Spatial variability of life-history parameters of the Atlantic chub mackerel ( Scomber colias ) , an expanding species in the northeast Atlantic

Summary: Atlantic chub mackerel is a pelagic species present in the Atlantic Ocean that in recent decades has expanded northwards in the eastern Atlantic. Fish samples were collected in scientific surveys and commercial catches between 2011 and 2019. We analysed the geographical variation of the biological parameters (age, length, weight and condition), as well as the length-weight relationship, maturity-at-length and spawning season onset and duration in five geographical areas (from south to north): the Canary Islands, Gulf of Cadiz, western Portuguese coast, northwestern Spanish coast and Canta-brian Sea. The influence of sea surface temperature (SST) on fish length was modelled as a potential driver of geographical variability. All biological parameters increased progressively northwards, while the spawning season was delayed and prolonged with increasing latitude, from January in the Canary Islands to May-August in the Cantabrian Sea, when SST was between 15°C and 19°C. SST had a positive effect on length in three study areas and a negative one in two of them, suggesting that each group is at a different position within their thermal tolerance range. Deviance from the geographical pattern of some biological parameters in the Gulf of Cadiz suggests that it could be a hinge or mixing zone between Atlantic African, Mediterranean and Atlantic Iberian population components.


INTRODUCTION
Distribution shifts of marine fishes towards polar areas due to global warming have been reported or predicted in the northeast (Perry et al. 2005, Rjinsdorp et al. 2009) and northwest Atlantic (Murawski 1993), in Australian waters (Cheung et al. 2012 and references therein) and in the Pacific Ocean (Alabia et al. 2018, Cheung et al. 2015. Moreover, changes in marine biodiversity have been foreseen as a result of climate change (Cheung et al. 2009). Variations in temperature, pH and oxygen dissolved in the ocean affect the physiology and dynamics of all marine organisms (Pörtner et al. 2005, Rijnsdorp et al. 2009, Ottersen et al. 2010 and, together with distributional shifts, alter trophic structures, the composition of marine communities and thus ecosystem balances (Vergés et al. 2014). While any alteration in marine ecosystems will have a major impact on the services they provide, it is especially evident that changes affecting species exploited by fisheries will have a direct impact on the economy and human health because of the importance of fisheries in the economic development of coastal communities and in the provision of healthy food (Hollowed et al. 2013).
Atlantic chub mackerel (Scomber colias, Gmelin, 1789) is a medium-sized migratory coastal pelagic fish distributed in warm and temperate waters on both sides of the Atlantic Ocean (Castro and Santana 2000), including the Macaronesian archipelagos and reaching the Mediterranean and southern Black seas (Collette and Nauen 1983, Whitehead et al. 1984, Collette et al. 2011. Although for many years the Atlantic chub mackerel was considered the same species as the Indo-Pacific chub mackerel (Scomber japonicus, Houttuyn 1789), morphologic and genetic studies carried out in the first decade of the 20 th century demonstrated that they are two different species (Infante et al. 2007, Catanese et al. 2010, Cheng et al. 2011. In the Canary Islands (FAO area 34.1.2), small pelagic fish are targeted by the artisanal purse-seine fleet, and the Atlantic chub mackerel is the species most commonly caught all year around (FAO 2020a). In the Gulf of Cadiz (FAO area 27.9.a.s), Atlantic chub mackerel is jointly with anchovy and sardine one of three target species fished by the purse seine fleet operating in this area. Additionally, a specifically dedicated seasonal purse-seine fishery targets the species during the second and third quarters of the year, supplying the regional canning industry and providing food for caged bluefin tuna. On the western Portuguese coast (FAO area 27.9.a.c), the northwestern Spanish coast (FAO area 27.9.a.n) and the Cantabrian Sea (FAO area 27.8.c), Atlantic chub mackerel has gained economic importance in recent decades, becoming an important resource for the fishing fleet, especially purse seiners, which are responsible for the bulk of the total landings of this species (ICES 2020(ICES , 2021a. In these last three areas, increasing landings of Atlantic chub mackerel have coincided with the decrease in European sardine (Sardina pilchardus) landings and restrictions on Atlantic mackerel (Scomber scombrus) catches.
