Feeding in Deep Waters: Temporal and Size-Related Plasticity in the Diet of the Slope Predator Fish Coelorinchus caelorhincus (Risso, 1810) in the Central Tyrrhenian Sea

Abstract

In-depth studies on the effect of size and period in the diet of the hollow-snout grenadier Coelorinchus caelorhincus in the Mediterranean Sea are scant and incomplete. We obtained 75 specimens of this species from the discard of deep trawl fishing on the slope of the central Tyrrhenian Sea. As corollary data, we estimated the length–weight relationship, the size frequency distribution, and composition of sexual maturity stages of the sampled individuals. We deepened stomach content analysis aiming at the evaluation of size and period’s effect in the fish diet by Costello’s interpretation of dietary indexes and correspondence analysis. The corollary results suggested negative allometric growth (b = 2.69), an asynchronous reproductive strategy (paucity of mature individuals) and a size-related bathymetrical distribution for this species (prevalence of small and intermediate-sized specimens). The prey importance index (PII) revealed that the hollow-snout grenadier is a generalist feeder on cephalopods (PII: 0–1200), fish (PII: 0–1000), crustaceans (PII: 4000–6000), and polychaetes (PII: 400–1800), and a light specialist at population level on the dominant prey among them. At the micro-taxa level, the species was found to be a generalist feeder on 10 groups of rare prey and a light specialist at population level on amphipods (PII: 1300–3200). Overall, results indicated the presence of two feeding gradients that determined an intermingled effect of size and period on fish diet. In particular, intraspecific competition and stability of food resources appeared as the factors that significantly harmonize the diet of Coelorinchus caoelorhincus in the context of the ecotrophic constraints of a deep-sea species.

1. Introduction

Determining feeding habits in fish biology is one of the most important aspects of understanding the qualitative–quantitative connections [1] between different species of fish and their prey. Intraspecific and interspecific competition for food resources [2] modeled trophic interactions in marine food webs [3,4]. Therefore, knowledge of the trophic relationships between and within species can contribute to the improvement of marine resource management [2,4].

When we refer to the term ‘generalist-opportunist’ to describe the diet of a predatory fish, we risk overlooking the fact that this feeding strategy may be more nuanced and complex than we think. In fact, ontogenetic and temporal effects intermingle to the extent that they ensure that the predator is adequately fed, according to the main ecological rules in animal feeding [3,5,6,7].

For example, food resource partitioning [8] can act within the species, determining intraspecific competition for prey [9]. The latter is targeted by the predator according to optimal foraging theory [10] (OFT, thereafter), along with its succeeding interpretations [11]. It balances between the need for proper food intake and the energy spent in feeding activity of the species. Interspecific competition is important as well, determining a harmonized species’ composition in fish assemblages [12,13] and trophic chains [14,15,16].

This is particularly true in deep continental slope environments where deep-water species have to feed on food resources that gradually decrease in abundance with increasing depth [17,18,19]. This naturally low abundance varies with recurrent surface-dependent hydrological and biological phenomena occurring at the bottom and in the water column [20,21,22,23]. In deep-water environments, the effect of seasonality on the variation of food resource abundance is dampened and temporally different compared to surface layers [24]. Consequently, food resources are a limiting factor that have been shaping the feeding strategies of the species inhabiting these environments [25,26].

In this setting, we considered the species that is the subject of this study.

The hollowsnout grenadier Coelorinchus caelorhincus (Risso, 1810) has a wide distribution, ranging from the Atlantic Ocean to the Mediterranean and the Black Sea [27]. C. caelorhincus is a benthopelagic species that lives at depths between 200 and 510 m, but has been captured both at shallower (90 m) and at deeper stands (850 m) [27,28]. The species shows a vertical distribution with smaller individuals distributed in shallower waters (<400 m) and larger individuals in deeper ones (>500 m) [29]. This may indicate ontogenetic migrations of the species toward deep waters [29]. As a result, this species is one of the most abundant in discards of deep-water trawl fishery [28,30], such as in the rose shrimp and Norway lobster trawl fishery in the central Tyrrhenian Sea [30,31]. Such discards have no commercial interest in the Mediterranean Sea [28,31]. Conversely, they have become commercially more important in ocean deep trawling, as their discards are used as a basis for fishmeal in fish farming [27].

