Feeding Habits and Prey Composition of Six Mesopelagic Fish Species from an Isolated Central Mediterranean Basin

Abstract

Mesopelagic fishes hold an important position in marine food webs, serving as a link between lower trophic levels and top predators and transferring energy from their deep mesopelagic habitat to shallower oceanic layers. Despite their ecological importance, research on mesopelagic fishes’ diet and feeding habits in the Mediterranean Sea is far from thorough. The present work attempts to assess the preying patterns and diet composition of four myctophid (Benthosema glaciale, Ceratoscopelus maderensis, Myctophum punctatum, Notoscopelus elongatus) and two sternoptychid (Argyropelecus hemigymnus, Maurolicus muelleri) species from the Corinthian Gulf (Ionian Sea, Greece), sampled during pelagic trawl surveys in 2018 and 2019. Stomach vacuity was high for myctophids caught during daytime, a pattern which sternoptychids did not follow. Estimated trophic indices revealed high dietary diversity (Shannon’s H’ index) for most investigated species, but a narrow trophic niche breadth (Levins’ normalized B_n index). Copepods and various marine crustaceans were dominant in all diets, classifying them under the zooplanktivorous trophic guild, while A. hemigymnus exhibited high concentrations of particulate organic matter in their stomachs and N. elongatus exhibited a higher consumption of fish. Diet overlap was significant among most studied mesopelagic species, as indicated by Shoener’s S index and confirmed by both a multidimensional scaling ordination and a hierarchical cluster analysis. Information on mesopelagic fishes’ diet composition in this poorly studied part of the Mediterranean is useful in further assessing and parameterizing marine food webs and midwater trophic interactions, as well as in quantifying the ensued energy transfer to top predators of commercial interest or conservation concerns.

1. Introduction

Mesopelagic fish have been characterized as the most abundant vertebrate group of marine organisms [1], with their current global abundance estimated to be approximately 10–30 times higher than what was previously thought (i.e., 1000 million tons) [2]. Fishes of the families Myctophidae and Sternoptychidae particularly, are considered to form some of the most substantial biomass assemblages of the mesopelagic ecosystem [3]. Due to their high abundance and worldwide distribution, mesopelagic fish are considered to exert high predation pressure upon their zooplankton prey [4,5]. These species also form sound scattering layers in the mesopelagic zone, more commonly referred to as deep scattering layers (DSLs) [6], which can be detected by echosounders.

Many species perform diel migrations towards the epipelagic layer during the night, in order to feed [7,8]. Mesopelagic fishes have been reported to play a crucial role in transferring nutrients and energy from the euphotic zone to the depths of their habitat and to effectively link secondary production to top predators. In turn, they are prey for larger pelagic fishes, cephalopods, seabirds and mammals [8], significantly contributing to a reverse upward energy flux from the mesopelagic to the epipelagic zone [9]. Myctophids and sternoptychids in particular have been found to contribute considerable quantities to the diets of commercial demersal and neritic species, such as the European hake (Merluccius merluccius) [10], the Atlantic mackerel (Scomber scombrus) [11] and the albacore tuna (Thunnus alalonga) [12]. This trophic interaction between mesopelagic fish species and their shelf slope dwelling predators seems to be predominantly taking place during their diurnal migrations to shallower depths, providing essential sustenance to these species of high market value during all of their life stages.

Knowledge of their diet and feeding patterns is also essential in order to quantify their contribution to biogeochemical cycles, including the process of carbon sequestration by the oceans, known as “the biological pump.” In addition, such information is necessary in order to better understand their trophic position within deep-sea food webs. Despite their important ecological position, data on their feeding ecology in the Mediterranean Sea, based on stomach content analyses, have been relatively scarce, while trophic studies employing stable isotope and fatty acid analyses, have in recent years, started to further elucidate mesopelagic fishes’ feeding ecology in the area (e.g., [8,13,14]).

These studies have revealed that mesopelagic fishes can be categorized into different trophic guilds according to their feeding dependence on suspended particulate matter, meso- and macrozooplanktonic crustaceans or even small fishes and cephalopods. Some species may exhibit less specialized diets, feeding upon a variety of different taxa at various percentages [4,8]. Unlike the diets of other fish families, which have been studied to a greater extent [15,16], the majority of mesopelagic species’ trophic ecology research in the Mediterranean has, until now, focused on certain areas of its western and central parts [17,18], while relevant studies have been conducted in neither the Ionian Sea (central Mediterranean) nor the eastern Mediterranean.