Despite its wide distribution, 90% of the Atlantic chub mackerel catches in the northeast Atlantic proceed from northwest African waters (FAO 2020b). More northerly, in the Atlantic Iberian waters, landings have increased significantly in the most recent years (ICES 2020). This increase has been associated with a higher availability of the fish, as it has expanded its distribrution northwards (Martins et al. 2013, Punzón et al. 2016, ICES 2020. This phenomenon has been suggested to be probably driven by the increase in the sea temperature (Costoya et al. 2015, Jurado-Ruzafa et al. 2019) as a result of climate change (Tasker 2008, Reid andValdés 2011). This distribution expansion makes Atlantic chub mackerel an interesting case study for the analysis of the impact of climate change on exploited species. In fact, Atlantic chub mackerel could compete in the new habitats with other important commercial pelagic species such as the congener Atlantic mackerel (S. scombrus), horse mackerel (Trachurus trachurus), sardine (Sardina pilchardus) and anchovy (Engraulis encrasicolus) (Bachiller and Irigoien 2015, Garrido et al. 2015a, b, Veiga-Malta et al. 2019, especially in areas where its presence has increased recently. This competition could affect the fishing yields as well as the pelagic ecosystem balance. Genetic, morphologic and parasite studies have shown significant differences between Atlantic chub mackerel from the western Atlantic and the eastern Atlantic (Scoles et al. 1998, Roldán et al. 2000, Costa et al. 2011, supporting the existence of two different populations at either side of the Atlantic Ocean. However, whether this species constitute a single stock or not in the eastern Atlantic is still unknown, as the studies performed in this area provide conflicting results (Scoles et al. 1998, Mele et al. 2014, Muniz et al. 2020). The information available on the northeast Atlantic suggests that in both African and European waters the species migrates from the wintering areas (mainly located in Mauritanian waters, southern Portugal and the inner part of the Bay of Biscay) towards northern waters in summertime and, in the case of the Bay of Biscay, towards the western Iberian Peninsula (ICES 2021a). However, knowledge regarding population dynamics, life-history traits, migration patterns and connectivity of Atlantic chub mackerel in the northeast Atlantic and environmental drivers of their variation is scarce and fragmented. Therefore, large-scale studies such as the preseent one are essential to understand not only the population structure of Atlantic chub mackerelin in the eastern Atlantic but also the influence of climate change on the dynamics of this expanding commercial species and, subsequently, on the ecosystems it colonizes and the fisheries targeting it.
The aim of this study was to analyse the spatial variation of the biological parameters of Atlantic chub mackerel between the traditional areas of distribution (Canary Islands and Gulf of Cadiz) and those of recent expansion further north on the Atlantic coast of the Iberian Peninsula (W Portuguese Coast, NW Spanish waters and Cantabrian Sea), as well as the role of sea surface temperature (SST) as a climate change-linked driver of life-history traits. The results of the present work will be useful for the future assessment of the species in European Atlantic waters, which has been promoted by ICES (2020).

MATERIAL AND METHODS
The study area covered most of the distributional range of Atlantic chub mackerel in the northeast Atlantic ( Fig.1), from the Canary Islands (FAO area 34.1.2) to the Cantabrian Sea (FAO area 27.8.c). Table 1 contains information about the characteristics of the sampling procedures in each study area.
The Spanish Institute of Oceanography (IEO, CSIC, Spain) and the Instituto Português do Mar e da Atmosfera (IPMA, Portugal) collect biological data of Atlantic chub mackerel (length, weight and maturity) in the Atlantic Iberian waters from regular (monthly or quarterly) sampling of the commercial fleet (including landings, observers on board and biological sampling in the laboratory), as well as from the acoustic and demersal research surveys carried out within the EU Data Collection Framework. In the Canary Islands, the IEO monitors fleet activity and collects biological information on Atlantic chub mackerel monthly from commercial landings exclusively. For the present study, biological information on Atlantic chub mackerel collected between 2011 and 2019 from the monitoring programmes mentioned above was analysed.
For all collected specimens, total length (TL), total weight (TW) and sex were recorded. TL was recorded to the nearest 0.1 cm and TW to the nearest 0.1 g in the Canary Islands, 0.01 g for laboratory sampling and 0.5 g for scientific surveys in the remaining areas. Maturity stages were determined through macroscopic examination of the gonads in all areas. In the Canary Islands, maturity classification was based on the Holden and Raitt (1974) maturity key (I, immature; II, maturing virgin or resting; III, maturing; IV, spawning; V, spent), while in the remaining ar-  Walsh et al. (1990) maturity key (I, immature; II, maturing; III,  mature; IV, spawning; V, partial post-spawning; VI,  final post-spawning).
More detailed information about the sampling schedules is provided in the supplementary material, including quarterly length distribution of sampled individuals in each study area (Table S1, Fig. S1).
In all study areas otoliths were collected from both males and females. Whole otoliths were mounted on black plastic slides, concave side up, covered with transparent resin and observed with a binocular microscope under reflected light for age estimation. Age readings from otoliths sampled from fish caught in the Canary Islands did not provide consistent results, so Canary data were not considered in the analyses involving age.
Since no sexual dimorphism has been found in this species (Techetach et al. 2019), all analyses were performed for both sexes combined, except for the study of the onset and duration of the spawning season (see details below).

Length-weight relationship
The length-weight (TL-TW) model used was where TW i and TL i are the total weight and length of the individual i, respectively, N the total number of individuals, and a and b the parameters to be estimated. The model was linearized taking logarithms (with base 10) on both sides, log 10 (TW i ) = log 10 (a) + blog 10 (TL i ), for fitting in R software (R Core Team 2021) using the common function lm for linear models.
In addition to TL, other covariables can be included in the model predictor. In particular, the TL-TW parameters a and b can be specific for each of the K categories of a factor (Note this K has not any relationship with Fulton's condition factor). Then the model including the interaction between length and the factor is formulated as follows.
where and are the a and b estimates, respectively, for category j (j=1,…,K) of the factor covariable, and the variables I j , j=1,…,K are binary dummy variables which take the value of 1 if individual i corresponds to category j. Then, the model leads to a and b specific estimates for each category of the factor covariable. In the current analysis, the model has been implemented for estimating a specific TL-TW relationship in each study area.

Condition
Condition was approached using Fulton's K condition index (Fulton 1902) to facilitate comparison with previous studies that used this proxy of condition.