The population structure [29,32,33], age and growth [34,35,36,37], and length–weight relationships [36,37,38,39,40,41,42] of this species were studied in different areas of the Mediterranean and European Atlantic [43].

Some dated studies on the diet of hollow-snout grenadier are available for both the Mediterranean Sea [44,45,46] and the Atlantic Ocean [47,48,49,50]. The first evidence of microplastic ingestion was recently provided in this species, as well as its association with feeding preferences [51]. Although these studies have reported an opportunistic zoo-benthic predator, it remains unclear how temporal and ontogenetic effects interact to determine the trophic habit of C. caoelorhincus, even within the generalist feeding strategy known for this species. This paper aims to provide up-to-date information on the feeding habits of C. caelorhincus, focusing on the influence of predator size and temporal variation on the fish’s diet, in the context of a deep-water adapted species. The results revealed a more complex feeding pattern than expected for a generalist-feeding species that evolved under conditions of limited food resources.

2. Materials and Methods

Samples were obtained from the bycatch of professional trawl fishing activities targeting Norway lobster (Nephrops norvegicus) on epi-bathyal grounds of the Latium continental slope (250–415 m) (Figure 1). Despite the fishery-dependent sampling, specimens were collected seasonally in four occasions between February and November 2017 so as to obtain a time-related data series, consisting of 75 specimens.

As deepwater seasonality does not coincide with coastal water seasonality in the Mediterranean marine environment [52], we grouped data into two periods (winter-spring = cold period and summer-autumn = warm period).

According to [53], individuals were measured for their total length (TL) and preanal length (PAL) to the nearest mm. Specimens were weighed to the nearest centigram and sexed after dissection. Based on [54], we identified three sexual maturity stages: juvenile immature (sex not assessed), maturing sub-adult, and mature adult. We grouped the specimens into three sizes (20 to 35 mm; 35–50 mm; 50–86 mm) based on PAL, as it resulted in a more reliable measure of fish size with respect to TL. In fact, the long and filiform caudal fin appeared to be often damaged by trawl fishing activities [53].

The length-frequency distribution was evaluated as a total and by period. A length–weight relationship (W = a × PAL^b) was fitted on natural logarithmic transformed data to detect the type of growth (parallelism test). The coefficient b is very informative, as it indicates the type of growth in the length-weight analyses [55]. A chi-square test with Yates’ correction for continuity was used on a 2 × 3 contingency table to investigate the difference in the frequency of occurrence of maturity stages between periods.

The stomach contents were frozen (20 °C) to preserve the ingested prey before sorting and taxonomical identification. Prey was observed both under a stereo-microscope and optical microscope in order to attain the lowest taxonomical level possible for prey identification. In doing this task, we used dicotomical keys and specific publications [56,57,58,59,60,61,62,63,64,65], as well as web sources [66,67,68,69]. Given the different condition of digested prey, only polychaetes and crustaceans were identified at a lower taxonomical level with respect to fish and cephalopods. Dietary indices, such as frequency of occurrence (F%) and relative abundance (N%) [70], were used to describe the feeding habits of species based on prey items found more than once in the stomachs. Prey’s weight or volume was not recorded as the condition of digested remains did not allow the use of the Index of Relative Importance (IRI), which was poorly informative and biased in this case [70]. Although not exactly the IRI (F% × N% × W% or Vol), we used values of the product F% × N% for each prey category (both at macro and micro-taxa level and stratified on size and period) as input for the analyzes (thereafter PII). Indexes were calculated both for macro-taxa (crustaceans, cephalopods, polychaetes and fish) and at family and order levels for polychaetes (onuphids, eunicids and lumbrinerids, nephtyds and trichobranchs) and crustaceans (decapods, isopods, cumaceans, mysids, amphipods, ostracods, euphausids and tanaids), respectively. Identification at the species level was achieved in a few cases only, so that statistical check was performed on data grouped by macro-taxa and by family and order in the cases of polychaetes and crustaceans, respectively. Based on the dietary indexes, we applied the Costello diagram and related diet interpretation [71,72].

We checked the statistical significance of the differences in the corresponding prey importance index (PII) of prey categories by the non-parametric Friedman ANOVA and Kendall concordance coefficient. Tests were run separately on prey data aggregated by macro- and micro-taxa level and stratified over the three size classes and the two periods in separate analyses, respectively.