Towards this direction, we analyzed the stomach contents of six myctophid and sternoptychid species sampled in the Corinthian Gulf (Ionian Sea—central Mediterranean). We estimated the trophic niche breadth and diet overlap indices, as well as the fractional trophic level of each species, in order to evaluate its position in the food web of the local ecosystem. The collected information on the diet of these widely distributed mesopelagic fishes will provide the necessary input to future trophic models, for an evaluation of their ecological role at regional scale, but also for the wider Mediterranean Sea.

2. Materials and Methods

2.1. Study Area and Sample Collection

The Corinthian Gulf is a deep (>930 m) isolated basin, located at the east end of the Ionian Sea (central Mediterranean Sea, Greece). The Gulf’s mesopelagic fish fauna includes some 15 species belonging to the families Myctophidae, Paralepididae, Sternoptychidae and Stomiidae [19]. Fish samples were collected in the Corinthian Gulf during November 2018 and April 2019 onboard the R/V PHILIA. The main sampling gear was a pelagic trawl with a 12 m × 7 m mouth opening and a 16 mm mesh (stretched) on the cod end. In addition, two hauls were performed with a smaller Sardonet pelagic trawl (mouth opening: 3.7 m × 2.1 m, mesh: 5 mm stretched) and two more with a scaled-down version of a Methot frame trawl [20] (1.5 m × 1.5 m frame, mesh: 1 mm stretched), aiming to capture smaller individuals. Sampling was carried out during both day and night (from dusk till dawn; see Figure 1 for individual station sampling times) and sampling depths ranged between 10–725 m (bottom depths: 150–865 m) (Figure 1). More sampling details are provided in Kapelonis et al. [19].

Sampled fish were either identified onboard and immediately preserved in 10% neutral-buffered formalin or packed and frozen (at −20 °C) until their transfer to the laboratory, where detailed taxonomic identification took place. The total length (TL) of each individual fish was measured to the nearest millimeter, followed by the subsequent removal of its stomach. Stomachs were preserved in formalin, prior to their dissection and examination under a microscope.

Mesopelagic species considered in the current study belonged to the families Myctophidae and Sternoptychidae. Four out of the total six investigated species were myctophids, namely Benthosema glaciale, Ceratoscopelus maderensis, Myctophum punctatum and Notoscopelus elongatus and two were sternoptychids: Argyropelecus hemigymnus and Maurolicus muelleri.

2.2. Laboratory and Data Analysis

The preserved stomachs were carefully washed of any persisting formaldehyde, and left to dry on absorbent paper, until the excess moisture was evaporated. Subsequently, the total weight of each stomach was weighed to the nearest 0.001 g, using a high precision digital scale. If there were signs of regurgitation, the stomach was excluded from all further analyses. Stomachs were then dissected and their contents were examined using a WILD M420 photomacroscope. Stomach fullness was recorded according to Bernal et al. [4], using a scale from 0 to 4, where “0” represented an empty stomach, “1” a stomach with at least one prey item and up to one quarter of its volume in fullness, “2” a stomach from 25% in fullness to half-full, “3” up to three quarters of stomach fullness and “4” a completely full or further expanded stomach.

Planktonic prey items were identified to the lowest possible taxonomic level according to the Mediterranean plankton identification keys of Trégouboff and Rose [21]. Organic matter that could not be identified, due to advanced prey digestion or due to specimens having actively fed on suspended organic material, was listed as particulate organic matter (POM). Each prey item was counted and similar items were weighed together, in order to estimate each prey’s numerical abundance (%N) and weight (%W) as a percentage of total prey items found in all of the stomachs of a particular species. The frequency of occurrence (%F) of each prey item was also estimated against the total number of stomachs of a predator containing any type of prey, and the Index of Relative Importance (IRI) [22,23] was subsequently calculated. IRI is an index combining numerical abundance values of each prey type, its estimated mass and frequency of occurrence, rendering it more representative for assessing dietary importance [24] when compared with conventional metrics. It is expressed by the formula IRI = (%N + %W) · (%F), while, when expressed as a percentage, it is expressed as follows: %IRI = (IRI/∑IRI) · 100 [17]. IRI and %IRI values were calculated for the highest possible taxonomic rank of each prey group, having excluded unidentified organic matter.