Spawning season
The spawning period was determined for each area based on maturity stage data of females, as males sampled for the present study were in the spawning capable stage almost all year around. Furthermore, the existence of spawning capable females is the limiting factor for effective breeding. For the analysis, the proportion of spawning capable over the total number of mature females was the selected index. In the case of the Canary Islands, only females in stage IV (Holden and Raitt's scale) were considered as spawning capable females, whereas in the rest of areas, they corresponded to females in stages IV and V (Walsh's maturity scale). Spawning season was defined as the period of consecutive months when the proportion of spawning capable females was higher than 25%.

Length at maturity
The percentage of mature females at length class was fitted to the logistic equation: where is the predicted mature proportion, a and b the estimated coefficients of the logistic equation and TL the total length.
The length at which 50% of specimens are mature (length at first maturity, L 50 ) was estimated as the minus ratio of the coefficients (-a/b). All maturity ogive parameters were estimated using the sizeMat (Torrejon-Magallanes 2020) package of the R software (R Core Team, 2021), applying a frequentist generalized linear model with 500 iterations in the boostrap resampling.

Geographical variability
Comparisons of length, age, weight and condition between study areas were delivered by the Kruskall-Wallis rank sum test (Kruskal and Wallis 1952; kruskal.test function of the R software), because variances were not homogeneus between areas.
For post hoc comparisons between group levels, the Wilcoxon pairwise Rank sum Test was used (pairwise. wilcox.test function of the R software) with the Benjamini and Hochberg (1995) method for multiple testing corrections. These functions are from the stats R package (R Core Team, 2021).

Temperature effects
To explain the potential geographical differences in biological parameters between study areas, SST was chosen as an explanatory environmental variable because it is directly linked to global warming. SST data for the study areas were downloaded from the SST Aqua MODIS, NPP, 4km, Daytime (11 microns), 2003 dataset at NOAA West Coast Regional Node ERDDAP data server (https://coastwatch.pfeg.noaa.gov/erddap). After data download, a 0.25×0.25-degree grid was used to average the SST, and subsequently the monthly averaged SST was assigned to the centroid of each grid cell. The R script for data download and averaging replication is available at https://git.csic.es/jtornero/erddap-sst-download.
A correlation analysis showed that TL and TW were highly correlated and that the condition factor, K, was also correlated with both length and weight (Fig.  S2). We used generalized linear models (GLMs; Wood 2017) to model the effect of SST and geographical distribution (study area) on TL exclusively, assuming that their variability would be reflected directly in TW and partially in K, because these variables were correlated.
GLMs are an extension of linear models for which the distribution of the response variable can be other than Gaussian. For this reason, a link function, g, is required between the linear predictor and the conditional expectation of the response variable Y, µ(X)=E(Y|X), and the GLM is formulated as where X=(X 1 ,…, X p ) are the covariables, µ(X)=E(Y|X) is the conditional expectation of the response variable Y, g is a link function between µ(X) and the linear predictor, and are the unknown model parameters.
A GLM assumes that the response variable follows a distribution belonging to the exponential distribution family; in our analysis, it is assumed that our response variable (i.e. TL) follows a Gamma distribution with a natural logarithm link because it is a strictly positive continuous variable.
In GLM, the effects of categorical variables are considered fork-1 of the J factor levels, with the remaining one being considered the base level. Hence the estimated coefficient of each factor level will indicate the deviation from the value of the base level.
The presence of influence points, outliers that greatly affect the regression estimates, was checked using the influence function of the mgcv package (Wood 2011), which was also used to perform the GLMs.

RESULTS
The TL of specimens collected for this study ranged between 11 and 54 cm (between 6 and 1498 g TW), corresponding to estimated ages between 0 and 14 years ( Table 1). The largest and oldest specimens were collected on the western Portuguese coast and the Cantabrian Sea.
Results of the Kruskal-Wallis test demonstrated that the medians of all biological parameters (age, TL, TW and Fulton's K) were significantly different between the study areas (p<0.01. Fig. 2). In fact, the Wilcoxon test revealed that all variables differed significantly be- Fig. 2. -Boxplot of biological parameters by study area. A, total length; B, age (years); C, total weight and D, Fulton's condition factor (K). Geographical study areas from south to north are 34.1.2, Canary Islands (red); 27.9.a.s, Gulf of Cadiz (orange); 27.9.a.c, W Portuguese coast (yellow); 27.9.a.n, NW Spanish coast (green); and 27.8.c, Cantabrian Sea (blue). Vertical line within the boxes represents the median, boxes represent the inter-quartile range (IQR) or distance between the first (25%) and third (75%) quartiles, the notch represents the 95% confidence interval of the median, the whiskers represent ±1.5 * IQR and dots represent the outliers.  Table 2 shows the basic descriptive statistics of the biological variables analysed in this study. The condition factor K ranged between 0.15 and 4.04. Individuals in the best condition were observed in the northern areas (NW Spanish waters and the Cantabrian Sea), but the geographical trend in K was not as clear as that observed in age, TL and TW. Similar mean K values were obtained in Atlantic chub mackerel from the Cantabrian Sea (27.8.c), the western Portuguese coast (27.9.a.c) and the Canary Islands (34.1.2) ( between 0.82 and 0.84). The highest ( =0.89) and lowest ( =0.75) mean values were found on the northwestern Spanish coast (27.9.a.n) and the Gulf of Cadiz (27.9.a.s), respectively ( Table 2). As expected, TW increased with TL and the values fitted to a power relationship, but this relationship differed significantly between study areas (p<0.01, Fig. 3), even when only the overlapping size range was considered. Though the intercept and slope was significantly different between all study areas (p<0.01; Fig. S4), the length-weight relationship was more similar among the groups from the Cantabrian Sea, the western Portuguese coast and the Canary Islands than for the groups from the northwestern Iberian Peninsula and the Gulf of Cadiz, which were more similar to each other, as observed in the condition factor. The final model predicts that the largest specimens from the northwestern Spanish waters and the Gulf of Cadiz are heavier than those from the other three areas.
