K_n was higher (p < 0.001) in late- (ORC3) and very late-spawning and spent fish (ORC4) compared to early-spawning (ORC1) and mid-spawning fish (ORC2) (Fig. 3B). In 2019, K_n values were more similar across ORC but highest (p < 0.001) in ORC1 and ORC4 (Fig. 3D). However, once more, the resulting significant differences in K_n between spawning status and years (two-way ANOVA, p < 0.001) might be due to survey coverage and time: mackerel females sampled in 2018 had most likely started feeding²⁶ since they were mainly collected in the Norwegian Sea (Fig. 2).

K_n values restricted to the 144 OPD samples (Supplementary, Fig. S6B, D) showed very much the same trend with ORC as just presented above for the population data set (Fig. 3B, D; Supplementary, Table S2), except for some situations where OPD-related K_n values were slightly higher, as for example ORC1 in 2018 (Supplementary, Fig. S6B) and ORC3 in 2019 (Supplementary, Fig. S6D). This illustration speaks for that the OPD samples were a representative subset, at least in these regards, see also the related weight-at-total length [W-at-TL] plot (Supplementary, Fig. S7).

Presence of postovulatory follicles and atresia

Both POFs and various degrees of vitellogenic atresia (Eα and Lα) were frequently annotated within the spawning season. For POFs, the corresponding volume fraction (Vν_i) showed an increase towards summer (Supplementary, Fig. S8A). However, a more pronounced presence of POFs was recorded in 2019 than in 2018, i.e., when the samples were collected within spawning areas (see above). Although atretic vitellogenic oocytes were detected in almost all months with samples taken, except in October, their presence was exceedingly low well off the spawning season (Supplementary, Fig. S8B). Thus, Eα atresia Vν_i peaked in July 2018, when mackerel most likely had ceased spawning and were feeding in the Norwegian Sea²⁶ (see above). Higher levels of Vν_i were generally observed for late (Lα) than for early (Eα) atresia, the former also being much more persistent; all examples of atresia outside the spawning season referred to this stage (Supplementary, Fig. S8B). Focusing on patterns within the spawning period as such, i.e., consulting ORCs, we observed that the Vν_i of both a atresia stages increased as spawning progressed from ORC1 to 3, seeing thereafter, at ORC4, a collective drop in Eα atresia but for Lα atresia this drop being restricted to 2019 (Supplementary, Fig. S9). No clear difference in K_n was seen between fish with or without α atresia, neither any evidence of a K_n effect across different TL on the presence of a atresia (Supplementary, Fig. S10).

PVO atresia was detected during the resting and early maturation period (August–April; Supplementary, Fig. S8B). September 2018 was an extreme case with PVO atresia being observed in all individuals analysed for OPD, amounting to a mean Vν_i of 8% (Supplementary, Figs. S8B, S10), despite that this month showed the highest mean K_n within the 2018 study year (Supplementary, Figs. S10, S11). K_n was typically around and above 1.0 in individuals where PVO atresia was observed (Supplementary, Fig. S10).

Oocyte diameter

Generally, phase-specific oocyte size represented by mean volume-based oocyte diameter (cODν_i) (where c stands for “corrected to formalin-preserved diameter”) varied little over the year. PVO2 and PVO3 showed indications of being smaller in macroscopically staged early-maturation months (October to January) compared to corresponding spawning and post-spawning months (March to September) (Supplementary, Fig. S2A). A partly different pattern was observed for late PVO phases; PVO4a was seemingly largest in October, whereas PVO4b and PVO4c in January 2019 (Supplementary, Fig. S2B). Likewise, no clear temporal pattern in size was recorded for phase_i-specific VOs (one-way ANOVA, p = 0.184 [VO1]—0.681 [VO3]) (Supplementary, Fig. S2C). However, cODν_i of FOM (germinal vesicle migration [GVM] and germinal vesicle breakdown [GVBD]) decreased along the spawning period (see below), except for GVBD in May to July 2018 (Supplementary, Fig. S2D). This high variation in July 2018 was due to one female in early GVBD and another female in late GVBD.