The trophic level (TROPH) of each fish species was then estimated using TrophLab, a stand-alone application for the estimation of trophic levels [25], using the quantitative routine for diet composition. In all cases, the default TROPH input values of the routine were used. The estimation of fish trophic levels can be formally expressed by the following formula:

$${TROPH}_i\mathrm{}=\mathrm{}1+\mathrm{}{\sum }_{j=1}^G{DC}_{ij}\times {TROPH}_j,$$

(1)

where DC_ij is the weight fraction of prey item j of species i, TROPH_j is the trophic level of prey item j and G is the total number of prey types in the stomach of predator i [25]. Species omnivory was expressed by the omnivory index (OI), which is derived from the sum of variances of a predator’s prey items’ trophic levels, as follows:

$$OI=\mathrm{}{\sum }_{j=1}^G({TROPH}_j-{TROPH}_i)\times {DC}_{ij}$$

(2)

the square root of which equals the standard error of the species’ TROPH (${\mathrm{S}\mathrm{E}}_{\tau }=\sqrt{OI}$) [26]. The OI receives a zero value when a species feeds on items of a particular trophic level and its value increases alongside the variety of consumed prey’s TROPHs.

The vacuity index (%VI) was estimated as the percentage of empty stomachs for each predator species [22,23] and was determined separately for day and night, in order to discern any preference between daytime and nighttime feeding. Prey diversity for studied fishes was assessed using the Shannon diversity index (H’) [27,28]. Shannon’s H’ was estimated according to the following formula:

$$Hʹ=–{\sum }_{i=1}^k{p}_i\mathrm{}\mathrm{l}\mathrm{n}{p}_i,$$

(3)

where k is the number of prey categories found in a predator’s diet and p_i is the frequency of observation of prey category i in the stomachs of that predator. Trophic niche breadth was also estimated by employing Levins’ normalized index (B_n) [29], using the relative frequency of each prey type, consumed by the sampled population (p_i), which was then normalized by the number of prey categories (R). B_n is therefore calculated by the following formula:

$${\mathrm{B}}_{\mathrm{n}}=\frac{1}{R · \sum _i{p_i}^2},$$

(4)

with i referring to each prey type [30]. B_n values vary from 1/R, when species tend to specialize on one type of prey, to 1, in which case they equally consume all prey types.

Diet overlap between mesopelagic species was assessed with Schoener’s index S [16,31], according to the following equation:

$$S=1-\frac{1}{2}·({\sum }_{i=1}^n\left|p_{xi}-p_{yi}\right|)$$

(5)

with p_xi and p_yi being the weight fractions of prey type i in the stomachs of species x and y. Values of Schoener’s S for species pairs vary between 0, when species exhibit totally unrelated diet compositions and 1, when compared diets are identical. Significant overlap in the diet of a pair of species is indicated by a value of S > 0.6 [32,33,34].

The diet overlap between individual specimens was also assessed by performing non-metric multidimensional scaling (nMDS), based on Bray-Curtis similarity and using prey item abundance data in Primer 6 (version 6.1.13) and PERMANOVA+ (version 1.0.3) software [35]. Data homogeneity between species was tested using the PERMDISP function with 999 permutations and because dispersion effects were found to be significant (p < 0.001), data were fourth root transformed in order to create the specimens’ diet resemblance matrix [4]. Average within-species diet similarity was assessed using SIMPER analysis, which was also used to determine the main contributing prey taxa in diet dissimilarity for each pair of investigated species. The between-species diet similarity dendrogram plot was constructed by implementing hierarchical agglomerative clustering and calculating the species average prey abundance values.

Finally, the importance of the identified prey items in the diet of each predator was depicted in two-dimensional Costello graphs by plotting the gravimetric contribution (%W) of each prey type against its frequency of occurrence (%F) in the species’ diet [36]. Prey type significance, as well as diet indices and analyses, were determined and performed for broader taxonomic prey groups, which were common for most investigated species, down to the rank of the taxonomic order, where identification was possible.

3. Results

Out of the total 437 stomachs examined, 234 contained at least one type of prey. The number of dissected stomachs ranged from 14 in the case of M. punctatum to 210 for C. maderensis. Mean lengths, length ranges as well as the estimated trophic indices of sampled mesopelagic fishes are given in

Table 1. Trophic indices of the mesopelagic fishes, sampled in the Corinthian Gulf (N: number of specimens; TL: total length (mm); TROPH ± SE: fractional trophic level and standard error; OΙ: omnivory index; %VI: vacuity index; H’: Shannon’s diversity index, B_n: trophic niche breadth Levins’ normalized index).