Spawning activity was detected in all study areas based on monthly variability in the proportion of actively spawning females, although on the northwestern Spanish coast it was negligible (<25%) compared with the remaining areas. In the Canary Islands the peak of spawning activity was recorded in January, when 58% of females were active. In the Gulf of Cadiz a peak of 70% was recorded in February, on the western Portuguese coast a peak of 61% was recorded in March, on the northwestern Spanish coast a peak of 20% was recorded in April-May, and in the Cantabrian Sea a peak of 27% was recorded in May. Assuming that spawning period corresponds to the consecutive months when the proportion of actively spawning females is greater than 25%, a certain prolongation of the spawning season is also observed towards the north. While in the Canary Islands spawning is concentrated in a single month (January), in the Gulf of Cadiz it lasts three months (February-April) and on the western Portuguese coast it lasts six months (January-June). Spawning activity on the northwestern Spanish coast did not reach the threshold established in this study and in the Cantabrian Sea. Though spawning females were detected from January to September, their proportion was less than 25% only in May and July (Fig. 4). In summary, the peak of spawning was delayed and the spawning season extended northwards. Fig. 3. -Plot of the modelled relationship between the total length (cm) and the total weight (g) by study area. Geographical study areas from south to north are 34. Regarding size at first maturity, significant differences in the maturity ogive were detected between areas in both the intercept and the slope (p<0.01, Fig. 5). Maturity ogives were estimated based on macroscopic maturity classification of ovaries, tso ahus certain degree of innacuracy is expected owing to the difficulty in distinguishing macroscopically between immature and recovering ovaries. The best fit of maturity ogives was obtained in the Cantabrian Sea (R 2 =0.73), while in the remaining areas the goodness of fit was lower (R 2 <0.42), especially in the Gulf of Cadiz (Table 3). However, the proportion of mature specimens was significantly related to length in all areas and fitted to the logistic model (p<0.01), although the adjustment can only be considered really good in the Cantabrian Sea. Fig. S5). Spawning season was detected to take place between 14°C and 21°C in all study areas. More especifically, mean SST during the spawning season decreased progressively from the Canary Islands (20.5°C) to northwestern Spanish waters (14.6ºC), increasing again in the Cantabrian Sea (18.1°C). Our results demonstrated that SST influences total fish length. In fact, our model, which includes the study area, mean annual SST and interaction between them as covariates, explains 33.3% of TL deviance (Table 4). In the Atlantic Iberian waters, the SST-TL relationship showed a positive slope in the Cantabrian Sea (mean SST =15.8°C), which decreased and became negative southwards, coinciding with the increase in SST: 16.4°C on the northwestern Spanish coast, 18°C on the western Portuguese coast and 21°C in the Gulf of Cadiz. Whereas in the Cantabrian Sea and on the northwestern Spanish coast the slope was positive, on the westsern Portuguese coast and in the Gulf of Cadiz it was negative. On the other hand, in the Canary Islands, with the highest mean annual SST (22.1°C), the SST-TL relationship showed the steepest slope ( Fig. 6). All these limitations imply that the results should be interpreted with caution. However, according to our analysis, a progressive increase in the L 50 towards the north was observed, from 15.5 cm in the Canary Islands to 25.2 cm in the Cantabrian Sea, with the exception, once again, of the Gulf of Cadiz (L 50 =24 cm).
The SST in the study areas fluctuated between 10.9°C and 25.7°C (Table 2), with a significant and progressive decrease in mean SST from the Canary Islands (22.1°C) to the Cantabrian Sea (15.9°C) (p<0.01,

DISCUSSION
Our results show a clear latitudinal gradient of biological parameters of Atlantic chub mackerel from the Canary Islands (division 34.1.2) to the Cantabrian Sea (27.8.c), with the exception of the Gulf of Cadiz (27.9.a.S), where the estimated values in general break this trend. Both the age range and mean age increase northwards, leading to an increase not only in length and weight, but also in length at first maturity and condition factor, though in the latter case the progression is not as evident as for the other parameters. The smallest, lightest and youngest specimens were collected in the Gulf of Cadiz. Growth was not estimated in this work because age data were not available for all study areas. However, the differences in length suggest that the age structure in Canary waters is likely to be biased towards younger ages than in the Iberian Peninsula groups (with the exception of the Gulf of Cadiz). Although the sampling in the Gulf of Cadiz was biased towards the second half of the year (Fig. S1), the lengths of the specimens caught in this area were smaller than in the other areas in all quarters of the year, so these differences were not due to sampling bias but rather to the biological attributes of this component of the northeast Atlantic population of Atlantic chub mackerel, as we discuss below.
