A first assessment of atresia in the Chilean jack mackerel Trachurus murphyi (Teleostei, Carangidae) from the South-eastern Pacific Ocean

The Chilean jack mackerel Trachurus murphyi support an important fishery in the South-eastern Pacific Ocean and has declined considerably over time. T. murphyi have asynchronous ovarian development with an oceanic spawning (September to January). Despite the economic importance of this resource critical issue such as characteristics and dynamics of atresia, which affect the reproductive output, are unknown. Females of T. murphyi caught in different spawning seasons were analysed to describe patterns of prevalence and intensity of atresia. General linear models (GLM) were performed by using prevalence and intensity atresia as response variables and gonadosomatic index, condition factor, sea surface temperature, postovulatory follicles and month/year as explanatory variables. The batch fecundity and relative fecundity were compared between females showing atresia, postovulatory follicles and non-atretic females. The stages of atresia identified were alpha, beta and gamma. The GLM showed that variations in the prevalence and intensity of atresia were explained for all variables, except condition factor for prevalence, and gonadosomatic index for intensity of atresia. According to our results, the dynamic of alpha and beta atresia could be considered a potential proxy for any disruption of spawning season. The batch fecundity and relative fecundity were lowest compared with other Trachurus spp. This is the first assessment comparing fecundity between atretic and not atretic females for this species.


Introduction
In the South-eastern Pacific Ocean (SEP), industrial fisheries for the highly migratory Chilean jack mackerel Trachurus murphyi Nichols, 1920. The main fishery includes the coast off Peru and Chile, extending from the central coast of Chile to the South-west Pacific Ocean, reaching the coast around New Zealand and Tasmania has been described (Serra 1991, Cárdenas et al. 2009, Vásquez et al. 2013. The peak in the catch was achieved in the 1990s, (ca., 4,600,000 tons) after which the catch for all fleets steadily decreased and for 2019 the South Pacific Regional Fisheries Management Organisation recommended a limit of capture of 591,000 tons (SUBPESCA 2018). The main reason explaining this situation was the strong fishing mortality and potentially unfavourable environmental conditions (Dragon et al. 2018).
In mature females, atresia can occur at any time during the reproductive cycle. Sometime, females delay its reproduction, inclusive for one year, increasing the prevalence and intensity of atretic oocytes. In highly migratory species, such as T. murphyi, it is expected that during the spawning period, some females will have atretic oocytes due to potential energy loss during annual migration. As atresia can have strong consequences in recruitment, the knowledge of the spatio-temporal dynamics of this process is important to assign the status of female in the spawning stock during spawning season. Atresia can have direct effect on the number of eggs that females can produce. Like other reproductive traits, atresia depends on the female's physiological conditions (e.g., age, length, condition) and environmental factors (e.g., temperature). Therefore, the main objectives were (1) to describe the phases of atresia in T. murphyi caught during the post-migration spawning period in order to evaluate the spawning history, (2) to determine the prevalence and intensity of atresia in different spawning periods and its relation with physiological conditions and environmental factors and (3) to determine the impact of atresia in the fecundity.

Ethical statement
Ethical approval was not required for this study, due all specimens were collected as part of research cruises that are performed for the evaluation of spawning biomass (DEPM) in the Chilean jack mackerel, Trachurus murphyi from oceanic spawning area off central Chile (Table 1). Fish arrived dead at fishing port. No samples were collected by the authors; permission was granted to use biological database and gonads of all fish.
In the SEP, Chilean jack mackerel conform a single large population (Poulin et al. 2004, Cárdenas et al. 2009) that can be found in three different habitats (Arcos et al. 2001) according to its life history: a) nursery ground area, b) coastal feeding area and c) a large, main oceanic spawning area off central Chile (Bertrand et al. 2004, Cubillos et al. 2008, Vásquez et al. 2013. Interestingly, otolith chemistry indicates that all adult Chilean jack mackerel come from the same oceanic spawning area (Ashford et al. 2011). Recently, Parada et al. (2017) indicated the presence of a potential new spawning and nursery ground area associated with the seamount region in the Challenger break and the East Pacific ridge (32°S-39°S and 105°-126°W).
