Diet of Adult Sardine Sardina pilchardus in the Gulf of Trieste, Northern Adriatic Sea

: Food availability is thought to exert a bottom-up control on the population dynamics of small pelagic ﬁsh; therefore, studies on trophic ecology are essential to improve their management. Sardina pilchardus is one of the most important commercial species in the Adriatic Sea, yet there is little information on its diet in this area. Adult sardines were caught in the Gulf of Trieste (northern Adriatic) from spring 2006 to winter 2007. Experimental catches conducted over 24-h cycles in May, June and July showed that the sardines foraged mainly in the late afternoon. A total of 96 adult sardines were analysed: the number of prey varied from a minimum of 305 to a maximum of 3318 prey/stomach, with an overall mean of 1259 ± 884 prey/stomach. Prey items were identiﬁed to the lowest possible taxonomical level, counted and measured at the stereo-microscope. Overall, sardines fed on a wide range of planktonic organisms (87 prey items from 17 µ m to 18.4 mm were identiﬁed), with copepods being the most abundant prey (56%) and phytoplankton never exceeding 10% of the prey. Copepods of the Clauso-Paracalanidae group and of the genus Oncaea were by far the most important prey. The carbon content of prey items was indirectly estimated from prey dry mass or body volume. Almost all carbon uptake relied on a few groups of zooplankton. Ivlev’s selectivity index showed that sardines positively selected small preys (small copepods < 1 mm size), but also larger preys (such as teleost eggs, decapod larvae and chaetognaths), conﬁrming their adaptive feeding capacity.


Introduction
Small pelagic fish play a crucial role in marine food webs, as key species in energy transfer from plankton to larger predators, marine mammals and seabirds [1,2]. Short lifespans, high fecundity rates and planktivorous feeding behaviour make small pelagic fishes sensitive to changing oceanographic conditions [3]. However, the ability of anchovies and sardines to adapt to different trophic conditions and the plasticity of their feeding behaviour have often been linked to their "ecological success" [4][5][6].
Fish exploitation and environmental changes may exert an important combined effect on small pelagic fish stocks. In the Mediterranean Sea, several authors have supported the hypothesis that food availability and/or trophic competition may exert a bottom-up control on the growth, size and body condition of small pelagic fish [7,8]. Therefore, studies on the trophic ecology of small pelagic fishes represent fundamental knowledge to improve the assessment and management of these species.
Previous studies based on numerical dietary indices described sardines as exclusively or almost exclusively phytophagous fishes. This was especially true for sardines living in upwelling areas, where it is commonly assumed that high phytoplankton production supports short and efficient food chains [9][10][11]. However, this assumption has been challenged following the assessment of carbon uptake by different food items [4,12]. It has layer, being deployed from the bottom to a height of 4 m above the bottom. Setting and towing times were adapted to seasonal and environmental conditions to maximize the fishing success. Fish from each tow were sorted, immediately frozen on board at −20 °C to stop digestive processes and preserved at the same temperature until laboratory analysis.   17:09 trawling 6 (0) -Zooplankton samples were collected during fishing operations, generally after the retrieval of the fishing net. Vertical plankton tows from about 3 m from the bottom up to the surface were performed with a standard WP2 net (mesh size 200 μm; mouth opening diameter 58 cm). Immediately on retrieval of the net, plankton samples were fixed and preserved in a seawater-buffered formaldehyde solution (4% final concentration).
Temperature profiles were measured using a PNF-300 Profiling Natural    17:09 trawling 6 (0) -Zooplankton samples were collected during fishing operations, generally after the retrieval of the fishing net. Vertical plankton tows from about 3 m from the bottom up to the surface were performed with a standard WP2 net (mesh size 200 µm; mouth opening diameter 58 cm). Immediately on retrieval of the net, plankton samples were fixed and preserved in a seawater-buffered formaldehyde solution (4% final concentration).
Temperature profiles were measured using a PNF-300 Profiling Natural Fluorometer probe in May, June and July 2006 and a portable thermometer YSI85 probe in September, October, December 2006 and February 2007.