The increase in size with latitude coincides with Bergman's rule (Bergmann 1847). Several works have attempted to verify Bergman's rule on ectothermic species, with contradictory results for fishes (Belk and Houston 2002 and references therein). Rypel (2014) suggested that the expresion of Bergman's rule in fishes is linked to the species' thermal niches and is only applicable to coldwater species. However, Atlantic chub mackerel is a temperate species (Castro and Santana 2000), so our results suggest that Rypel's hypothesis could also be extended to this type of species or, at least, to those with a wide distribution range, such as Atlantic chub mackerel (Whitehead et al. 1984). Recent studies show that pelagic fish, including Atlantic mackerel, also belonging to the genus Scomber, follow Bergmann's rule, confirming that temperature is one of the most important determinants of body size (Hattab et al. 2021). Similarly, latitudinal differences in length at maturity have been observed in the horse mackerel (Trachurus trachurus) in the Mediterranean Sea (Ferreri et al. 2019), which some authors have associated with primary production and temperature (Bonanno et al. 2016;Basilone et al. 2017). Pörtner et al. (2005) concluded that eurytherms are able to dynamically adjust the range of their tolerance windows according to temperature fluctuations, so adapted eurytherms show better fitness than native stenotherm species. These authors further suggested that natural selection should favour the adapted eurytherm species, which are energy-efficient, grow fast and reproduce successfully. These results are especially relevant when one considers global warming and its combined impact with fisheries exploitation, as the most commonly exploited pelagic species have shown a drastic decline in growth, condition and size (Van Beveren et al. 2014;Brosset et al. 2017). Given the wide distribution of Atlantic chub mackerel, it can be considered a highly adaptive eurytherm, so its ability to colonize new habitats is likely to be high, and it may continue to expand northwards as long as water becomes warmer within its temperature tolerance or preference, so it is able to adapt to global warming even better than native species.
This type of geographical gradients of life-history traits has been reported in other species from both freshwater and marine habitats (Vila-Gispert et al. 2002, Blanck andLamouroux 2006), as well as different life stages, from early ones (Castro et al. 2002, Takahashi et al. 2012 to adults (Huret et al. 2019). Hughes et al. (2017) relate the latitudinal gradient of life-history traits to temperature, while other authors relate it to other environmental processes such as upwelling events (Gertseva et al. 2017), meso-and large-scale oceanographic processes (Castro et al. 2009, Stocks et al. 2014, fishing pressure (Gertseva et al. 2010), food quality and availability (Perrotta et al. 2005) or a combination of several of them (Huret et al. 2019). The reality is that energy trade-offs in fish are complex and depend on the balance between energy inputs and the metabolic costs of maintenance, growth or maturity. Additionally, all these metabolic processes are strongly influenced by the physical and biological environment in which individuals develop their life stages (Kooijman 2009) and by their life strategies. Because of this, linking geographical clines of biological parameters to specific environmental drivers is difficult, especially in widely distributed species such as the Atlantic chub mackerel.
In the case of pelagic fish, latitudinal changes of biological parameters have been documented in several life-history traits. For example, in Arripis trutta from Australia the number of large and old fish was reported to increase northwards, from the cold waters of Tasmania to the warm waters of southern Queensland, although the initial growth rate was higher in the southern areas (Hughes et al. 2017). Furthermore, Huret et al. (2019) observed an increase in growth rate, body size and length at maturity with latitude in European anchovy, and suggested a combination of factors (food availability, maintenance costs, environmental seasonality and temperature) as drivers of these latitudinal gradients.
The present study demonstrates that geographical differences in age and length structure, condition, spawning and maturity of NE Atlantic chub mackerel are significant, showing a clear latitudinal gradient with small and early maturing individuals in the south (the Canary Islands). As the maturity classification was based on macroscopic staging, the results for this parameter must be interpreted with caution; however, considering that we found significant differences between areas in the other biological parameters, the differences in length at first maturity detected in this study can be expected to be real. Latitudinal gradient is also detected in reproductive phenology, with earlier, shorter and more intense spawning activity in the southern study areas than in the northern ones. Latitudinal variability of reproductive season onset and duration has been re-ported in other fish species too. Huret et al. (2019) reported a delay and shortening of the reproductive season of European anchovy in more northern latitudes, linked to seasonal variability of temperature and food availability and the ability of individuals to accumulate energy reserves. Barbee et al. (2011) found that, in the diadromous fish Galaxias maculatus, spawning season onset occurred earlier at high latitude (New Zealand), where spawning fish were older and larger than on the south coast of Australia. These authors linked this geographical variability of life-history traits to key environmental parameters that vary seasonally, such as temperature, day-light duration and productivity. In our case, latitudinal differences in reproductive phenology could be related not only to temperature and food availability, but also to population structure. However, more analyses are required to corroborate this hypothesis.
The maximum length observed in the present study was between 40 and 54 cm (mean between 21.8 and 31.3 cm), while the maximum age recorded was between 6 and 14 years (mean between 1.5 and 2.8 years). These results are in accordance with the observations of previous studies (Velasco et al. 2011, Jurado-Ruzafa et al. 2021, Navarro et al. 2021b.