Chilean jack mackerel can spawn throughout their whole distribution (Taylor 2002, Gerlotto et al. 2012, and the time extension of the spawning period varied between area and years (Perea et al. 2013). For the main oceanic spawning area from central coast of Chile (32ºS-38°S, 80ºW-92°W) the spawning period extends from September to January when mature fish begin the annual migration from the feeding area to the oceanic spawning area; that extends more than 1,800 km off the coast (Cubillos et al. 2008, Leal et al. 2013. Like other Trachurus species, T. murphyi present asynchronous ovarian development (Leal et al. 2013). George (1995 suggests an indeterminate fecundity for T. murphyi based in the frequency of oocytes size distribution, length of the spawning season as well presence of atresia. A generalized prevalence of atresia and resorption of mature oocytes at the end of the spawning season has been described for fishes with indeterminate fecundity (West 1990). Accordingly, fecundity estimates should be made using the most advances development oocytes (hydrated oocytes) instead the vitellogenic oocytes (Saborido-Rey & Kjesbu 2005). For T. murphyi, is not easy to find females harbouring hydrated oocytes during the spawning season (George 1995, Leal et al. 2013, consequently vitellogenic oocytes could be counted, and in this case, atresia must be quantified (Saborido-Rey & Kjesbu 2005). Atresia is

Sample collection
Mature females (n= 3,650) were sampled along the spawning area off central Chile (30º-39ºS; 71°-88°W), according to the methodology described by Sepúlveda et al. (2009). Briefly, for each successful trawl a set of females were selected at random. Females caught were longer than 23 cm fork length. Fork length (FL, cm) and body weight (BW, g) were measured and gonad tissue samples were collected on board. Besides, sea surface temperature (SST) was recorded at each trawl.
Gonads were processed for histology using standard protocols. The ovaries were fixed in 10% neutral buffered formaldehyde at board of fishing vessels. In the laboratory a cross-section of tissue was sampled from the middle parts of the ovary. Each sample was embedded in paraffin, sectioned transversely with a thickness of 5 μm and stained with haematoxylin and eosin (H+E). The histological methodology was followed Brown-Peterson et al. (2011) and applied for Leal et al. (2013) to describe and characterize the reproductive maturity phases for this species.
The significance of potential differences in FL and BW between spawning periods was analysed using Kruskal-Wallis test. Differences between immature and mature females for FL and BW were analysed using Mann-Whitney test. Gonadosomatic index (GSI) that indicate the tendency of the degree of gonadic maturity in its reproductive cycle, were recorder for each specimen. GSI was calculated as GSI= (GW/BW) × 100 (Zimmermann 1997); where GW is gonad weight (g). Condition factor (K) was calculated as an overall measure of robustness of the fish in relation to reproductive maturity phases. K was estimated as K= (BW/ FL 3 ) x 100 (Zimmermann 1997). The mean values of GSI and K, obtained for the different maturity stages during each sampled period, were compared using the Kruskal-Wallis tests. Potential relationships between GSI and K for the total samples were analysed with Spearman's correlation coefficient. The presence and age of postovulatory follicles (POFs) were identified according to Hunter & Goldberg (1980) and Hunter & Macewicz (1985). The POFs were classified into three age classes: day 0= new POF, indicating that the females spawned approximately 24 hours prior; day 1= spawning occurred between 24 and 48 h ago; and day 2= old POF in which spawning occurred more than 48 hours before being caught.
The prevalence of atresia (PA) was defined as the number of fish containing ovaries with at least one phase of atretic oocytes relative to the number of all individuals examined, expressed as a percentage (Ganias et al. 2008). The presence/absence of atresia was modelled using a generalized linear model (GLM) with a binomial error distribution for the response variable and a logit link function (Ganias et al. 2008). Prior to selection of variables to be included in the GLM for presence/absence of atresia, Spearman's correlations were performed to detect collinearity between the quantitative variables (FL, K, GSI and SST). The explored explanatory variables were FL, K, GSI, POF, SST and month/year (MY). Three models were performed. The first model included as response variable all atresia phases identified (alpha, beta and gamma) and their combinations, the second model included only A-atresia, and the third model included only B-atresia. The selection of the best models was checked with Akaike's Information Criterion (AIC) and Deviance (D 2 ). Additionally, the Receiver Operating Characteristic (ROC) curve was used to evaluate the sensitivity and specificity of the models.
