Identifying and understanding the main factors that affect the spatial distribution of marine organisms is important to evaluate the current distribution patterns and predict potential impacts of human activity (Lasram et al., 2010; Morfin et al., 2012). Changes in species distributions may be driven by environmental seasonal variation, as well as by prey availability (Carney, 2005; Morfin et al., 2012). Seasonal variations of life cycle events in animals and plants characterize the seasonal phenology and long term dynamics of a species, which is one of the most sensitive indicators to environmental changes (Cormon et al., 2014; Scranton and Amarasekare, 2017).

Seasonality takes place in all marine ecosystems, but their duration and intensity varies according to the geographical area, in general terms being more evident in tropical waters than in temperate waters (Valiela, 1995). Nevertheless, at a more regional scale, the western Mediterranean Sea is characterized by having a high seasonality (Coll et al., 2010); with a marked thermocline in summer and a lack of nutrients on the surface layer versus a mixing of the water column and an upraise of nutrients to the photic layer during winter (Margalef, 1985). Variations in environmental factors drive the distribution of nutrients and primary production, in fact phytoplankton blooms in this area peak in winter-spring coinciding with the stabilization of the water column and again in autumn, when the waters start to mix again (Estrada, 1996; Salat et al., 2002). The seasonal changes in environmental and biological variables in the entire Mediterranean basin is a well-studied subject (Psarra et al., 2000; Bosc et al., 2004; Zveryaev, 2015) and some studies have also been done on the effect of seasonality at different levels of biological organization (Gaertner, 2000; De Souza et al., 2011; Puerta et al., 2016). Nevertheless, the analysis of the interplay of seasonality with higher level trophic organisms tends to be scarce and patchy. Environmental factors have shown to affect species distribution (Katsanevakis et al., 2009; Pennino et al., 2013; Navarro et al., 2015; Puerta et al., 2015) and these environmental factors also show strong intra-annual variation in the Mediterranean Sea (Salat et al., 2002), therefore it could be relevant to take seasonality into account when analyzing species spatial patterns in this basin.

Several studies have used species distribution models (SDM) (Guisan and Zimmermann, 2000) to investigate the spatial patterns of marine species and most agreed on biological, environmental and human related variables affecting species distributions in the Mediterranean Sea (Katsanevakis et al., 2009; Navarro et al., 2015). On this context, several Mediterranean commercial marine species, including the European hake (Merluccius merluccius), present seasonality on their spatial distribution (Demestre and Sánchez, 1998; Paradinas et al., 2015; Sion et al., 2019; Lloret-Lloret et al., under review). European hake is one of the most important demersal target species for commercial fisheries in the Mediterranean basin (Sánchez et al., 2007), but it is reaching overexploitation levels in numerous areas (Lleonart and Maynou, 2003; Fernandes et al., 2017) to the point that it is currently listed as Vulnerable species by the International Union for the Conservation of the Nature (IUCN)(Di Natale et al., 2011). Therefore, achieving a better understanding of spatial patterns of European hake and how they relate to seasonal processes could be of use to inform ecosystem-based management actions that result on more sustainable fisheries.

Due to the ecological and economic importance of European hake, different published studies had been focused on its spatial distribution (e.g., Demestre et al., 2000; Abella et al., 2005; Druon et al., 2015; Sion et al., 2019). These studies indicated that European hake occupies a wide bathymetric distribution range (from 20 to 1,000 meters), inhabiting the shelf and upper slope in the Mediterranean Sea (Fisher et al., 1987; Demestre et al., 2000; Orsi-Relini et al., 2002). Oceanographically, this species is mainly present in cooler bottom waters (from 9.65 to 19.23°C) (Maravelias et al., 2006). Seasonal variations in the density of European hake have been recorded in some areas of the Mediterranean Sea and the Atlantic Ocean with movements to deeper and cooler waters during summer (Fariña et al., 1997; Maravelias and Papaconstantinou, 2006; Maravelias et al., 2006; Lloret-Lloret et al. under review).