Mean Fulton's K ranged between 0.82 and 0.89 considering all study areas, except in the Gulf of Cadiz, where K was slightly lower (mean = 0.75). In Atlantic chub mackerel, condition factor is significantly related to TL (Fig. S6). Therefore, the low values observed in the Gulf of Cadiz may be due to the small size of specimens from this area. Fulton's K values between 0.7 and 1.1 have been reported for the Atlantic chub mackerel from the Atlantic Portuguese coast (Alves 2016, Santos et al. 2017, Barboza et al. 2020, suggesting that the mean condition indices obtained in the study period and areas are in the lower part of this range. Fish body condition is closely related to feeding, metabolic rates and the capacity to store energy reserves, i.e. to energetic trade-offs (Saborido-Rey and Kjesbu 2005), all of which are modulated by environmental factors such as temperature or prey availability and diversity, i.e. marine community biodiversity. Changes in the pelagic ecosystem induced by global warming, fishing pressure or any other environmental driver may lead to a decrease in the health status of native pelagic fishes (Shephard et al. 2014, Muhling et al. 2017. The biological characteristics of Atlantic chub mackerel from the Gulf of Cadiz (young, small and low-condition specimens) should be also addressed from the environmental-induced point of view. In this area, there has been an increase in landings of Atlantic chub mackerel since 2007. While this increase coincides with the recent expansion of the species throughout Ibero-Atlantic waters, it also coincides with the decline of the Ibero-Atlantic sardine stock, which led to severe restrictions on fishing opportunities for this species. These limitations may have led to a change in the behaviour of the fleet, which may have redirected its effort towards Atlantic chub mackerel to compensate for the decline in sardine catches (ICES 2021a). This increase in fishing effort could be behind the juvenescence of Atlantic chub mackerel in this area. However, as explained below, the Gulf of Cadiz could also be either a mixing area between the Atlantic and Mediterranean populations or a hinge area in the latitudinal and longitudinal biological gradient observed for Atlantic chub mackerel in the NE Atlantic and Mediterranean.
Regarding the onset and duration of the spawning season, most of the existing studies of Atlantic chub mackerel are based on the analyses of the monthly prevalence of active females and the gonado-somatic index, showing some variability among regions. The spawning season in North African waters has been reported from December to March in Morocco (Techetach et al. 2010, Wahbi et al. 2011 and from January to March in Mauritania and Senegal (ICES 2020). In the Atlantic islands the spawning season of Atlantic chub mackerel has been reported from November/December to March in the Canary Islands Pajuelo 1996, Jurado-Ruzafa et al. 2021), from January to April in Madeira (Vasconcelos et al. 2012) and from March to July/ August in the Azores (Carvalho et al. 2002). Finally, in the Atlantic Iberian waters spawning season has been reported from December to March in waters of southern Portugal waters (ICES 2020), from December to May in western Portuguese waters (Nunes et al. 2019) and, lastly, from March to July on the northwestern Spanish coast and the Cantabrian Sea (Villamor et al. 2017Navarro et al. 2021a). The spawning period of Atlantic chub mackerel in the Mediterranean Sea has been reported from April to August (Rizkalla 1998, Cengiz 2012, Allaya et al. 2013. Our results show a clear latitudinal gradient of spawning season onset and duration that matches with previous studies. Reviewing all this information together, there seems to be a latitudinal trend from the Strait of Gibraltar towards the poles, but also a longitudinal one from the Strait of Gibraltar towards the Mediterranean Sea and the central Atlantic. Changes in reproductive phenology are usually associated with environmental drivers (Rogers andDougherty 2019, Slesinger et al. 2021), although they could be related to the age structure of populations, as spawning season onset and duration may change depending on the age of reproductive specimens (Lambert 1987, Wright andTrippel 2009). However, with the available data we cannot distinguish whether the phenological differences observed in this study are due to environmental drivers or to the differences in the population structure of each study area. This is especially important in the context of global warming, which is leading to distribution shifts of fish species because reproductive phenology affects population resilience (Lowerre- Barbieri et al. 2017), and more attention should be paid to this issue in the near future.