We also used a polar representation of the PIIs of prey categories stratified by size classes or period of the predators. In this manner, we obtained a novel graphical interpretation of the general feeding pattern and intensity displayed by the predator on prey. Specifically, we considered the open polygons that appear in the graph when representing PIIs of prey categories according to size classes or periods at macro and micro-taxa level, separately. Their shape can be a measure of the general similarity or difference between ingestion’s patterns displayed by predators of different sizes or periods on common or prevalent prey. Considering the polygons as closed figures, we hypothesized that the area inscribed in each was a graphical measure of the overall feeding intensity exhibited by different predators toward their prey. Correspondence analysis (CA) was used as interpretative tool for the overall data using PII values of PII for each prey category stratified on both size and period at once. We interpreted the level of association between each row or column point and the two extracted dimensions by analyzing the ratio between the squared cosines of points along dimensions. Squared cosine is calculated on the angle between each extracted dimension and the line joining column or row points to bi-plot origin [73,74]. It is considered a good proxy for the magnitude of association between column or row points and dimensions [73,74].

To interpret extracted dimensions: (1) we used Kendall’s tau rank correlation coefficient to check for the non-parametric correlations that exists between PII of prey and mean length of size classes; then (2) for each prey category, we calculated the absolute value of the percent difference in relative abundance N% between periods calculated bilaterally as

D% ((X_C) − (X_W)) = |(X_C)/(X_W)^∗100) − 100|

(1)

D% ((X_W) − (X_C)) = |(X_W)/(X_C)^∗100) − 100|

(2)

where X is the relative abundance of the prey category in the cold (C) and warm (W) periods. Finally, we compared medians of absolute values of Kendall’s tau coefficient between prey associated to dimension 1 (D1) and prey associated to 2 (D2) by Mann–Whitney U test. We performed the same analysis in comparing medians of the absolute value of the percent difference between prey associated to D1 and prey associated to D2. All analyses were performed by STATISTICA 7.0 [75].

3. Results

3.1. General Biological Data

In the cold period, we sampled a total of 41 hollow-snout grenadiers, having a size range comprised between 23 mm and 73 mm PAL, and a weight range between 1.3 g and 34.6 g. Referring to the warm period, sampled individuals were 34, with a size range from 24 mm to 86 mm PAL, and a weight range from 1.7 g to 61.3 g. The frequency distribution by size class of the total sample showed an almost equal amount of individuals for the smaller and intermediate size classes, compared to a lower number found in the largest size class (Figure 2a).

Different frequency distributions were obtained by dividing the sample into the two periods. In the cold period, the intermediate size class represented the modal class, while the number of individuals decreased with the length of the fish in the warm period (Figure 2a).

The length-weight power relationship is represented in Figure 2b. Based on ln-transformed data, a parallelism test demonstrated a negative allometric growth (F = 18; df = 1; b < 3; p < 0.001).

Only four mature females were observed (one in the cold period and three in the warm period, respectively), and two maturing males were observed in the warm period. The difference in the occurrence of the three maturity stages was (Figure S1) not significant between the two periods considered (contingency table 2 × 3, X² = 1.37, d.f. = 2, p > 0.05). The individuals most represented were immature juveniles and subadults, although a size of first maturity is not available as a reference.

3.2. General Data on Diet

The taxonomical details of the identified prey are reported in

Table 1. Taxonomical details of prey found in a sample of Coelorinchus caelorhincus seasonally collected in the bathyal slope of the Central Tyrrhenian Sea. Different text indentation and prey items in bold (and corresponding abundance as number) indicate different levels of taxonomical identification and taxonomical units used for the statistical analyses, respectively.