SDMs generally focus on abiotic factors, however species spatial patterns are also related to the availability and preference of prey, hence the importance of considering information on feeding strategies when analyzing the spatial distribution of a species (Navarro et al., 2016). In the case of European hake, many studies have analyzed its feeding ecology (Bozzano et al., 1997; Carpentieri et al., 2005; Cartes et al., 2009) but few have directly studied the relationship between trophic behavior and spatial distribution in the Mediterranean Sea (Johnson et al., 2012, 2013).

European hake has been described as an ambush demersal predator (Pitcher and Alheit, 1995), which also feeds on the water column performing nocturnal diel migrations, important for the juvenile individuals (Bozzano et al., 2005; Aguzzi et al., 2015). Trophic studies of this species have generally focused on immature individuals (Oliver and Massuti, 1995; Bozzano et al., 2005) except for a few number that include a wider range of sizes (Bozzano et al., 1997; Carpentieri et al., 2005; Mellon-Duval et al., 2017). Despite these studies, there is still a lack of information on the ecology of European hake, especially regarding variation of the diet with seasonality.

The study of the diet of marine fish normally relies on stomach content analyses (Hyslop, 1980). Although this technique provides quantitative diet composition, it is limited by degradation and ingestion rates, amongst other constraints (Hyslop, 1980). In addition, high levels of regurgitation have been recorded for European hake (Modica et al., 2011), which difficult the analysis of the diet based on stomach content analysis. As an alternative, the use of stable isotopes analysis and isotopic mixing models are additional techniques to examine the diet of marine predators (Davis and Pineda-Munoz, 2016). Despite the extended use of these methodologies, only a small number of studies have used them to characterize the feeding ecology of European hake in the western Mediterranean Sea [e.g., (Sinopoli et al., 2012; Fanelli et al., 2018)] and more specifically in the northwestern Mediterranean Sea [e.g., (Ferraton et al., 2007; Albo-Puigserver et al., 2016; Mellon-Duval et al., 2017; Rueda et al., 2019)].

In this study, we aimed to investigate the seasonal differences in the spatial distribution and trophic habits of European hake in a highly exploited area of the northwestern Mediterranean Sea, considering ontogenetic variations. To analyze the seasonal spatial distribution, we employed a hierarchical Bayesian species distribution modeling approach (B-SDM) and we used the mean weight values estimated from two oceanographic surveys conveyed in winter and summer 2013, as a proxy of body size (Chih-Lin et al., 2010) and consequently age class. We expected that juvenile (i.e., lower mean weight) and adult individuals (i.e., higher mean weight) will distribute differently within the same season and between seasons and these differences might be explained by feeding preferences on top of environmental variables. We included depth and sea bottom temperature (SBT) as environmental variables, in addition to potential preys’ abundance and mean weight distribution as trophic components. Furthermore, we used stable isotope values of δ¹³C and δ¹⁵N to determine the ontogenetic and seasonal changes in diet using Bayesian mass-balanced isotopic mixing models (MixSIAR) (Stock and Semmens, 2016) and corrected standard ellipses area (SEA_c) and Bayesian standard ellipses area (SEA_b) using SIBER – Stable Isotope Bayesian Ellipses in R (Jackson et al., 2011) to estimate niche width and overlap. Then, we examined if spatial differences did coincide with seasonal differences in the trophic habits and plasticity. This is to our knowledge one of the first studies to use this multidisciplinary approach to analyze the spatial and seasonal variations of European hake distribution and trophic ecology, accounting for ontogenetic changes. Specifically, we hypothesized that seasonal changes observed in European hake distribution could be partly due to prey availability, in addition to environmental variability, which could change between the juvenile and the adult stages of the population due to dissimilar trophic preferences or feeding capabilities.

The spatial distribution of marine species is known to be affected by interannual and seasonal environmental factors and habitat parameters, but also by species interactions factors (Gilinsky, 1984; Hixon and Carr, 1997; Carney, 2005; Morfin et al., 2012). In the present study, we applied a multidisciplinary approach that analyses the spatial distribution of European hake, using its main environmental and feeding related drivers explaining European hake distribution. For this, we applied Hierarchical Bayesian species distribution models (B-SDMs) and stable isotope analysis, considering a seasonal and ontogenetic perspective. Overall, our results showed seasonal and ontogenetic differences in the spatial distribution of European hake in the northwestern Mediterranean Sea. In addition to the effect of particular environmental factors, these spatial changes were partially explained by the distribution of the main prey consumed by European hake. We also gathered information about the ontogenetic and seasonal variation in European hake diet to provide further insight into how feeding preferences may affect species spatial occurrence.