Diel Feeding Cycle
At laboratory fish were defrosted, measured to the nearest 1 mm of total length (TL) and weighed on an analytical balance to the nearest 0.0001 g of total wet body mass. A total of 343 sardines (Table 2) were dissected and their stomachs were removed and preserved individually in a buffered 4% formaldehyde-seawater solution. Afterwards the stomachs were dissected and their contents were washed with distilled water on a Petri dish under a stereo-microscope. Subsequently, the stomach contents were filtered through pre-dried and pre-weighed glass microfiber filters (Whatman ® grade GF/F, 25 mm Ø) and dried at 60 • C until constant mass. To determine the diel feeding periodicity of sardine, the Fullness index (F) was calculated as F = (DM/SWM) × 1000 g SWM where DM is the dry mass of the stomach content and SWM is the somatic wet mass of fish. Feeding periodicity was described by plotting mean Fullness index (F) calculated in fish from the same tow against the time of day.

Diet Composition and Dietary Carbon
Only sardine specimens caught during the period of maximum feeding activity were analysed to describe the diet composition since prey were less digested and easier to identify. Digestive tracts' dissection took place under a stereo-microscope and the stomach content of each fish was washed out onto a Petri dish. All the material contained both in the cardiac stomach and in the fundulus of the stomach was considered as "stomach content". Regurgitation during sampling was not observed since no food was found in any oesophagus.
For each sampling period (Table 2), sardines were randomly divided in 3 groups of 8 individuals. The stomach contents of the 8 sardines were pooled and diluted in a known volume of 0.22 µm filtered seawater. After homogenization, subsamples representing one fish (1/8 of the original pool), were analysed under the stereo-microscope (Leica M205-C, up to 160× magnification). This procedure was repeated for 3 pools of 8 sardines each, in order to produce replicates. Prey items were identified to the lowest possible taxonomical level and counted. When specific characters were missing or damaged, copepod specimens of the genera Paracalanus, Ctenocalanus, Clausocalanus and Pseudocalanus were classified as the "Clauso-Paracalanidae" group. The prosoma length of all copepods and the maximum dimension of each other prey were measured using an ocular micrometer (accuracy of 6 µm). The original size of incomplete prey was reconstructed by means of morphometric relationship, obtained by the measurements of whole individuals captured in zooplankton samples. Prey size distribution was represented by grouping sizes in classes of 50 µm.
The carbon content of prey items was indirectly estimated applying relationship between dry mass or body volume and carbon content (Table A1).

Feeding Selectivity
Feeding selectivity was estimated by Ivlev's electivity index E [34], calculated as follows where r i is the relative abundance of prey category i in the stomachs of fish (as a percentage of the total stomach contents) and a i is the relative abundance of the same prey at sea. E ranges from −1 to +1; negative and positive values indicating avoidance or positive selection for a prey category, respectively, and zero value indicating neutral selectivity.
Mesozooplankton samples collected in concomitance with fishing operations, were considered to define the food availability at sea. Taxonomic composition was analysed in subsamples sufficient to count and identify at least 1000 specimens. Mesozooplankton abundance was expressed as number of individuals per cubic meter of seawater. The volume of filtered water for each sample was estimated by multiplying the net-mouth area by the sampling depth.

Feeding Selectivity
Feeding selectivity was estimated by Ivlev's electivity index E [34], calculated as follows E = (ri − ai)/(ri + ai) (1) where ri is the relative abundance of prey category i in the stomachs of fish (as a percentage of the total stomach contents) and ai is the relative abundance of the same prey at sea. E ranges from −1 to +1; negative and positive values indicating avoidance or positive selection for a prey category, respectively, and zero value indicating neutral selectivity. Mesozooplankton samples collected in concomitance with fishing operations, were considered to define the food availability at sea. Taxonomic composition was analysed in subsamples sufficient to count and identify at least 1000 specimens. Mesozooplankton abundance was expressed as number of individuals per cubic meter of seawater. The volume of filtered water for each sample was estimated by multiplying the net-mouth area by the sampling depth.   Mesozooplankton composition and abundance were analysed in samples collected in correspondence to the fishing hauls dedicated to fish diet analysis. Mesozooplankton abundance ranged from 2044 to 13,212 ind. m −3 , measured in December and July, respectively ( Figure 3). Copepods were generally the most abundant group with the exception of July and September when Cladocerans numerically dominated (6961 and 6352 ind. m −3 , respectively). Copepods were mainly represented by the order Calanoida from February to September (72-91% of Copepods) and by the Cyclopoida of the suborder Ergasilida in October and December (54% and 42%, respectively). Cnidarians were particularly abundant in June (2404 ind. m −3 ), mainly with Muggiaea spp. and other Siphonophorans. Meroplankton, composed essentially of Echinoderms and Molluscs, showed its maximum (up to 2525 ind. m −3 ) in June and July ( Figure 3). Chaetognaths' abundance ranged from the minimum of 4 to the maximum of 78 individuals m −3 in May and December, respectively. Detailed information is available in the Table A2.