Species	Argyropelecus hemigymnus	Benthosema glaciale	Ceratoscopelus maderensis	Maurolicus muelleri	Myctophum punctatum	Notoscopelus elongatus
N	79	16	210	92	14	26
TL range	22–44	30–65	32–79	24–52	56–74	68–122
Mean TL (±SD)	32.0 ± 4.6	44.6 ± 9.8	57.0 ± 7.5	36.6 ± 8.6	62.1 ± 6.4	80.6 ± 13.8
%VI	41.77%	56.25%	55.71%	17.39%	66.27%	73.08%
Prey item N	88	15	1591	661	14	13
TROPH ± SE	2.6 ± 0.46	2.73 ± 0.44	3.14 ± 0.43	3.01 ± 0.29	3.03 ± 0.51	3.98 ± 0.7
OI	0.212	0.194	0.185	0.084	0.260	0.490
H’	1.304	0.683	1.830	1.572	1.33	1.213
Bn	0.282	0.196	0.545	0.386	0.36	0.291

, whereas detailed prey composition data for each species are presented in

Table 2. Diet composition of the six mesopelagic fish species sampled in the Corinthian Gulf (%N: numerical abundance; %F: frequency of occurrence; %W: weight contribution; %IRI: index of relative importance).

	Argyropelecus hemigymnus				Benthosema glaciale				Ceratoscopelus maderensis				Maurolicus muelleri				Myctophum punctatum				Notoscopelus elongatus
Prey Items	%N	%F	%W	%IRI	%N	%F	%W	%IRI	%N	%F	%W	%IRI	%N	%F	%W	%IRI	%N	%F	%W	%IRI	%N	%F	%W	%IRI
Crustacea (Total)	90.91	40.00	17.12		53.33	57.14	37.50		99.56	73.60	85.10		99.09	79.00	78.88		50.00	50.00	33.33		23.08	25.00	15.38
Unidentified Crustacea	10.23	10.00	6.05	9.31	53.33	57.14	37.50	100.00	0.88	12.80	9.13	2.86	1.66	7.00	2.81	0.77	28.57	33.33	27.78	66.05	7.69	12.50	15.38	4.11
Copepoda (Total)	75.00	21.67	8.77						96.98	49.60	56.97		96.52	66.00	70.05		21.43	16.67	5.56		15.38	12.50
Unidentified Copepoda	73.86	18.33	8.35	86.15					19.74	21.60	12.86	15.73	53.71	33.00	42.38	78.50	21.43	16.67	5.56	15.81	15.38	12.50		2.74
Calanoida (Total)	1.14	3.33	0.42						77.25	28.00	44.11		42.81	33.00	27.67
Unidentified Calanoida		1.67	0.21	0.02					77.25	28.00	44.11	75.92	22.39	16.00	16.84	15.54
Acartiidae													0.76	2.00	1.60	0.12
Calanidae (Total)													8.47	4.00	1.74
Unidentified Calanidae													3.48	1.00	1.07	0.11
Calanus sp. (Total)													4.99	3.00	66.84
Unidentified Calanus sp.													0.91	1.00		0.02
Calanus helgolandicus													4.08	2.00	0.67	0.24
Euchaetidae (Total)													0.61	3.00	2.67
Unidentified Euchaetidae													0.45	2.00	2.14	0.13
Euchaeta marina													0.15	1.00	53.48	1.33
Clausocalanidae													0.30	1.00	1.07
Ctenocalanus vanus													0.30	1.00	1.07	0.03
Paracalanidae	1.14	1.67	0.21										10.29	7.00	3.74
Paracalanus parvus	1.14	1.67	0.21	0.13									10.29	7.00	3.74	2.43
Malacostraca	5.68	8.33	2.30						1.70	11.20	18.99		0.91	6.00	6.02
Amphipoda									0.19	1.60	1.56	0.06
Decapoda (Total)									0.06	0.80	1.44		0.15	1.00	0.53
Decapoda (larva)									0.06	0.80	1.44	0.03
Lucifer typus													0.15	1.00	0.53	0.02
Euphausiacea	5.68	8.33	2.30	3.80									0.30	4.00	5.48	0.57
Mysida									1.45	8.80	15.99	3.43	0.45	1.00		0.01
Fish	2.27	3.33	0.84	0.59					0.25	10.40	8.17	1.96	0.15	1.00	4.68	0.12	14.29	16.67	16.67	18.14	61.54	50.00	69.23	93.15
Polychaeta									0.00	0.80	0.24	0.00	0.76	2.00	0.40	0.06
Particulate organic matter	5.68	55.00	79.54		46.67	42.86	62.50		0.19	15.20	6.49			18.00	16.04		35.71	33.33	50.00		15.38	25.00	15.38

.