Concentrating on the spawning period, mean cODν_i increased in a smooth, likely slightly concave way from PVO2 to PVO4c (Fig. 4) but the growth in size picked up quickly from cortical alveoli (CA) to GVBD following an apparent linear trajectory up to GVM (Fig. 4, insert). Detailing the underlying phase-specific changes in mean cODν_i as a function of ORC, noticeable differences existed among and within PVOs and VOs (Supplementary, Figs. S12, S13). The mean cODν_i of PVO2-4a phases were rather stable whereas PVO4b-c oocytes displayed a more complex pattern (Supplementary, Fig. S12). Mean cODν_i at the CA phase was approximately similar before and during spawning (Supplementary, Fig. S13A) whilst oocytes in VO1 clearly became smaller as spawning progressed (though with one outlier; Supplementary, Fig. S13B). Mean cODν_i of VO2 and VO3 showed a rather mixed picture (Supplementary, Fig. S13C, D).

Switching to studying smoothed OSFDs across individuals (Fig. 5) (OD ≥ 100 µm), the frequency density of oocytes in the 230–800 µm range became gradually dampened as spawning progressed, see also weak indications of the same pattern following model smoothing restriction in OD to OD ≥ 230 µm (Fig. 4, Supplementary, Fig. S14). Spent ovaries (ORC4), except for six females, consisted only of PVOs (Fig. 5).

Oocyte packing density and oocyte development

Mackerel showed a highly dynamic line of oocyte production throughout their reproductive cycle. Overall, the number of oocytes per gram of ovary (OPD) declined abruptly from PVO2 and PVO3 (millions g⁻¹) via PVO4a–c (hundred thousand g⁻¹) and, finally, to CA ending with GVBD (a few thousands or hundreds g⁻¹) (Supplementary, Fig. S15). The phases PVO2, PVO3 and, PVO4a were always present in high densities (Supplementary, Fig. S15A, B). Their density showed a dome-shaped pattern, increasing from May to October 2018 (PVO2-3) and to November 2018 (PVO4a), then declining until the end of the forthcoming spawning period (Supplementary, Fig. S15A, B). Oocytes in PVO4b-c phases were also omnipresent, however, the number increased from August until November, i.e. during the macroscopic “resting period” (Supplementary, Fig. S15B). The number of PVO4b tended to be higher than PVO4c during most of the study period (Supplementary, Fig. S15B). The onset of maturation took place in October by the appearance of CA (Supplementary, Fig. S15C). The number of CA continued increasing from October until January, then decreased. Early vitellogenesis started in November, when primary vitellogenic oocytes (VO1) were noticed (Supplementary, Fig. S15C). Similarly to CA, the number of VO1 showed a dome-shaped pattern from October to June (Supplementary, Fig. S15C). The high variation recorded in several oocytes phases in March might be attributed to the lower number of individuals in different phases (n = 5) (Supplementary, Table S2). Mackerel was spawning capable (SC), in our study area, from March until July, when late vitellogenic oocytes (VO2-3) and FOM (GVM + GVBD) phases were present (Supplementary, Fig. S15D, Table S1). At least for 2019, the density of GVM and GVBD oocytes were inversely related, i.e. GVM oocytes declined in number when GVBD increased in number (Supplementary, Fig. S15D). The average OPD of GVM and GVDB was 900 and 460 oocytes g⁻¹, respectively (Supplementary, Fig. S15D).