Diel Feeding Cycle
The diel pattern of stomach fullness confirmed the diurnal feeding behaviour of S. pilchardus. The fullness index (F) was generally low in the morning, gradually increased at noon and peaked at dusk. Overnight, feeding activity had nearly ceased by sunrise. In all months studied, the highest values of stomach fullness were observed between 18:00 and 21:00 and the lowest values around 4:00. The mean Fullness index varied from a minimum of 0.04 in June and July to a maximum of 1.9 in May ( Figure 4). In May, the maximum fullness value was twice as high as the maximum values observed in June and July. In October, it was not possible to describe feeding periodicity because catches did not yield sufficient numbers throughout the day. No sardines were caught in December and February.

Composition of the Diet
A total of 96 sardines were analysed: 24 specimens collected in the same tow in May, June, July and October ( Table 2). The total length (TL) of adult S. pilchardus ranged from a minimum of 138 mm to a maximum of 200 mm, with a mean of 169.5 ± 12.4 mm ( Table 2).

Diel Feeding Cycle
The diel pattern of stomach fullness confirmed the diurnal feeding behaviour of S. pilchardus. The fullness index (F) was generally low in the morning, gradually increased at noon and peaked at dusk. Overnight, feeding activity had nearly ceased by sunrise. In all months studied, the highest values of stomach fullness were observed between 18:00 and 21:00 and the lowest values around 4:00. The mean Fullness index varied from a minimum of 0.04 in June and July to a maximum of 1.9 in May ( Figure 4). In May, the maximum fullness value was twice as high as the maximum values observed in June and July. In October, it was not possible to describe feeding periodicity because catches did not yield sufficient numbers throughout the day. No sardines were caught in December and February. Copepods were the most abundant food category in the stomachs of captured sardines in all months, with the sole exception of June when tintinnids predominated numerically. Among copepods, the Clauso-Paracalanidae group and the genus Oncaea were by far the most important, alternately dominating. Copepod nauplii were always found, with

Composition of the Diet
A total of 96 sardines were analysed: 24 specimens collected in the same tow in May, June, July and October ( Table 2). The total length (TL) of adult S. pilchardus ranged from a minimum of 138 mm to a maximum of 200 mm, with a mean of 169.5 ± 12.4 mm ( Table 2).
A total of 15,109 prey items belonging to 73 taxa were identified ( Table 3). The number of prey varied from a minimum of 305 to a maximum of 3318 prey/stomach, with an overall mean of 1259 ± 884 prey/stomach. Large differences were observed between months ( Table 4). The numerically most abundant prey categories (N%) were copepods (85.0%) and Dinophyceae (9.5%) in May; tintinnids (56.4%), eggs and larvae of teleosts (12.2%), Dinophyceae (11.3%) and copepods (10.8%) in June; copepods (48.4%) and Crustacea larvae (21.1%) in July; copepods (50.4%) and chaetognaths (44.3%) in October.     Copepods were the most abundant food category in the stomachs of captured sardines in all months, with the sole exception of June when tintinnids predominated numerically. Among copepods, the Clauso-Paracalanidae group and the genus Oncaea were by far the most important, alternately dominating. Copepod nauplii were always found, with relatively small amounts, from 0.38 to 7.35% of copepods. A variable amount of invertebrate eggs with diameters ranging from 33 to 88 µm, probably belonging to copepods, was observed in the gut contents. Nevertheless, we did not consider these eggs as "prey" because it was impossible to determine whether they were ingested intentionally or as egg masses carried by the captured copepods. Chaetognaths represented an important food category in October in terms of numbers (44.3%), immediately after copepods.
We found microphytoplankton in all seasons with 26 taxa representing numerically about 10% of the prey in May, June and July and only 1% in October. In contrast, tintinnids were present with four taxa and were especially present in June when they dominated the number of prey (55.13%), probably related to a bloom of Tintinnopsis radix. In May, we observed a very high number of pollen grains (4616 ± 637.4 cells/stomach), but they were not included in the diet analysis. In spring, pollen may be very abundant in coastal surface waters, but we have not yet found evidence of the ability of sardines to digest it.