Stomach emptiness of caught specimens was high for all species at the time of sampling, except for M. muelleri, which had an estimated vacuity index (%VI) of 17.39%. Other species exhibited %VI values ranging from 41.77% for A. hemigymnus to 73.08% for N. elongatus. The majority of examined stomachs came from specimens that had been sampled during daytime. Only C. maderensis comprised a high number of specimens fished during the night (

Table 3. Vacuity index (VI%) of examined stomachs per species for specimens caught during the day (D) and night (N) in the Corinthian Gulf. The numbers of examined individuals are given in parentheses.

VI%	D	N
Argyropelecus hemigymnus	41.03% (78)	100% (1)
Benthosema glaciale	100% (4)	41.67% (12)
Ceratoscopelus maderensis	76.83% (82)	42.19% (128)
Maurolicus muelleri	17.39% (92)	- (0)
Myctophum punctatum	75.00% (12)	0.00% (2)
Notoscopelus elongatus	76.00% (25)	0.00% (1)

). In the case of C. maderensis stomach vacuity was lower during nighttime, calculated to 42.19% for 128 specimens, compared to 76.83% for 82 specimens during the day. For the rest of the mesopelagic fishes, sample sizes did not allow for safe comparisons between daytime and nighttime sampling.

Overall, a total of 2382 individual prey items were recorded and classified in all examined stomachs. In the majority of species, various types of crustaceans were the most important prey type identified, as indicated by their relative importance index. Total crustacean relative importance in stomachs ranged from 6.85% in the diet of N. elongatus to 100% for B. glaciale, not considering POM or other unidentified digested food items. In A. hemigymnus, C. maderensis and M. muelleri, copepods showed the highest rates, with %IRI values of 86.30%, 91.66% and 98.45%, respectively (Table 2). Regarding copepods, those identified in higher taxonomic resolution were members of the order Calanoida. More specifically, identified calanoid copepods contributed 77.25% to the numerical abundance in the diet of C. maderensis and 42.81% in M. muelleri (Table 2). Prey contents of M. muelleri specimens’ stomachs had not undergone extensive digestion by the time of sampling (all sampled during daytime), which allowed for higher level of taxonomic identification of prey taxa. Identified calanoid families in M. muelleri’s diet were Acartiidae, Calanidae, Clausocalanidae, Euchaetidae and Paracalanidae. At a finer level, calanoid prey determined at the species level in M. mulleri’s stomach contents included Calanus helgolandicus, Ctenocalanus vanus, Euchaeta marina and Paracalanus parvus, while Lucifer typus was identified in the order Decapoda.

The presence of the class Malacostraca was generally less important in the diet of most examined specimens. Prey items belonging to this class were classified to the orders Amphipoda, Decapoda, Euphausiacea and Mysida. Overall, the dominance of crustacean prey items was evident in most of the examined species’ diets, with the majority of studied fishes exhibiting high consumption of calanoid copepods or other unidentified Crustaceans, as outlined by the Costello graphs (Figure 2).

Other semi-digested fragments of marine organisms were found inside the examined stomachs and were assorted to classes Actinopterygii and Polychaeta. More specifically, these fragments included fish otoliths, eyes, fish fin rays, small scales and polychaetae parapodia, but were relatively rare in most of the examined stomachs with the exception of N. elongatus specimens, in which fish remains were the most important identified prey item. Various fish fragments were also found to contribute a relatively important proportion in the diet of M. punctatum (%IRI = 18.14).

Estimated trophic levels ranged from a minimum of 2.6 ± 0.46 (SE) for A. hemigymnus to a maximum of 3.98 ± 0.7 for N. elongatus (Table 1 and Table 2). The highest omnivory index score was also estimated for N. elongatus at 0.49, while the lowest was estimated for M. muelleri at 0.084. Diet diversity values estimated with the Shannon–Weaver index (H’) were remarkably high for most species, ranging from 0.683 to 1.830. The lowest diet diversity index value belonged to B. glaciale, while C. maderensis, M. muelleri and M. punctatum presented the highest values (H’ = 1.830, 1.572 and 1.330 respectively). Nevertheless, as is evident by the low niche breadth index (B_n) values (Table 1) and backed up by the high abundance and frequency of appearance of certain prey groups in stomachs (Table 2), the mesopelagic fishes under study each seemed to prefer specific prey types, with copepods and other crustaceans being a common component in all cases.