For females with alpha-atresia, the intensity of atresia (IA) was calculated as the number of alpha-atretic oocytes divided by the total number of vitellogenic oocytes (alphaatresia + non-atretic follicles) following Nunes et al. (2011). Simple counting from histological sections was used as a proxy of intensity of atresia (Ganias et al. 2008, Nunes et al. 2011. The IA was evaluated using a generalized linear model (GLM) with a normal distribution for the response variable and a log link function. The explanatory variables included in the model were FL, K, SST and MY.
Batch fecundity (BF) was estimated via gravimetric methods on hydrated females (Hunter & Macewicz 1985b) in order to compare contribution of hydrated eggs in spawning seasons showing atresia or POF. Three female stages were identified: females non-atretic oocytes (n= 71), females with atretic oocytes (n= 41) and females with presence of POF (n= 66). Batch fecundity (FB), defined as the number of oocytes released per spawning event, was determined as the number of hydrated oocytes in actively spawning ovaries. Two pieces of ovary were removed from the middle parts of the gonad and weighed (approx. 0.2 g), and hydrated oocytes were counted. These hydrated females were caught during the spawning period S6 (November and December to 2014). Relative batch fecundity (RF), defined as the number of hydrated oocytes per gram of body weight, was determined by dividing BF by BW. Significance differences in BF and RF amongst three female stages were evaluated using Kruskal-Wallis tests and the Dunn a posteriori test. In this case, effects of the explicatory variables FL, K and female stages for BF and RF were tested with a generalized linear model (GLM) with a Poisson distribution for the response variable and a log link function. GLM analysis and ROC curve were performed in R Software (R Core Team 2016).
Immature females and developing females were observed at low frequency during several seasons spawning (Fig. 1). Conversely, potentially spawning capable females (range: 2-81%) and actively spawning females (range: 8-18%) were recorded in all samples except January 2013, increasing the proportion of regressing (61%) and regenerating (17%) females (Fig. 1). The absence of females in the actively spawning reproductive phase and the high frequency of advanced maturity phases (regressing and regenerating) indicated that the long spawning period ended by January.

Atresia morphology
Three main sequential stages of atresia were observed: Alpha (A), Beta (B) and Gamma (G) (Fig. 3). For A-atresia three morphological stages were identified: initial (A1), intermediate (A2) and advanced (A3) (Fig.  3b, c, d). During A-atresia, follicular diameter diminished gradually. The first signal for A1-atresia was an internal, non-smooth zona radiata, which subsequently showed a small fragmentation (Fig. 3b). Some oocytes may show thickening of zona radiata before their fragmentation. Additionally, the disintegration of the nucleus is evident from dark basophilic staining. At this stage, the oocyte retains its shape and sometimes looks like a healthy oocyte.
In A2-atresia, the disintegration and fragmentation of the zona radiata were evident and broke up into irregularly shaped segments; also, unstained vacuoles were apparent inside the ooplasm (Fig. 3c). For the A3-atresia phase, the follicle had the appearance of beta-atresia, but few yolk globule follicles were evident when compared with the other stages (Fig. 3d). A decreased prevalence was evident for the three A-atresia sub-stages from the initial stage to the advanced one, 67% in A1-atresia (initial), 52% in A2atresia (intermediate) and 24% in A3-atresia (advanced). These sub-stages of A-atresia can also co-occur, where the combination of sub-stages A1-atresia and A2-atresia was the most prevalent (PA= 38%). B-atresia was characterized by the presence of many empty vacuoles and some blood vessels (Fig. 3e). The G-atresia was characterized by yellow-brown granular pigments inside the granulose cells (Fig. 3f).