Food Carbon
An estimation of prey carbon content showed that S. pilchardus obtained almost all carbon from metazoans, especially copepods, chaetognaths, crustacean larvae, and the eggs and larvae of teleosts (Table 4). This was true even when considerable amounts of microphytoplankton and microzooplankton were present in the stomach contents. For example, in June, tintinnids were the most abundant prey in the stomach contents (56.4%); however, they accounted for only 0.3% of the total carbon ingested. Zooplanktonic prey of large size and high carbon content were the most important energetic food source, even when only a few specimens were ingested, such as chaetognaths in July (4.9% by number, 47.3% by carbon content), teleost eggs and larvae in June (12.2% by number, 62.9% by carbon content) or Crustacea larvae in June (5.5% by number, 24.0% by carbon content).
Differences among months were even more pronounced when dietary carbon content was considered. In fact, only one prey category accounted for most of the total carbon intake in each month: 97% of chaetognaths in October, 96% of copepods in May, 63% of teleost eggs and larvae in June, and again 47% of chaetognaths in July.
The overall size spectrum of prey ranged from 17 µm for the diatom Thalassiosira spp. to 18,388 µm for chaetognaths ( Figure 5), but in all seasons most prey had sizes from 400 to 1000 µm. Isolated peaks occurred only in May in the 50-100 µm size range (mainly corresponding to Dinophyceae) and in June in the 200-250 µm size range (mainly corresponding to Tintinnids) ( Figure 5). Overall, most of the prey carbon was in size classes from 600 to 1000 µm, especially in May and June, with an isolated peak in June and July in the 1700-1750 µm size class (corresponding to Crustacea larvae). cnidarians and echinoderm larvae. Partially avoided prey (mesozooplanktonic specimens of taxa more abundant at sea than in the gut contents, in terms of relative abundance) were: "Others", Appendicularia, mollusk larvae and polychaetes larvae.

Selection of the Prey
The Ivlev index was calculated for 16 prey groups, excluding those prey (Bacyllariophyceae, Dinophyceae, Tintinnina, eggs of Invertebrata) that were too small to be effectively retained by the mesh of the WP2 plankton net. All taxa found at abundances <1% in both food and environment (e.g., ctenophores, nemerteans, phoronids, ostracods, stomatopods, amphipods, isopods, mysidaceans, thaliaceans, cephalochordates) were grouped into a category labelled "Others".
The Ivlev index values ( Figure 6) confirmed the preference of adult sardines for large prey such as chaetognaths, decapod larvae, eggs and larvae of teleosts. Nevertheless, some smaller prey such as harpacticoid and cyclopoid (Ergasilida) copepods (with sizes of 150-880 µm and 125-625 µm, respectively) and cirripede larvae (220-660 µm) were also positively selected. Other prey items were selected only occasionally (copepod nauplii, cyclopoid copepods and Cladocera).   Completely avoided potential prey (mesozooplanktonic specimens of taxa found in the sea but never observed in gut contents) that had negative Ivlev index values were cnidarians and echinoderm larvae. Partially avoided prey (mesozooplanktonic specimens of taxa more abundant at sea than in the gut contents, in terms of relative abundance) were: "Others", Appendicularia, mollusk larvae and polychaetes larvae.

Trophic Environment
In the Adriatic Sea, Sardina pilchardus spawns from October to May at water temperatures between 9 and 20 • C, with peaks between 11 • C and 16 • C, and salinity between 35.2 and 38.8 [35]. Gamulin and Hure [36] noted that sardines disappear from fishing grounds from October to March, which corresponds to their spawning season. Tičina et al. [37] suggested that sardines leave the most productive shallow waters of the north in the fall for the deeper waters of the south, where they find the stable and relatively warm waters necessary for spawning. This is probably the reason why we did not catch any fish in winter. Vučetić [23] also reported that she had difficulty obtaining adequate samples of sardines in winter.
According to other authors, sardine migration is favoured by the search for the optimal trophic conditions, with zooplankton concentration being higher in May-June in the shallow coastal areas, especially meroplankton [24,38], while in winter it becomes more available in the upper layers of the offshore areas, especially for large copepods [39][40][41][42]. In any case, migration to open and deeper waters is assumed to be associated with spawning and overwintering [43]. In the study area, copepods were the most abundant mesozooplanktonic organisms, with the only exception being in summer when cladocerans (with the summer swarming species Penilia avirostris) predominated in numbers. Meroplankton, consisting mainly of echinoderms, mollusk and crustacean larvae, were abundant in the spring. This is the typical seasonal outline of the neritic and estuarine plankton community in the northern Adriatic [44].