Our species distribution models showed that adult individuals had a tendency to be present in deeper areas, with lower bottom temperatures. These environmental preferences have already been described for this species in the Mediterranean Sea (Demestre and Sánchez, 1998; Maravelias et al., 2006; Katsanevakis et al., 2009; Sion et al., 2019). Particularly, bathymetry has repeatedly been reported as a main driver in hake spatial distribution (Recasens et al., 1998; Orsi-Relini et al., 2002; Maynou et al., 2003; Paradinas et al., 2015). The importance of temperature and depth as key factors on SDMs has also been reported to vary intra-annually. For example, in the Aegean Sea (Central Mediterranean Sea) in 2016, the effect of temperature in the distribution of European hake was more pronounced during months of thermal stratification (Yalçln and Gurbet, 2016). Indeed the influence of environmental factors on determining abundance of demersal fish in general are reported to be intensified during summer and autumn in the Aegean Sea (Katsanevakis et al., 2009). Furthermore, a seasonal and ontogenetic migration along the bathymetric gradient was also observed for European hake outside the Mediterranean Sea, for example in the Galician waters (Atlantic Ocean) (Fariña et al., 1997).

Our results showed a spatial segregation between adult and juvenile European hakes, with adults mainly present in the southern area during winter and in the northern area during summer. Ontogenetic spatial differentiation and areas of aggregation for this species have previously been related to life cycle events (Demestre and Sánchez, 1998), nursery areas (Druon et al., 2015), areas with high food availability (Sion et al., 2019) or changes in diet (Carpentieri et al., 2005; Garofalo et al., 2018). In general, it seems that adult females choose deeper areas whereas juveniles prefer the shallower continental shelf (Demestre and Sánchez, 1998). For instance, a study from late spring of 2015 using the same modeling approach than us detected three main hotspots area for European hake nurseries in the northwestern Mediterranean Sea, one of the three located northern of the Ebro Delta, similar to the distribution found here for adults’ hake in winter (Paradinas et al., 2015). In this context, European hake in the western Mediterranean Sea presents a spawning peak during autumn and winter (Maynou et al., 2003), hence the aggregation found close to the Ebro Delta during winter in our study might also correspond to a reproduction/nursery area. Furthermore, changes in the spatial distribution between seasons have also previously been detected in our study area, with individuals intensifying their spatial differentiation during summer and spring (Demestre and Sánchez, 1998).

In addition to the importance of the environmental variables on the distribution of European hake and although bathymetry showed the highest impact in all four models, the distribution of European hake main prey, both fish and crustaceans, also provided valuable information explaining the distribution of hake in the northwestern Mediterranean Sea. Specifically, the spatial distributions and abundance of potential preys were used as indirect proxies of trophic behavior in species spatial occurrence (Navarro et al., 2016). However, just including prey distribution was not enough in this case, as European hake is an ambush predator and its mouth gape is a limiting factor for prey selection and ingestion (this being especially true for juvenile individuals) (Johnson et al., 2012). This is why we also considered the information about potential preys’ size as an explanatory variable. In this regard, our results indicated that European hake’s juveniles in winter were associated with low abundance of relatively large fish preys. The size of these individuals could potentially outbound the mouth gape limitations of the predators, and thus juvenile’s European hake would mostly be feeding on small crustaceans as has been described in previous feeding studies (Mellon-Duval et al., 2017). However, in summer, juveniles are correlated with smaller and more abundant fish prey implying a potential capacity of also ingesting available small fish, in addition to crustaceans. Nevertheless, these results on their own were not sufficient to drawn a robust conclusion and the use of stable isotope analyses complemented the SDMs results providing additional information of European hake seasonal and ontogenetic trophic metrics.