PERMDISP analysis showed significant dispersion of prey item abundance data (F: 27.01, df _{number of groups} = 5, df _{number of samples} = 126, p < 0.001). Average within-species similarity, determined with SIMPER analysis, was low for all species (

Table 4. Average within-species similarity of prey item abundance.

Species	Av. Similarity (%)
Argyropelecus hemigymnus	28.30
Benthosema glaciale	100
Ceratoscopelus maderensis	33.75
Maurolicus muelleri	35.33
Myctophum punctatum	33.33
Notoscopelus elongatus	33.55

), ranging from 28.30 in A. hemigymnus to 35.33 in M. muelleri, with the exception of B. glaciale, for which the level of taxonomic resolution of prey did not permit for the observation of variation between individual specimens. Between-species cluster analysis showed high similarities between M. punctatum and N. elongatus and between C. maderensis and M. muelleri (>75%, Figure 3) in terms of prey item relative abundance. The main contributors to most species pairs’ diet dissimilarity proportions were calanoids and unidentified copepods, while, in the case of N. elongatus, fish were the most prominent differentiating prey group, when compared with other species (

Table 5. Key prey contributions to mesopelagic fishes’ diet dissimilarity in terms of prey item abundance.

Species Groups	Key Prey Taxon Contribution (%)
A. hemigymnus—B. glaciale	Unidentified Crustacea: 41.04
A. hemigymnus—C. maderensis	Calanoida: 35.35
A. hemigymnus—M. muelleri	Unidentified Copepoda: 34.76
A. hemigymnus—M. punctatum	Unidentified Copepoda: 34.76
A. hemigymnus—N. elongatus	Fish: 36.78
B. glaciale—C. maderensis	Calanoida: 35.12
B. glaciale—M. muelleri	Unidentified Crustacea: 38.73
B. glaciale—M. punctatum	Unidentified Copepoda: 35.62
B. glaciale—N. elongatus	Fish: 45.29
C. maderensis—M. muelleri	Calanoida: 41.93
C. maderensis—M. punctatum	Calanoida: 34.00
C. maderensis—N. elongatus	Calanoida: 32.76
M. muelleri—M. punctatum	Unidentified Copepoda: 29.96
M. muelleri—N. elongatus	Fish: 30.61
M. punctatum—N. elongatus	Fish: 41.12

).

Diet similarity among individual specimens and prey overlap among species of both mesopelagic fish families varied widely, as indicated by the non-metric MDS analysis (Figure 4) and the diet overlap values of Shoener’s index (S > 0.6,

Table 6. Schoener’s index of diet overlap estimated using prey weight contribution (%W) to the diets of the examined mesopelagic fishes. Values in bold numbers signify a considerable diet overlap (S > 0.6) [32,33,34] between compared species as mentioned by Karachle and Stergiou [16].

Shoener’s S Index	Benthosema glaciale	Ceratoscopelus maderensis	Maurolicus muelleri	Myctophum punctatum	Notoscopelus elongatus
Argyropelecus hemigymnus	0.783	0.599	0.638	0.785	0.556
Benthosema glaciale		0.436	0.421	0.840	0.543
Ceratoscopelus maderensis			0.601	0.511	0.283
Maurolicus muelleri				0.461	0.232
Myctophum punctatum					0.647

). All investigated species exhibited significant diet overlap in prey weight contribution with at least another, with A. hemigymnus showing significant similarity in prey composition (POM being excluded) with most other species. The highest Schoener’s index value was estimated between M. punctatum and B. glaciale (S = 0.840), followed by that between M. punctatum and A. hemigymnus (S = 0.785).