Diel Feeding Cycle
In May, June and July, sardines exhibited high stomach fullness during the dusk and early night hours. Evidence of daytime feeding by S. pilchardus has been previously observed in the Adriatic Sea, both in adults [23] and late larvae [45], as well as in the northern Aegean Sea [16] and in the Catalan Sea [19]. The late afternoon feeding peak coincides with the migration of zooplankton to the surface. Zooplankton species are nearly transparent, so only some specific pigmented body parts can be seen by a visual predator [46]. Such prey are difficult to detect in light with a natural angular distribution (low image contrast). Nevertheless, planktivorous fish may increase the contrast of their prey by searching for it at an angle greater than 48.6 • to the vertical, as this makes the prey appear bright against a dark background [47]. This finding could explain the feeding peak that occurs when the sun is generally low on the horizon and the angle to the vertical is greater.
When analysing the stomach contents of fish caught by commercial purse seines operated at night under artificial light, a fundamental problem arises because artificial lights attract different species of mesozooplankton in different ways, creating unnatural conditions. This leads to a bias as the natural diet of small pelagic fishes is described under artificial conditions: both qualitative and quantitative aspects are distorted.

Diet Composition and Seasonality
The decision to analyse only samples caught in the late afternoon, during intense feeding activity, was justified by the possibility to detect prey not yet digested and to describe the composition of the bulk of the diet. We identified an average number of 466 to 2446 prey items per stomach. This abundance is comparable to the results obtained by Nikolioudakis et al. [17] in the northern Aegean (from 83 to 3334) and by Costalago et al. [15] in the western Mediterranean (from 1498 in winter to 4843 in summer). In contrast, the results are quite different when compared to those of Costalago et al. [15] in the Iberian Atlantic (from 40,126 in winter to 1,231,010 in summer). However, when phytoplankton is excluded, the range found by Costalago et al. [15] in the Atlantic (1339 prey/stomach in winter and 4094 prey/stomach in summer) is more similar to that found in this work in the northern Adriatic (from 460 to 2208 prey/stomach).
The results of the present study show that the diet of adult S. pilchardus in the northern Adriatic consists mainly of zooplankton. The diet was numerically based on copepods in all months, with the only exception being in June when the diet was more differentiated. Among the copepods, the Clauso-Paracalanidae group was the most abundant, followed, in decreasing order, by the genus Centropages, unidentified Calanoida, the families Oncaeidae and Corycaeidae, the species Euterpina acutifrons and the genera Temora and Oithona. These copepod groups have also been reported by other authors as important prey on the Spanish Atlantic coast [48], on the Portuguese coast [49], in the northwestern Mediterranean [6, 15,19], in the northern Aegean [17], in the eastern Aegean [21] and in the central Adriatic [23,26].
Other prey categories reached high abundances only in a single month: tintinnids (in June), decapod larvae (in July) and chaetognaths (in October). Tintinnids have been confirmed as food for sardine in the Mediterranean [50] and on the Atlantic coast of Spain [49,51,52]. Crustacean larvae have also been found by other authors [15,17,19,21,23,26,49]. In contrast, chaetognaths have only been observed by Vučetić [23] in the central Adriatic. Teleostea eggs were also found in sardine stomachs in previous studies [49,51,53], but only in small quantities in the Mediterranean [15,26].
Dinoflagellates (Dinophyceae) were consistently present in the diet from May to July, while diatoms (Bacillariophyceae) were found in surprisingly low amounts. We observed a mean of 113.6 phytoplanktonic cells/stomach, a value lower than those found by Costalago et al. [15] in the northwestern Mediterranean (from 364.8 to 1192.8) and especially on the Iberian Atlantic coast (from 38,787.1 to 1,226,915.8).

Dietary Carbon
Early studies on the diet of Sardinops sagax from the Benguela region, based on abundance data or the volumetric method, indicated that sardines were phytoplanktivorous, non-selective filter feeders [54]. Later studies have shown that zooplankton make up the largest proportion of the diet, although phytoplankton play an important role in certain regions (upwelling systems) or in particular periods of the year [55,56]. The contribution of phytoplankton to the carbon content of the diet of adult sardines varies widely, ranging from 14-19% on Portuguese coasts [49] to <3% in the northern Aegean [17]. In the northern Adriatic, we found that the carbon uptake of sardine completely relies on zooplankton, while the contribution of phytoplankton to the total carbon content of the diet was 0.02%. The main food categories in terms of carbon content were: chaetognaths (49.9%), copepods (30.8%), teleost eggs and larvae (10.3%) and decapod larvae (8.0%). It should be noted that in the present study, the contribution of copepods to the carbon content of the diet was underestimated as their egg masses were not taken into account.
The seasonality of the diet, expressed as carbon content, showed even more marked differences than the number of prey. Copepods, chaetognaths, eggs and larvae of teleosts and decapod larvae accounted for almost all the carbon content of the diet in all seasons considered. Similarly, Costalago and Palomera [19] found that, regardless of their numerical importance, decapod larvae and copepods contributed more than any other prey to seasonal differences in diet.