Prevalence of atresia
Overall and during all months/year were identified atretic oocytes, being Alpha (A)-atresia the most frequently stage observed (Fig. 4). For A-atresia (n= 1,485), the 39% co-occur with a combinations of Beta (B)-atresia, Gamma (G)-atresia, while with B-atresia (n= 107) the 74% of ovaries co-occurs with G-atresia (Fig. 4). Females in regenerating reproductive phases not showed atresia. Alpha-atresia was the most prevalent stage in the remaining reproductive phases, including females with hydrated oocytes that showed A-atresia and B-atresia (Fig.  2). Females in the regressing reproductive phase showed prevalence of atresia >80% due to higher co-occurrence of A-atresia and B-atresia (Fig. 2). Also, 27% of mature females recorded post ovulatory follicles (POFs) in all reproductive phases, except the regenerating phase. Thirtytwo per cent of females exhibited POFs in co-occurrence with atresia, mainly A-atresia. The prevalence of POFs at age 0 was higher in females with hydrated oocytes. The frequency of B-atresia and G-atresia increased with POFs at age 1 and age 2.
The GLM showed that MY, SST and incidence of POF were the variables showing higher effect on PA in the three evaluated models. As FL was significantly correlated with GSI (Spearman's correlations ρ, P < 0.001) was excluded from the analysis. GSI were not significative for explain the presence of atresia in model 2 (Alpha-atresia) while K in all evaluated models (Table 2). In the three models, the proportion of explained variance (deviance) was low but significant. Additionally, the prediction of the models was good (ROC= 0.8) ( Table 2).

Intensity of atresia
Intensity of atresia (IA) was high in all spawning periods (range: 45-77%); January 2013 showed the highest value of IA (92%), confirming that this month corresponded to the end of the spawning period (Fig. 4). GLM showed that MY, SST, POF and K were variables that explained the IA, with MY being the most important variable (Table 2).

Discussion
Trachurus murphyi showed three mains sequential atretic stages (Alpha, Beta and Gamma), like others Trachurus species (Macewicz & Hunter 1993, Karlou-Riga & Economidis 1997, Gordo et al. 2008; being prevalence and intensity of atresia influenced by months/years of spawning seasons. T. murphyi has an extensive spawning season (5 months), can spawn an estimate of 5-14 batches for each reproductive period (Ruiz et al. 2008, Perea et al. 2013, and this condition can explain the co-occurrence of atresia in the different months of the spawning season analysed. Prevalence of Beta-atresia increased strongly at the end of spawning season (January); therefore, the increased proportion of this atresia could be a potential proxy to determine the length of spawning seasons, and it would allow us to detect any change in the current or future reproductive period. During December 2013, the prevalence of atresia (>80%) was higher than observed for the same month in other years, suggesting that spawning season began early this year or was shorter. The turnover rate for the atretic oocytes in Chilean jack mackerel is unknown, but atretic follicles in fishes are thought to be short lived (Hunter & Macewicz 1985a, Kjesbu et al. 1991, Kurita et al. 2003. Turnover rates at environmental temperatures for alpha-atretic oocytes have been estimated as 7-9 days for Engraulis mordax Girard, 1854 at 15-16 °C but with differences between stages of atresia (Hunter & Macewicz 1985b); 4 to 7 days in Atlantic herring Clupea harengus Linnaeus, 1758 for temperatures from 4.2 to 11 °C depending on the month of catch (Kurita et al. 2003); 10 days in cod Gadus morhua Linnaeus, 1758 at 8 °C with a spawning period of 50 days (Kjesbu et al. 1991); and 9 days in sole Solea solea (Linnaeus, 1758) (Witthames & Walker 1995). Understanding trends and dynamics of atretic oocytes in T. murphyi allows us to identify any change in the spawning period, such as a potential skipped spawning (Rideout & Tomkiewicz 2011) or previous spawning events as has been assessed for Mediterranean sardine Sardina pilchardus (Ganias et al. 2008).