In this context, differences in δ¹³C and δ¹⁵N values pointed out to potential differences in trophic habits of adults and juveniles between winter and summer. Furthermore, the low overlap between the different isotopic niches showed a strong degree of niche differentiation between seasons and stages, also confirmed by the mixing model. More specifically, the wide isotopic niche observed for adults suggests a more diversified feeding behavior, contrasting with the narrow isotopic niche breadth recorded for juveniles. This is in contrast with previous studies, in the same area and in the Tyrrhenian Sea (central Mediterranean Sea), that have described juveniles as being more opportunistic than adults (Modica et al., 2013). Our isotopic results also suggest a seasonal differentiation in adults’ diet, with higher consumption of crustaceans during summer compared to winter. To a certain extent, this relates to our observations of the SDMs, as adults’ hake during summer were correlated with small-size crustaceans. Related to juveniles, the differences in δ¹⁵N values when compared to adults, and the positive trend with body size (Supplementary Figure 5) indicated ontogenetic variations in diet; with the lower values recorded for juveniles suggesting a consumption of lower trophic level organisms. In addition, MixSIAR outputs showed that cluster 3 (formed uniquely of the amphipod Vibilia armata) represented the highest diet proportion for juveniles in winter, followed by cluster 4 (formed exclusively of crustaceans) and the same clusters but in the opposite level of importance dominated for summer. This concurred with previous findings that reported a diet based on crustaceans for juveniles of European hake (Bozzano et al., 1997; Cartes et al., 2004, 2009; Ferraton et al., 2007).

However, partly contrasting the results of the B-SDM, juveniles larger than 20 cm in winter showed a higher proportion of cluster 5 (formed by several crustaceans and small fish, i.e., Sardinella aurita and Engraulis encrasicolus) when compared to juveniles from summer. Consumption of small pelagic fish has also been recorded for this species for individuals as small as 7 cm in nearby areas (Gulf of Lions) (Mellon-Duval et al., 2017). Within this context, the SIA results suggested that summer juveniles may have been feeding on higher trophic levels prey sources, as they showed higher δ¹⁵N values (Navarro et al., 2011) compared to juveniles in winter. However, rather than an increase of piscivorous diet during summer, it could be due to the fact that at this time of the year, juveniles feed more on cluster 4 than cluster 3, which occupied a slightly higher isotopic position. Nevertheless, it is important to consider that changes in δ¹⁵N values can also reflect variations in the baseline rather than in diet, especially when comparing values from different seasons (Costalago et al., 2012; Mellon-Duval et al., 2017). Thus, further and more extensive analysis that include this aspect should be done to verify this seasonal variation in the diet.

Overall, our trophic analysis results were partly consistent with published data (Bozzano et al., 1997; Cartes et al., 2004, 2009; Ferraton et al., 2007), where juveniles showed a crustaceans dominated diet and switch to a more piscivorous diet as they reached larger body size. Here we observed an evident ontogenetic variation in diet, with increased consumption of fish preys as individuals reach intermediate and large body size. We also observed that the species of crustaceans consumed varied with ontogenetic stage and season. However, not all the studies concur on the body length and the reason why this switch occurs. It has been suggested that the diet’s switch corresponds to an increase in individual’s mobility in the water column (Mellon-Duval et al., 2017) along a proportional increase in mouth dimensions (Karpouzi and Stergiou, 2003) that facilitates ingestion of larger preys (Modica et al., 2013). Piscivorous diet has also been associated with the retinal changes and the improvement in vision acuity (Bozzano and Catalán, 2002) that permit European hake detects prey from higher distances and even in more turbid waters. These physiological changes altogether will ease prey selectivity and an intake preference for higher energy preys (Stagioni et al., 2011; Modica et al., 2013).