Feeding Selectivity
Although filter feeding is considered the main feeding mode in S. pilchardus [48,57], the ability to switch to a particulate feeding mode was described by Garrido et al. [49] for sardines under experimental conditions: filtering was adopted when small particles (≤724 µm) were offered as food, while particulate feeding started when bigger prey (≥780 µm) were proposed. When a wide range of prey sizes was offered, both feeding modes were simultaneously adopted. In the studied area, prey showed a wide range of sizes (from 17 µm to 18,388 µm) without a clear distribution (Figure 6), suggesting that adult sardines in the northern Adriatic may use both filter and particulate feeding regardless of the season.
The Ivlev index pointed out that sardines actively selected teleost eggs and larvae, decapod larvae and chaetognaths, prey that were less abundant in the plankton samples than in stomach contents. Nevertheless, not only large prey were positively selected, but also small copepods such as Oncaea spp. and Euterpina acutifrons (whose maximum size is <750 µm) were preferred. The high selectivity for these small copepods reported in this and other works [17,19] and even for anchovies [58,59], could probably be explained by their tendency to associate with detritus and/or gelatinous zooplankton (e.g., [60,61]), which induce them to aggregate into patches, making them easy prey to be pursued and caught by filtering sardines.
Further evidence of selectivity in food intake is the fact that cladocerans were only marginally present in the stomach contents, even though they were dominant in the summer plankton. This result contrasts with Costalago and Palomera [19], who found that cladocerans are highly selected by sardines in the northwestern Mediterranean during summer. The importance of cladocerans in the diet of small pelagic fishes is not clear: some authors emphasise their importance [18,62,63], but others found that high concentrations of cladocerans could even have unfavourable effects on the feeding activity of small planktivores [64].
Some of the most abundant meroplankton (echinoderm larvae) were never found in the stomachs and thus were completely avoided by the sardines. The inconsistency of prey composition in stomach contents in relation to plankton has also been noted by other authors [17,19,23]. In laboratory studies, sardines, when fed with wild mixed prey assemblages, also showed a preference for copepods and decapods over other zooplankton [11]. The constant presence of copepods in gut contents suggests that certain prey characteristics may be more likely to induce predation by fish. In planktivores, prey detection may be strongly influenced by prey movement [65,66], shape and colour of prey body [67], relative orientation between prey and predator [68] and light intensity [69].
The observed low selectivity for calanoid copepods could be due to the ability of these prey to escape sardine predation [17,70]. The swimming behaviour of copepods generally varies between continuous and intermittent locomotion, but can also have even more complex features, as in the genus Clausocalanus, involving a rapid and continuous movement in intertwined small loops [71]. The short pauses of motion may provide copepods with brief moments of invisibility to predators, and the subsequent sinking may increase their perceptual ability [72,73]. In addition, copepod species appear to modify their escape behaviour depending on the strength of the stimulus they encounter [74].
Explaining food selection in small planktivorous fishes is quite difficult as little is known so far about the vertical distribution of planktonic species, their ability to swarm and their escape capacity. This, in turn, complicates considerations about the probability of an encounter between prey and predator. The predator is often able to compensate for the prey's adaptations, resulting in equal capture success [74]. In any case, for sardines, the factor determining the selection/avoidance of a potential prey, as well as the switch from filter feeding to the particulate feeding mode, is the final energy uptake achieved when the gains from consumption exceed the losses from capture. On the other hand, sardines can influence planktonic community structures and food web functioning thanks to their ability to select for food.
Studying the adaptive capacity of small pelagic fishes is central to better management of fisheries' resources. Given the importance of mesozooplankton in understanding the ecology of small pelagic fishes, scientific surveys should be carried out with regular environmental monitoring, including plankton sampling. Data Availability Statement: All data generated during this study are included in this article with the exception of temperature and salinity data which are available on request from the corresponding author.
Acknowledgments: Special thanks to the crew of Cuba and Acquario fishing vessels for their assistance in the field work and to Alenka Goruppi for the revision of the taxonomical list of zooplankton and the help for the editing of figures. We thank the three anonymous reviewers of this manuscript for their useful suggestions.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.