T. murphyi have a reproductive strategy to disperse over their spawning oceanic area associated to SST warmer than 15-16 °C (Cubillos et al. 2008). According to Vasquez et al. (2013), the reproductive activity of T. murphyi is highly correlated with the Subtropical Convergence or Subtropical Front (STF), which separates relatively warm and salty tropical waters from colder and fresher Subantarctic waters (Vasquez et al. 2013, Parada et al. 2017. The results indicate that SST is an important variable explaining the presence of atresia in T. murphyi, and eventually SST higher or lower than the optimal temperature for spawning, could generate a mismatch between reproduction and optimal environmental conditions for progeny development (Miranda et al. 2013). As indicated by Hunter & Macewicz (1985a) and Witthames & Walker (1995), spawning events and atresia are temperature-dependent processes, and females can spawn or delay during migration if the temperature regime is not suitable (Ma et al. 1998, Haslob et al. 2013. For yellowtail kingfish Seriola lalandi Valenciennes, 1833 and Sardina pilchardus (Walbaum, 1793), a temperature exceeding the optimal range can cause an aborted spawning event or high prevalence of atresia, respectively (Poortenaar et al. 2001, Ganias 2009, Stuart & Drawbridge 2013.
In T. murphyi, condition factor (K) significantly affect the intensity but not the prevalence of atresia, suggesting that K is a good index for explain the intensity of atresia because a fish in bad condition will have not enough energy to develop the oocytes and consequently will have a higher intensity of atresia (Hunter & Macewicz 1985a). Cubillos & Alarcón (2010) suggested that T. murphyi accumulates energy reserves (autumn-winter) prior to the reproductive process (spring-summer), indicating that this species is capital breeding. As K not showed differences between immature and mature female and the lowest correlation between K with GSI could be an evidence that T. murphyi is an income breeding (i.e., continues to feed during migration) (Bucholtz et al. 2013). To confirm the dynamics of somatic energy, the trends of K should be evaluated during all reproductive cycle of T. murphyi. Besides, the relation of GSI and K should be complemented with the evolution of hepatosomatic index (HSI), which is a very good indicator of total energy reserves and fish condition (Zimmermann 1997, Nuñes et al. 2011. In this study, high values of prevalence of atresia especially the beta-atresia, must be associated with the end of spawning season, which is consistent with an indeterminate type of fecundity (Gordo et al. 2008). In species with indeterminate fecundity such as T. murphyi, the recruitment of oocytes can continue for a long period into the spawning period if adequate food is available (Murua & Saborido-Rey 2003). Atresia might be considered an important mechanism for the adjustment of fecundity through the resorption of oocytes in unfavourable seasons (Abaunza et al. 2003). The batch fecundity (BF) is an important biological variable for indeterminate spawners and is used in the daily egg production method (DEPM) (Hunter & Lo 1997, Zwolinski et al. 2001. In this first assessment, the FL, K, presence or absence of atresia and POF affected BF. Fecundity of non-atretic females differed significantly from POF females, coinciding with Macewicz & Hunter (1993), who indicated that for T. symmetricus, the batch fecundity was lower for recently spawned females than females without evidence of a previous spawning event (POF).
Regarding the relative fecundity (RF), the results showed that T. murphyi recorded the lowest values (69 oocytes g -1 in non-atretic females) when compared with other Trachurus spp., 172-250 oocytes g -1 in T. trachurus (Karlou-Riga & Economidis 1997, Abaunza et al. 2003, 112 oocytes g -1 T. symmetricus (Macewicz & Hunter 1993). The fecundity is known to vary due to nutritional conditions of the spawner (Kurita et al. 2003), but during the spawning season analysed (November-December 2014) K did not show a significant relationship with RF, suggesting that K is not a good proxy for RF. This result agrees with those for Trachurus trachurus, where K and lipid content are not reliable indices as proxy for fecundity (Van Damme et al. 2014). In T. murphyi, are needed further studies on seasonal dynamics of atresia through the whole maturation cycle, and thus be able to understand the trend of atresia with the body condition (i.e., GSI, K and HSI), environmental variables (i.e., temperature) and fecundity of this specie during the whole spawning season.