Relative fecundity

From a more general perspective, temporal patterns in relative fecundity (RF_i) estimates mirrored those for OPD_i, simply because the former is given by multiplying the latter with ovary size (see below). Hence, mean RF_i (i.e. number of oocytes g⁻¹ body weight) showed a similar seasonal trend as OPD_i for almost all oocytes phases, except for PVO2 and PVO3 which presented a rather flat RF_i over the sampling period (Fig. 6A). Overall, the mean figure for these two early PVO phases was > 2000 oocytes g⁻¹ body weight (Fig. 6A). Mean RF_i for later PVO phases (PVO4a–c) was typically < 250 oocytes per g⁻¹ body weight (Fig. 6B). Regarding CA to VO3, a dome-shaped pattern was found from the resting to end-of-spawning period (October 2018 to June 2019) (Fig. 6C). RF_i of FOM (GVM and GVBD) appeared at a higher level in 2019 than in 2018 (Fig. 6D). Trends in total length-based relative fecundity (RF_TLi) compared well with those just outlined for RF_i (Supplementary, Fig. S16), considering here TL_j to be a more resilient body size biometric than W_j ((whole) body weight).

Setting the lower threshold value for PVOs at 230 µm, i.e. at PVO4c (Supplementary, Table S1)—the current result (see above and below)—instead of traditionally 185 µm¹⁸, i.e. at PVO4a (Supplementary, Table S1)—during the enumeration work significantly (paired t-test, p < 0.001) affected the resulting, aggregated RF_i (Supplementary, Fig. S17). Note here that a few females contained only PVO4a (at ORC4) and thereby dropped out in this pairwise comparation (ORC1-3). The corresponding reduction from RF_PVO4a-GVBD to RF_PVO4c-GVBD was around 29 and 20% for 2018 and 2019, respectively. Mackerel obviously ate during the spawning season, seen by no obvious drop in K_n with ORC (Supplementary, Fig. S17), a topic explored further on above.

Quantifying de novo oocyte recruitment

Successive estimates of relative fecundity (RF_ij) from PVO2 to VO3 evidenced that the spawning period (ORC1–4) is a time of most active transfer of one type of oocytes to the next type of oocytes, but not equally applicable to all oocyte phases. There was a sign of an initial decline in respective RF_i PVO2-4b, i.e. from prespawning (ORC0) to early-spawning (ORC1), but then levelling off, from ORC1 to ORC4 (very late- or post-spawning) (Fig. 7A–D). Importantly, this description did not apply to RF_i of PVO4c which showed a pronounced decline with ORC (Fig. 7E), with the subsequent RF_i of CA presenting a resembling, but more chaotic picture (Fig. 8A). RF_i of VO1 followed a right-skewed dome-shaped trend vs. ORC (Fig. 8B), whereas RF_is of VO2 and VO3 were more in line with an on-going fall with ORC (Fig. 8C, D). At the assumingly representative, aggregated level (PVO4c–GVBD), RF_i exhibited a dome-shaped pattern throughout spawning, though with large individual variation at a given ORC (Fig. 9). So, de novo oocyte recruitment is evidently important in mackerel, exemplified foremost for 2019 where the RF_i in question increased by almost 65% between ORC0 and ORC1 (Fig. 9B). For 2018, RF_i increased from ORC1 to ORC2, but for this year missing ORC0 data excluded the possibility to track any initial RF_i change (Fig. 9A). For a so-called “standard individual”, i.e. summing the grand mean of phase_i-specific RF_i, a higher final mean RF_i was registered; compared to the estimates just presented (Fig. 9A, B), the difference ranged from 7% at ORC1 in 2019 up to 102% at ORC4 in 2018 (Fig. 9C, D).

Number of batches

Considering the seemingly representative, prespawning mean RF_i in 2019 (ORC0, 528 oocytes body g⁻¹) (Fig. 9B) plus PVO4c de novo recruiting (160 oocytes body g⁻¹—found by subtracting ORC4 figures from ORC0 figures) (Fig. 7E), i.e. totally 688 oocytes body g⁻¹, divided by batch fecundity (40 oocytes by body g⁻¹)²⁷, the typical mackerel female apparently produced ~ 17 batches in this particular year. Consulting instead the grand mean batch fecundity reported by ICES¹⁸ (~ 30 and 34 oocytes by body g⁻¹ in 2016 and 2019, respectively), the number of batches increases to ~ 20–23. So, this number on batches released per individual is strongly dependent upon the batch fecundity, an issue not pursued further here.

Source: Ecology - nature.com