Overall, our results show that seasonality plays a role when looking at the distribution and feeding behavior of European hake. Seasonality can affect prey availability and thus predators’ diet. In our study, despite diet variations were more distinct between stages, differences between seasons should not be disregarded. Previous studies have also recorded diet variation with seasonality in the Western Mediterranean Sea (Mellon-Duval et al., 2017) and non-Mediterranean areas (Velasco and Olaso, 1998), with mixed conclusions. A study from the north-eastern Mediterranean Sea showed an increase in fish ingestion in winter and dominance of crustaceans during summer (Stagioni et al., 2011). This is partly in line with our results, as despite high proportions of cluster 2, 3, and 4 (formed exclusively of crustaceans) were observed during both seasons, consumption of cluster 5 (formed of fish and crustaceans) was overall higher during winter. Some studies have also recorded an increase on juveniles feeding on euphausiids in spring, coinciding with the aggregating reproductive behavior of this species (Ferraton et al., 2007), as well as an increase of gobiids consumption during autumn. Furthermore, a study from the Gulf of Lyon (northwestern Mediterranean Sea) found lower proportion of pelagic fish ingestion in adults’ diet in spring when compared to autumn (Mellon-Duval et al., 2017).

Despite valuable information, and the novelty and benefits of using a continuous covariate in the MixSIAR analyses, which is a relatively novel technique (Francis et al., 2011; Stock and Semmens, 2016; Gagne et al., 2018; Stock et al., 2018; Gorman et al., 2019), this study is subjected to some limitations. Unfortunately, this study only covers one year and it is not possible to prove repeatability of the seasonal pattern observed in the analysis. Moreover, the use of the mean weight as a proxy of European hake size and life stages in the B-SDM is appropriate to identify areas mostly occupied by juveniles and large adults in winter and summer. This type of variable can be of interest when wanting to account for changes related to life stages but where data of individuals’ body length is scarce, not available or not feasible. For example, oceanographic surveys where total biomass and abundance per haul is usually recorded but the individual sampling of all specimens for biological analysis is unrealistic. However, this kind of proxy also presents some limitations, not directly informing of the less extreme phases of the life cycle, as intermediate values of mean weight cannot be categorized into adults other juveniles and are likely composed of mixed life stages. Furthermore, according to the length-weight relationship for European hake in the western Mediterranean Sea (Morey et al., 2003), for an individual of 25 cm, which is the length above which we classified individuals as “adults” (Bozzano et al., 1997; Lleonart, 2002), the corresponding mean weight is around 0.0895 kg. This means that for our data, only a reduced number of hauls had mean weight values over this threshold (Supplementary Table 6). This translates into an under-representation of adult individuals, especially during summer. This is partly related to the fact that all the data for this study was collected through bottom trawling, whereas generally larger individuals of European hake are caught trough long net gears. This means that adults were not very abundant and that our maximum individuals’ size is 50.2 cm, but European hake in these waters can reach body lengths of more than 80 cm. Therefore, the diet and spatial distribution described here is not representing the entire population, omitting the largest individuals.

Our findings could have clear implications for a future seasonal-based adaptive fisheries management, as local depletion of prey, or variation in prey size and/or condition may affect hake presence in an area. This is of particular interest for this study as overfishing has led to a general decrease on demersal species (FAO-MED, 2018), some of which are prey to European hake, as well as a decline in small pelagic fish (e.g., Sardina pilchardus and Engraulis encrasicolus) (Coll and Bellido, 2019). Indeed, previous studies have already expressed their concerns about the underlying effect of these alterations for hake’s population stock and distribution (Sion et al., 2019). Combining this information with future climate-change scenarios could result on major variations of the potential distribution of European hake in this area. Therefore, this study presents important information in the context of phenological processes of exploited species and their potential changes due to climate change. It does also emphasize, the need for additional monitoring efforts that consider a seasonal sampling of the marine ecosystems. Further information about spatial trophic analysis with season and ontogenetic stages is necessary if we want to fully understand species ecological roles, spatial-temporal population and food-web dynamics within marine ecosystems.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The authors declare that all fieldwork was approved by the Ethics Committee of CSIC, in accordance with the Spanish and EU legislation on the protection of animals used for scientific purposes.

EL-L, MC, JN, and MA-P: conceptualization and design of the experiment. MC, JN, NL and MA-P: samples collection. NL and MA-P: samples preparations. EL-L, MA-P, and JG: formal analysis. EL-L: writing–original draft preparation. MP, JN, MC, JG, MA-P, NL, and EL-L: writing–reviewing and editing. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.