Prey Item	N°
Teleostea	23
Crustacea tot.	288
Crustaceans ind.	42
Isopoda tot.	21
Isopoda ind.	1
-Cirolanidae tot.	12
--Cirolanidae ind.	4
--Natatolana sp. tot.	8
--Natatolana sp. ind.	7
---Natatolana borealis	1
-Gnathiidae tot.	8
--Gnathia sp.	8
Cumacea	2
Decapoda tot.	13
Decapoda ind.	9
-Stenopodidae	4
Mysida tot.	17
Mysida ind.	12
--Pseudomma sp.	5
Amphipoda tot.	164
Amphipoda ind.	107
---Eusirus longipes	3
Gammaridae tot.	46
-Gammaridae ind  -Ampeliscidae	19 53
--Ampelisca sp. tot.	27
--Ampelisca sp. ind.   -Maeridae	26 1
--Othomaera schmidti	1
-Lysianassidae	5
-Phoxocephalidae tot.	2
-Phoxocephalidae ind.	1
---Harpinia dellavallei	1
-Vibillidae	1
---Vibilia armata	1
Euphausiacea tot.	8
Euphausiacea ind.	5
--Meganyctiphanes sp. tot	3
--Meganyctiphanes sp.	2
---Meganyctiphanes norvegica	1
Otracoda tot.	11
Ostracoda ind.	4
--Cypridina sp. tot	7
--Cypridina sp.	5
---Cypridina mediterranea	2
Tanaidacea	10
--Apseudes sp.	10
Cephalopoda	80
Polycheta tot.	81
Polycheta ind.	19
-Eunicidae	18
-Lumbrineridae tot.	7
-Lumbrineridae	7
-Nephtidae tot	1
--Nephtys sp.	1
-Onuphidae tot.	35
-Onuphidae ind.	35
-Trichobranchidae tot	1
--Trichobranchus sp.	1

. Taxonomical precision was the highest among crustaceans, for which the species level was obtained in seven cases. Polychaetes were determined at the genus level (two cases) at last, whereas cephalopods and fish stopped at macro-taxa level.

3.3. Costello’s Diagrams: Ontogenetic Variation in the Diet

Thanks to Costello’s interpretation, dietary data indicated that the species exhibits an opportunistic–generalistic feeding habit, yet with relevant differences in prey’s importance between size classes (Figure 3). The hollow-snout grenadier fed on a large array of prey. Most of the species were more (several crustacean families) or less (onuphids for the small and intermediate sized individuals) rare. Others showed either a slightly dominant importance (amphipods and total crustaceans) or a high within-phenotype component (total polychaetes), all in the diet of intermediate-to-large fish (Figure 3).

Data from the Friedman ANOVA showed borderline results (Friedman. ANOVA, X² = 7.00, N = 3, gl = 3, p = 0.07) for the macro-taxa level, with a high Kendall concordance coefficient (tau = 0.78, mean rank = 0.67). Borderline results indicated that predators exploited some groups of prey in a similar manner, while others did so in a different manner. On the one hand, some prey groups exhibited a similar ontogenetic trend of prey PII along predator’s size classes. This was the case of crustaceans and polychaetes, which had higher values in size class 2 and 3, in particular in the former, compared to size 1 (Figure 4b). On the other hand, the predator exploited other prey groups by different ontogenetic trends. This was the case for fish and cephalopods, which increased and decreased in PII with fish size, respectively (Figure 4b). The high concordance conversely suggested that PII values of fish had the lowest rank within all size classes, while crustaceans and polychaetes were in the same order in size classes 2 and 3. Size 1 had cephalopods as the second prey, after crustaceans (Figure 4b).

Results were significant for the micro-taxa level (Friedman. ANOVA, X² = 18.64, N = 3, DF = 10, p < 0.05) with a lower Kendall concordance coefficient (tau = 0.62, mean rank = 0.43). In this case, results indicated a much stronger heterogeneity in the patterns displayed by predator in exploiting its prey (Figure 4a). For example, five groups over 10 had prevalence in intermediate individuals, decapods and lumbrinereids increased in importance with fish size, tanaids, and ostracods decreased (Figure 4a). In fact, the concordance in PII of prey categories between size classes was lower than in the macro-taxa level. This indicated that there were different rankings of prey categories within single size classes (Figure 4a). It is worth noting that a similar shape of open polygon size classes was drawn for prey categories at the macro (Figure 4b) and micro-taxa (Figure 4a) level, respectively. It indicated a general resemblance between ingestion’s patterns displayed by different-sized predators on common or exclusive prey. Furthermore, note the difference between the areas inscribed in each polygon when considered as a closed figure (Figure 4a,b). Based on these observations, predator size classes differentiated their feeding preferences much more at the micro-taxa level than the macro-taxa level, in which only cephalopods were different. Additionally, intermediate-sized individuals showed a higher feeding intensity than larger ones, and smaller individuals showed the narrowest. This was particularly evident at the micro-taxa level.