4. Discussion

The present work examined aspects of preying patterns and diet composition of six mesopelagic fishes sampled in the Corinthian Gulf during surveys dedicated to the study of the mesopelagic zone in this deep enclosed basin. The sizes of individual specimens covered almost entirely the size ranges of the species recorded in the Corinthian Gulf [37] and are, thus, representative of the species’ populations in the study area. The investigated mesopelagic fishes exhibited highly diverse diets overall, but with definite species-specific feeding patterns. Various types of crustaceans were dominant in the diets of both sternoptychids and myctophids (with the exception of N. elongatus). Malacostracans, copepods or various unidentifiable, semi-digested crustacean segments were nonetheless omnipresent in the stomach contents of all mesopelagic specimens. Advanced digestion prevented most crustacean prey from being identified with high taxonomic precision, as prey items were mostly dissolute by the time of sampling. Mesopelagic fishes have been reported to feed mostly during the night [4], following their zooplanktonic prey to shallower depths [38,39], meaning that, with most analyzed specimens having been collected during the day, difficulty in high resolution prey identification came as no surprise. Most mesopelagic fishes showed high stomach vacuity during daytime, as expected for myctophid species, with sternoptychids presenting significantly lower %VI values. POM was also present in significant amounts in all of the examined stomachs, suggesting that, in the absence of prey of higher energetic value, mesopelagic fish may actively feed upon suspended organic matter in the water column.

In the Corinthian Gulf, A. hemigymnus is mainly detected in the deep scattering layers occurring at 160~250 m throughout the day [19], where plankton is more abundant in comparison with deeper layers, and therefore this species does not need to perform the extensive diel migrations of myctophids to feed [40]. Argyropelycus hemigymnus specifically has been reported to feed during both day and night in the Atlantic Ocean [4,41,42] and is considered a partial or non-migrator in the study area [19]. Argyropelecus hemigymnus’ diet also presented the highest proportion of weight in unidentified organic matter (%W = 79.54%), significantly affecting its trophic level estimation, which was the lowest among the species examined. Nevertheless, it exhibited a relatively high prey diversity (H’ = 1.304), with various identified and unidentified copepods dominating its stomach contents, similar to findings from the western Mediterranean [4].

A small microfiber was also found, having been consumed by an A. hemigymnus specimen. Although the presence of microplastics has in recent years been reported in high numbers in the gastrointestinal tracts of mesopelagic fishes in other oceanic areas e.g., [43,44,45], that was the sole case of microplastic ingestion in our samples. There is, however, a high probability of fibers from the working area contaminating the samples when the stomach contents are investigated, resulting in misleading conclusions [46].

Prey diversity and trophic niche breadth were found to be high for M. muelleri in the Corinthian Gulf. Our estimated fractional trophic level for the species was in line with findings from other studies, employing both stable isotope [14] and stomach content analysis [18]. Maurolicus muelleri inhabits the upper mesopelagic zone in the region (between 80 and 225 m) [19] and in the current stomach content analysis the species exhibited the lowest vacuity index (%VI = 17.39%). In addition, the high number of relatively freshly caught prey found in their guts indicated high feeding intensity during daytime. The freshness of M. muelleri specimens’ prey items allowed for a higher degree of precision in classification, identifying calanoid copepods down to the species level, such as Calanus helgolandicus and Euchaeta marina, as well as the decapod Lucifer typus. Euchaeta marina is referred to as a strictly surface copepod [17] and C. helgolandicus is characterized by a segregated bimodal vertical distribution with a surface population residing at depths of 0–200 m [47,48,49]. This suggests that M. muelleri in the Corinthian Gulf performs vertical migrations in order to feed, as has been described in previous studies targeting the species [50,51,52]; however, it is thought to mainly feed during the transitional periods between day and night, as light during the night is not sufficient for prey detection for this species [53]. Changes in this behavior have nevertheless been reported in northern latitudes based on factors such as ontogenetic stage, feeding and predation [54], as well as on light intensity [50,55].

The large number of C. maderensis specimens sampled during both day and night permitted for the identification of a relative variety of identified prey taxa, with their diet nonetheless being dominated by calanoid copepods. This coincides with stomach content findings from the western Mediterranean [4], where genera such as Pleuromamma, Paracalanus and Clausocalanus represented the greatest portion of prey composition, although, in the latter case, the species have also exhibited feeding upon larvaceans, something we did not observe in our specimens. In the Corinthian Gulf, C. maderensis demonstrated the highest diversity of consumed prey taxa among the investigated species (H’ = 1.830) and therefore the highest proportion of available food resources used (B_n = 0.545). The wide assortment of identified prey items found in the diet of C. maderensis and its clear consumption on calanoid copepods classify this species somewhere in between a generalist and a specialized feeder, exhibiting a mixed feeding strategy.