Regarding the size at first maturity (L 50 ) of Atlantic chub mackerel, our results based on macroscopic observations showed a latitudinal trend, with the lowest value in the Canary Islands (15.5 cm) and increasing progressively towards the north (17.5 cm on the western Portuguese coast and 25.2 cm in the Cantabrian Sea). Once again, the L 50 values for the Gulf of Cadiz (24 cm) break with this trend, despite the fact that in general the individuals in this area are the smallest. Previous studies reported higher values in the Bay of Biscay (L 50 =29 cm; Lucio 1997) and in Portuguese waters (L 50 =27 cm; Martins 1996), although more recent investigations showed L 50 values in the Bay of Biscay similar to those reported here (25 cm; Villamor et al. 2017, Navarro et al. 2021a). In western Portuguese waters, Nunes et al. (2019) reported an L 50 equal to 22.6 cm and 19.2 cm for females and males, respectively; while Gonçalves et al. (2016) in southern Portugal reported an L 50 for both sexes combined of 18.6 cm. These Portuguese values are slightly higher than our estimations (17.5 cm). In the case of the Canary Islands, our estimate of L 50 is the smallest one and lower than previous estimations reported in the same area (18-19 cm; Pajuelo 1996, Jurado-Ruzafa et al. 2021). This difference is likely due to differences in the data used to calculate the L 50 , which in previous studies included post-spawning and spawning specimens collected exclusively during the spawning period. In Atlantic Moroccan waters, L 50 values increased from 23 cm in the north (Techetach et al. 2010) to 25 to 27.5 cm in the south (Wahbi et al. 2017). High values of L 50 were reported in the Azores (27.8 cm; Carvalho et al. 2002), followed by Madeira (22 cm; Vasconcelos et al. 2012) and the Mediterranean Sea (16.8 and 19 cm; Cengiz 2012, Cikeš and Zorica 2012, Techetach et al. 2019. Spatial patterns of size at first maturity do not seem to be as clear as spawning onset and duration, although the L 50 values presented here should be interpreted with caution, as stated above. Nevertheless, certain latitudinal and longitudinal trends of maturity patterns can be glimpsed and should be corroborated by histological studies. In any case, the spatial differences observed in size and age structure and condition suggest the existence of different life-history patterns, so spatial differences in maturation can also be expected. This aspect would be especially relevant for stock assessment purposes. The spatial population structure of the Atlantic chub mackerel in the eastern Atlantic, the migration processes through the distribution range and the connectivity, including the Mediterranean Sea, remain unknown (ICES 2021a). The geographical differences in life-history traits estimated in this and previous studies suggest that there may be at least two population components: one in the Canary Islands and one in the Iberian Peninsula. Furthermore, the Iberian unit could include two subcomponents in the Cantabrian Sea (Bay of Biscay) and the Portuguese coast, whose degree of connection could vary depending on the abundance and migration intensity of Atlantic chub mackerel in each area (ICES 2020 and references therein). Additionally, the Gulf of Cadiz could be considered a hinge area, or even a mixing area between the northeast and central east Atlantic populations of Atlantic chub mackerel as well as between the notheast Atlantic and Mediterranean stocks, as suggested by ICES (2021a). The results of the growth analysis carried out by Velasco et al. (2011) support the hypothesis of a mixing area, as they found no differences in growth between the Gulf of Cadiz and the Alboran Sea, suggesting that the Strait of Gibraltar is not a geographical barrier for this species. However, in general, results of studies targeting the stock structure of Atlantic chub mackerel in the northeast Atlantic and the Mediterranean Sea show differences depending on the methodological approach. Some genetic studies showed no significant differences between the Atlantic chub mackerel from the Mediterranean and eastern Atlantic areas (Scoles et al. 1998, Zardoya et al. 2004). However, significant regional differences have been found between smaller areas in the east Atlantic Ocean, including the Mediterranean Sea and Macarronesian islands, based on the analysis of morphology, meristic characteristics (Allaya et al. 2016, Bouzzammit and El Ouizgami 2019, Muniz et al. 2020) and associated parasites (Mele et al. 2014). Muniz et al. (2020), based on otolith morphology analysis, suggested the existence of one group in the northeast Atlantic islands (Azores, Madeira and the Canary Islands) and another one on the Iberian Portuguese coast. Correia et al (2021), according to microchemical analysis of the same otoliths, suggested the existence of at least four stock components in the northeast Atlantic (Azores, Madeira, Canary Islands and Iberian Portuguese coast). There is no information on the migratory behaviour of this species in the study area, although by analogy with the previous study on the species in northwest African waters (García 1982), we assume that the Atlantic chub mackerel performs annual reproductive migrations all along its distributional range. If so, connectivity between the different population components may exist, but the extent of this migration and therefore the degree of potential connectivity remain unknown. Many of these studies have been approached on a local or regional scale, and thus lack the global vision required to investigate the population structure of a species as widely distributed as the Atlantic chub mackerel. Jansen et al. (2013) proposed for the northeast Atlantic mackerel (S. scombrus) that the population structure is more a dynamic cline rather than connected contingents, and this could also be the case of the Atlantic chub mackerel. However, more studies targeting migration behaviour, genetics, morphometry and meristic characterization from a global perspective should be carried out to confirm meaningful biological management units for assessment purposes.
Considering that the increase in abundance of Atlantic chub mackerel in the Atlantic Iberian waters is assumed to be a relatively recent event (Villamor et al. 2017), it is striking that individuals in the northernmost (i.e. most recently occupied) study areas show the greatest diversity of age and size and the best condition. There are two possible explanations for this: a) the presence of older and larger individuals of Atlantic chub mackerel could be due to the fact that they have a greater capacity for movement and therefore a greater capacity for expansion to new suitable habitats in a global warming scenario; and/or b) fishing pressure or predation could be less intense in the north than in northwest African waters, reducing the presence of larger and older individuals in this area. With the information available, we cannot determine the causes of the age and size structure observed in each study area. According to Saunders and Tarling (2018), greater body size in mesopelagic fishes is a necessary attri-bute to reach colder regions, which supports our first hypothesis that the largest individuals are those with the greatest migratory capacity.
In most studied fish species, great size, age and condition are associated with high reproductive potential owing to the existence of maternal effects (Green 2008). In addition, a prolonged reproductive season increases the probability of offspring finding optimal environmental windows for their development, thus increasing their survival (Mertz and Myers 1994). Given the biological characteristics of the groups analysed in this study and the duration of the reproductive season in each area, the geographical variability of the reproductive potential of this species and its impact on stock resilience should be analysed in the future. This variability may be a mechanism of adaptation to the environmental conditions (temperature, productivity and turbulence) that could favour or jeopardize offspring survival and hence recruitment.