3.4. Costello’s Diagrams: Temporal Variation in the Diet

Costello’s interpretation confirmed the general remarks of a generalist feeder species, but there were differences between periods (Figure 5). The cold period showed generally higher values of prey indexes compared to the warm period, excepting cephalopods, cumaceans, and tanaids (Figure 5). Furthermore, the differences between periods were less marked for more abundant prey, such as total crustaceans and amphipods (Figure 5), compared to rare prey and some important period-related prey, such as polychaetes and cephalopods (Figure 5).

Statistical check on the macro-taxa level indicated temporal trends of PII were similar along periods, with cold period prey having higher PII values compared to warm ones in all cases, except cephalopods (Friedman. ANOVA, X² = 5.40, N = 2, gl = 3, p = 0.14) (Figure S2a). Concordance coefficient was very high (tau = 0.90 mean rank r = 0.80) indicating that PII’s values had same rank for crustaceans and fish within each period, whereas cephalopods and polychaetes reversed their rank in the warm compared to cold period (Figure S2a). At the micro-taxa level results were similar (Friedman. ANOVA, X² = 13.56, N = 2, gl = 10, p = 0.19), but with a lower concordance (tau = 0.68 mean rank r = 0.36). Furthermore, in this case, the PII of the cold period were higher compared to warm ones, except the values of two rare prey categories (cumaceans and tanaids, having similar values) (Figure S2b). The geometric considerations made regarding the shape and area for ontogenetic variation also applied also to temporal trends (Figure S2a,b). Based on these observations, fish that fed in the cold period had a higher feeding intensity compared to the warm period, in particular at the micro-taxa level. Nevertheless, the general ingestion pattern appeared similar between periods.

Correspondence analysis showed that the variability of the sample was fully described by two dimensions (total inertia = 0.36, χ² = 9798.30, df = 58, p < 0.001), with D1 and D2 describing 58.89% and 41.11% of total variance, respectively. As PII’ s values, some prey groups were more associated to D1 (

Table 2. Data used for interpretation of results from correspondence analyses run on values of the prey importance index (PII: F% * N%) of prey categories found in the stomachs of a seasonally collected sample of Coelorinchus caelorhincus caught on epi- and meso-bathyal grounds of the central Tyrrhenian Sea. R is the ratio between squared cosines of both row (predator’s sizes) and column (prey) points over dimension 1 and dimension 2, respectively. Ass. dim. is the most associated dimension to each row or column point. Tau is the Kendall’s rank correlation coefficient between mean size by size classes and PII values according to prey category in a given period. |D%| is the absolute value of percent difference in relative abundance (N%) calculated bilaterally between periods (C: cold, W: warm), as subscripts of prey categories, for macro-taxa prey categories (C: Crustaceans; P: Polychaetes; F: Fish; CE: Cephalopods;) and for micro-taxa prey categories (Eun: Eunicids; Ost: Ostracods; Anf: Amphipods; Mys: Mysids; Onu; Onuphids; Tan: Tanaids; Lum; Lumbrinerids; Dec: Decapods; Euf: Euphausids; Cum: Cumaceans; Iso: Isopods). PSG and PEG stand for Prey-Sharing Gradient and Prey-Exclusivity Gradient, and corresponding letters (a) and (b) as superscripts indicate which of the two gradient prevails for a given prey by a given period in the corresponding column. Symbols ↑, ↓, and ≈ stand for PII monotonically increasing, decreasing and about constant, respectively, along sizes of the predator. Sizes are size classes of the predator and numbers ₁, ₂ and ₃ indicate small (20–35 mm), intermediate (36–50 mm) and large (51–86 mm) specimens, respectively. NA: not applicable.