Fish consumption was noteworthy in some species that reach larger sizes (e.g., N. elongatus, for which the predominant prey was fish, but also for M. punctatum and to a lesser extent for C. maderensis), which subsequently allows for the consumption of larger prey. Bernal et al. [4] also identified fish remains in the stomach contents of most of the large-sized mesopelagic fish species that they studied from the western Mediterranean, which, as in our case, contributes to a relatively high fractional trophic level of their predators. Particularly for N. elongatus, for which fish were found to be the most important prey, its trophic level was estimated to be the highest among the investigated mesopelagic fishes from the Corinthian Gulf. Fish are generally euryphagic and induce high uncertainty in trophic estimations as prey, consequently increasing the index of omnivory of their predators. This is the reason both M. punctatum and N. elongatus were found to be the most omnivorous among our examined mesopelagic fishes, as these were the species with the highest fish consumption rates in their diets. However, records of fish as prey should be considered with caution, especially in instances where only scales have been found as proof of fish consumption, due to the possibility of unintentional cod-end feeding.

Estimated TROPH values in the current study were generally found to be in line with trophic level estimations for the investigated species based on stable isotope analysis of specimens collected from the western Mediterranean [8]. Only B. glaciale exhibited considerable trophic level divergence between the two areas and its lower trophic level in our study can be attributed to the high POM content in the specimens’ stomachs. Bernal et al. [4] identified calanoid and non-calanoid copepods, ostracods and euphausiaceans as the main prey of the species, while adult specimens showed high consumption on calanoids of the Pleuromamma genus, categorizing B. glaciale as a selective plankton feeder [56,57]. In our case, the state of our B. glaciale specimens’ stomach contents and the small number of non-empty stomachs did not allow for high resolution taxonomic categorization of their prey and induced too high an uncertainty in our findings to safely arrive at quantitative results. Although the small number of B. glaciale specimens were sampled exclusively during the night and in depths close to the surface, which is an indication of night-time feeding migration, the observed stomach contents included semi-digested crustacean segments and POM, which may also suggest an active dependency on suspended organic particles in the water column. B. glaciale has been previously described as an opportunistic feeder that may also feed during the day [58] at its daytime depth stratum, while its usual zooplanktonic prey remains mostly in the upper 100 m [59].

Significant diet overlap was observed in many cases, both between and within species in our investigated specimens. Broad diet overlap and low diet specialization have also been previously reported for mesopelagic fishes in the oligotrophic western Mediterranean [4]. The low feeding segregation in the region has been attributed to the relatively low species diversity of midwater fish fauna [40,56] and to the increased abundance of food resources, which together lead to a decrease in the competition for food. This was also highlighted in our study by the high proportion of diet dissimilarity estimated between specimens of the same species, indicating low intraspecific competition.

The diet composition of mesopelagic fishes has been found to be relatively consistent throughout different ontogenetic stages, with copepod prey at various developmental stages being generally common [60]. Overall, it has been shown that the abundance and size of consumed prey by fishes increase with their developmental stage [61], but this rule does not always seem to apply for myctophids or sternoptychids. Species that display a discrepancy to this trend, such as A. hemigymnus, are more short-deep body shaped and therefore exert less energy efficient movement compared with other more hydrodynamic shaped species. This factor would render those species less effective in going after more mobile prey and also explains the high POM weight contributions we observed in the stomach contents of the species.

With the exception of A. hemigymnus, the rest of the mesopelagic fishes have been found to perform extensive diel vertical migrations, with C. maderensis and B. glaciale displaying higher abundances in the upper layers of the study area during the night [19]. Most of the individuals included in this study were sampled during the day, when most of the species remain in their deeper daytime depths, where environmental parameters, such as dissolved oxygen and temperature, are relatively stable. These factors have been reported as limiting for the vertical distribution of mesopelagic communities and may differentiate migration patterns among regions with different oceanic regimes, consequently affecting prey availability and diet composition [62].

Mesopelagic fishes are generally characterized and grouped into three distinct trophic groups according to their feeding habits: zooplanktivorous, micronektivorous and generalists [63]. Our findings from stomach content analysis and the estimation of trophic indices pointtowards the zooplanktivorous trophic guild (with the exception of the micronektivorous N. elongatus) as they exhibited consumption of various types of copepods and crustaceans. The presented information on the diet and feeding strategies of the mesopelagic fishes is essential for understanding their ecological role in Mediterranean deep-sea ecosystems. Species inhabiting the mesopelagic zone constitute an important prey component of other marine species, supporting the diet of commercial fishes of high economic value [9,64] and species of conservation concern [65]. The information on the biological and ecological aspects of the six fish species, presented here, contributes to a better understanding of their position in the regional marine food web, which is essential for a holistic and effective implementation of the ecosystem approach to fisheries.