Our model indicates that SST does indeed play a relevant role in the size of Atlantic chub mackerel and, presumably, on its growth. While on the western Portuguese coast and the Gulf of Cadiz this effect is negative, in the remaining areas it is positive. The rates of physiological processes usually increase with temperature up to a certain limit, above which the temperature effect becomes negative. This temperature effect is also observed in the growth of fish, so at low temperatures growth increases as temperature increases until a tipping point (optimum temperature) when growth decreases with increasing temperature (Jobling 1995). The temperature range at which fish populations exhibit this behaviour changes geographically owing to acclimatation mechanisms (Sunday et al. 2011). Therefore, growth is expected to increase with temperature in areas that are below their optimum temperature and to decrease in areas that have exceeded this threshold. Assuming the total length as a proxy of the growth rate of Atlantic chub mackerel for each study area, and based on the slope of the SST-TL relationship, we could conclude that S. colias in the Cantabrian Sea and on the northwestern Spanish coast is farther from its optimum temperature than it is on the western Portuguese coast, where it is slightly above it, and in the Gulf of Cadiz, where it is clearly above it. In the Canary Islands, where mean annual SST is considerably higher than in the rest of the study areas, the effect of SST on the TL of Atlantic chub mackerel was the most markedly positive. We do not know the reasons behind this relationship but it could be related to the comparatively small specimens of Atlantic chub mackerel inhabiting the Canary waters. Further knowledge of this relationship between SST and growth would help to better understand the dynamics and population structure of Atlantic chub mackerel in the northeast Atlantic in order to properly calibrate the assessment and manage this resource adequately under future global warming scenarios.
Many fish species have been reported to be moving to higher latitudes as a consequence of climate change (Perry et al. 2005, Rijnsdorp et al. 2009). One of the most emblematic cases in recent decades is that of the Atlantic mackerel (S. scombrus), whose distribution has expanded northwards and westwards in the NE Atlantic Ocean, reaching the waters of Svalvard and Iceland (Berge et al. 2015) driven by stock size and temperature (Astthorsson et al. 2012, Olafsdottir et al. 2019. Similarly, landings of Atlantic chub mackerel in the Atlantic Iberian waters have increased during the last few decades (Martins et al. 2013, Villamor et al. 2017, ICES 2020, likely associated with an increase in stock size and its northward expansion linked to global warming. The arrival of new species in ecosystems can have a major impact on trophic balances and the dynamics of other species (Cheung et al. 2009, Hollowed et al. 2013, Muhling et al. 2017). The Atlantic chub mackerel is an opportunistic pelagic species that feeds mainly on euphausiids and decapod crustaceans, but also significantly on eggs, larvae and juveniles of other pelagic and demersal fish species (Torres et al. 2013). In fact, Garrido et al. (2015a) identified sardine eggs as an important component of the diet of Atlantic chub mackerel and reported a considerable overlap in the prey spectrum between European sardine and juvenile chub mackerel. Taking this into account, it is probable that the expansion of Atlantic chub mackerel in Atlantic Iberian waters played an important role in the decline of Iberian sardine populations observed in the same period. To corroborate this hypothesis, the trends of both species in the Atlantic Iberian waters in the coming years should be monitored, especially when the sardine population has started to recover (ICES 2021b). Another question that would be worth investigating further is whether the change in the distribution of S. scombrus could have reduced the occupation of certain pelagic niches in Atlantic Iberian waters, which could be being occupied by warmer-water species such as the Atlantic chub mackerel. Furthermore, its expansion in Atlantic Iberian waters may also impact on predators, as it has been reported as an important prey for other commercial species such as European hake, sharks, bluefin tuna and marine mammals (Torres et al. 2013, Varela et al. 2013, Giménez et al. 2017.
This paper outlines the population structure of the Atlantic chub mackerel in the northeast Atlantic from a broad geographical perspective and provides information on the regional variability of key biological parameters commonly required for assessment, such as length-weight relationships, the maturity ogive and size composition. However, there are still many gaps to be filled. In relation to population structure, it would be desirable to extend the study to North African and western Mediterranean waters in order to carry out not only analyses of life-history parameters such as growth and maturation, but also other stock identification studies such as genetic, morphometric and parasite analyses with a wider perspective. Similarly, it would be necessary to analyse how the biological differences detected between the study areas affect the productivity and resilience of the species. This would require time series of fecundity, egg and larval quality and recruitment. Additionally, analysing biodiversity changes in the pelagic ecosystem of the study areas would allow us to understand the impact of new species arrivals on fisheries yields. Ulti-mately, from an ecological point of view, this species is a promising case study for monitoring the impact of climate change on pelagic ecosystems.

FUNDS
This study was co-funded by the European Union through the European Maritime and Fisheries Fund (EMFF) within the national programme of collection, management and use of data in the fisheries sector and support for scientific advice regarding the Common Fisheries Policy.

ACKNOWLEDGEMENTS
We thank all the crew members, scientists and technicians involved in the scientific surveys, fish sampling and laboratory work: without them this work would not have been possible. Our gratitude also goes to Ricardo S. Leal for his guidance in obtaining the SST satellite data. The INVIPESCA network is thanked for organising the fifth edition of SIBECORP and giving us the opportunity to present and publish our results in this special issue. The IMPRESS project and its participants are thanked for contributing to the discussion on the impact of these results on the future analytical assessment of Scomber colias in the northeast Atlantic.