Prey Category	R	Ass. dim.	tau	|D%|	(a) PSG or (b) PEG
Euf C	0.002	2	0.33	92.39	(b) Size 2
Onu C	0.010	2	−0.33	67.72	(b) Size 1
Mis C	0.046	2	0.33	17.57	(b) Size 2
Anf W	0.077	2	0.33	17.62	(b) Size 2
CE W	0.143	2	−0.33	211.86	(b) Size 1
Cum C	0.176	2	na	100	na
Dec W	0.176	2	na	100	na
Lum W	0.176	2	na	100	na
F C	0.196	2	0.33	30.04	(b) Size 2
Iso W	0.256	2	−0.82	83.59	(b) Size 1
Tan W	0.256	2	−0.82	133.90	(b) Size 1
Eun C	0.326	2	0.33	236.68	(b) Size 2
Tan C	0.365	2	−1	57.25	(b) Size 1
Size 1	0.367	2	na	na	na
Euf W	0.508	2	0	48.02	(b) Size 2
Eun W	0.539	2	−0.33	70.30	(b) Size 2
Iso C	0.702	2	1	509.24	(a, b) borderline
Size 2	0.728	2	na	na	na
Lum C	0.809	2	1	na	(a, b) borderline
Ost C	1.359	1	0.33	10.22	(a, b) borderline
Cum W	1.462	1	1	na	(a) ↑ (1 < 2 < 3)
Anf C	2.021	1	1	21.39	(a) ↑ (1 < 2 < 3)
P W	2.398	1	−0.33	58.58	(a) ↑↓ (3 < 1 < 2)
CE C	2.979	1	−1	67.93	(a) ↓ (3 < 2 < 1)
Ost W	3.872	1	−1	11.38	(a) ↓ (3 < 2 < 1)
Mis W	21.782	1	−0.33	14.95	(a) ↓ (3 < 1 ≈ 2)
C C	22.582	1	1	5.90	(a) ↑ (1 < 2 < 3)
Dec C	48.908	1	1	na	(a) ↑ (1 < 2 < 3)
Onu W	63.404	1	−0.33	40.38	(a) ↓ (3 < 1 ≈ 2)
C W	72.244	1	−0.33	5.57	(a) ≈ (1 ≈ 3 < 2)
P C	88.358	1	1	141.43	(a) ↑ (1 < 2 < 3)
F W	124,999.000	1	0.82	42.94	(a) ↑ (1 = 2 < 3)
Size 3	199,999.000	1	na	na	na

) in given periods. This is the case of crustaceans and polychaetes for both periods, cephalopods in the cold and fish in warm period, as well as for other rare or moderately abundant prey (Figure 6, Table 2) depending on the period. The prey categories most associated with D2 were mostly rare prey, except cephalopods in the warm period and fish in the cold period, either associated with the small size class (size 1) or intermediate size class (size 2) depending on the period (Figure 6; Table 2). Median Kendall tau was higher in prey associated with D1 (Table 2) compared with prey associated with D2, although the results had a borderline significance level (Mann-Whitney U Test, U = 59.5, Z = −1.53, p = 0.09). On the contrary, the median percentage relative difference in abundance between periods was higher in prey associated to D2 (Table 2) compared with prey associated with D1 (Mann–Whitney U test, U = 32, Z = 2.76, p < 0.01). Prey associated to D2 showed prevalence in a given size class (1 or 2), whereas prey associated with D1 had generally monotonic patterns (increasing or decreasing) according to fish size (Table 2). Therefore, D1 was interpreted as a prey-sharing gradient. It is intended as the status of a prey for which predators exhibit an intraspecific competition such that a monotonic trend of prey importance is observed along size classes. D2 was interpreted as a prey-exclusivity gradient. In this case, exclusivity stands for the prevalence of a given prey by a given period in a particular size class of the predator compared to the others.

Polar representations of PIIs according to both size and period were provided in Supplemental Information (Figure S3a,b), together with associated Friedman statistics and comments to results (Analysis S4).

4. Discussion

4.1. Corollary Biological Data

The hollow-snout grenadier is a common fish in the grounds comprised between 200 and 500 m depth, where the species shows its highest density and biomass in the waters of the Mediterranean Sea and the Northeast Atlantic Ocean [28]. This explains the importance of its catches in the deep Mediterranean and ocean trawling [30,35,76], as also found in this study.

The present size range was similar to that obtained in the Ionian Sea by other authors [35,45], or slightly higher when compared to a sample from the Aegean Sea [37]. Given the sampling depth, the size frequency distribution of the sample was in line with the bathymetrical size distribution known for the species. In fact, pelagic juveniles originating from pelagic eggs [48]—are firstly distributed at shallower depths (<400 m). Subsequently, they progressively migrate to deeper stands (>500 m) typical of adult specimens [29,34,35]. However, exploring grounds down to the deepest bathymetric limit of the species will help further confirm current observations.

Contrasting information is available for the spawning period. D’Onghia et al. [35] hypothesized two spawning periods, one in spring and the other in autumn, as well as [34] in spring and summer. Conversely, [77] argued one spawning event between May and November. Such an uncertainty appears to be linked to the paucity of mature individuals [34,35] as found also in this work (only females, in particular summer-autumn period). This could indicate that the species has rather evolved an asynchronous and opportunistic reproductive cycle, with rapid and unpredictable gonad maturation and spawning occurring when trophic conditions are optimal [78,79,80,81,82,83]. This reproductive strategy can be associated with low metabolism and longevity (8–10 years for a maximum size of 25–30 cm in the Mediterranean [34,35,36,37]), typical of deep-water species, such as macrourids [28], and adaptive in response to decrease in food availability along slopes of deep grounds. An asynchronous reproductive strategy could be an explanation about the difficulty in estimating size at first maturity, which is still unknown for this species [67].

Fanelli and Cartes [84] found that similar macrourid species (Nezumia aequalis and Hymenocephalus italicus) had a slower metabolic response to temporal-related variation in prey abundance than Hoplostetus mediterraneus, due to different assimilation times and trophic habits.

On one side, the general paucity of food resources affects all individuals, given the negative allometry of the length–weight relationship, as obtained in this study. However, it particularly affects large specimens (mature individuals), which actually showed the highest deviations from the general regression model compared to smaller specimens. This indicates that food constraints affect mature individuals in particular, as they have high-energy requirements depending on reproductive activity, consistent with the hypothesis of asynchronous spawning. Comparing rate of growth obtained in this study with other data showed agreement [37] or disagreement [39] alternatively. As linked to trophic conditions, the ecological and hydrological local characteristics of the water masses strongly influence the species’ rate of growth [84,85,86,87,88]. Whatever the case, a negative allometric growth indicates that food is a first order constraint for this species, and this generally applies to deep-water species [89,90,91,92].

4.2. General Data on Diet

The hollow-snout grenadier is known as a bentho-pelagic feeder in a large array of macro zoo-benthic supra-benthic prey in the Atlantic Ocean and Mediterranean Sea [28,44,45,46,47,50,51,93] and this was generally confirmed by this study. In fact, taxonomical diversity of prey found in stomachs ranged throughout four different taxa of animals: cephalopods, fish, polychaetes, and crustaceans.

Based on Costello’s diagrams [71], the results showed that C. caelorhincus is a generalist feeder for more (fish, then cephalopods) or less rare prey and a specialist on crustaceans at population level. In agreement with other scientific studies carried out in the Mediterranean, our results showed that crustaceans and polychaetes were the most representative prey of the diet of C. caelorinchus [44,45,46] followed in abundance order by fish and cephalopods. In contrast to our findings, cephalopods and fish have not always been indicated as common prey for C. caelorinchus in previous studies. For instance, [44,46] did not observe cephalopods and fish in the Spanish western Mediterranean and Aegean seas, respectively, and [45] did not find cephalopods in the Ionian.

As far as crustaceans are concerned, we have observed that amphipods and isopods represented the two dominant orders, while other taxa (e.g., tanaids, ostracods, and euphasids) were probably rare or occasional prey. Therefore, the feeding mode of C. caelorinchus spans numerous benthic crustacean taxa, such as decapods, amphipods, isopods, tanaids, mysids, and ostracods; but also pelagic species, such as the euphasid Meganictyophanes norvegica.

Amphipods are an important component in the diet of C. caelorinchus [94] and our analysis showed that the families Ampeliscidae, Gammaridae, and Lysianassidae were some of the most abundant taxa among crustaceans. Some species belonging to these families were found to be particularly abundant also in the study by [45] and in some studies conducted on other macrourid species in the Mediterranean [95,96]. Of note is the presence of the maerids Othomaera schmidtii in the stomach contents, as this rare species is associated with the bathyal plane [97]. It has been indicated as one of the main prey items for some macrourids [84], but was often found as a single individual in different areas of Italian waters [59,60,62,63]. However, amphipods are an important component in the diet of several bony and cartilaginous fish [98].

The order Isopoda was represented by two species, the cirolanids Natatolana borealis and the gnathids Gnathia sp. While parasitic isopods of the genus Gnathia have already been indicated as prey of C. caelorinchus [45,46], the suprabenthic scavenger N. borealis has never been observed in the stomach contents of this macrourid. It is important to underline that N. borealis is a common and abundant isopod in deep environments [99,100,101,102] and this species was already reported in other studies concerning the diet of other deep-sea macrourids fish [84,95,96].

Among the crustaceans observed, tanaids of the genus Apseudes, decapods, mysids, ostracods, and euphasids have already been reported in the diet of C. coelorinchus [45,46]. However, among euphasids, Meganictyophanes norvegica has been found in association with the diet of other macrourids species but never in hollow snout grenadier [44,96]. It is important to emphasize that all the cited authors reported the presence of copepods in the Mediterranean area, while such a group was not found in our sample.

Out of the five families identified for polychaetes, nephtyds and trichobranchs were not included in the analysis as they were found as single items in two different stomachs. Nevertheless, data showed some degree of prey selection (high F%) towards total polychaetes, with predators contributing to dietary niche width by a high within-phenotype component [51,72]. Polychaetes are known as errant or sedentary benthic organisms common and abundant on mobile deep-sea substrates [103,104,105]. Previous studies on C. coelorinchus and other deep-sea macrourids have often highlighted the presence of polychaetes in the diets of these species [45,46,51,84,94,95,96,106,107,108,109].

Specifically, the eunicids and onuphids families dominated the polychaetofauna in the stomach contents of sampled C. caelorinchus, in partial agreement with [45], who found mainly capitellids and eunicids; while [46] observed the presence polychaetes larvae only. In the stomach contents of other macrourid species, the eunicid family is often the most abundant, but the presence of other families—such as aphroditids [45], nepthyds [84], polynoids [96], glycerids, goniadids, flabelligerids, terebellids, and pectinarids [45,84,95,96,109]—has also been observed. The typical front rostrum of the hollow-snout grenadier is used to shake the soft sediment in which the prey are dug out of their burrows or walk in proximity of them or on the substrate, as already hypothesized by [93,108]. Indeed, many species in the genus Coelorinchus are benthic foragers and prey on infauna and epifaunal polychaetes [108,109]. Whatever the predation behavior, C. caelorhincus appeared to seek these worms intensely [51], as suggested by their position in the corresponding Costello’s diagram. Overall, the comparison with previous studies carried out on C. caelorinchus highlights that our data showed minimal overlap with the data from [44], who found only decapods, amphipods, and isopods in the diet of this macrourid species, while polychaetes have not been further identified. Differently, [45] observed the same taxa found in our work for crustaceans and a family (eunicids) for polychaetes. Sever et al. [46] found the same crustacean taxa except tanaids, ostracods, and cumaceans, and no further information was provided for polychaetes. However, the difference in diet composition may be related to the changing availability of prey for the predator due to seasonality, habitat differences, variations in the composition of the local macrozoobenthic community [50,84,110,111,112,113,114,115,116,117,118,119], and vertical temporal migrations of their mesopelagic prey [114,120].

4.3. Ontogenetic Variation in the Diet: The Sharing Gradient

Ontogenetic variation in the exploitation of a given food resource implies an intraspecific competition occurring for that food resource [3,9]. Exploitation (contest) intraspecific competition applied to present case according to its definition [9]. It is known that predator skill in predation generally increases along with ontogenesis (fish size in this case). Therefore, adults feed more efficiently than young do when intraspecific competition for a given food resource occurs [3,121,122]. This is particularly true for deep-water macrourids [108,123,124,125,126], for which the effect of food partitioning [8] is intense at both the interspecific and intraspecific level because of depth-related decrease in food resources [116,127]. Thus, the importance of a contended prey generally increases with fish size [3,128,129,130,131,132,133], yet decreasing trends were also observed. The latter take place due to changing habitat during the ontogeny of both prey and predator or changing trophic needs of the latter [134,135,136,137,138]. In fact, the mouth morphology of fish has been implicated as a major determinant of variation in the types and sizes consumed by several species [128,139,140,141,142]. For example, the intensity in food resources competition has modeled the macrourids of the articulated feeding apparatus that had evolved, such as a mandibular barbell and an anterior light organ [143].

Ontogenetic changes in diet are common in fish, including macrourids [106,108,124,125,126,144], and our observations illustrated several examples of increasing or decreasing ontogenetic ingestion patterns in the diet of the hollow-snout grenadier.

In accordance with CA results, it emerged that this type of prey was associated to prey sharing dimension depending on the period. Specifically, prey sharing by predators occurred preferentially in those periods in which prey had the lowest percent change in its abundance relative to the opposite periods. In particular, prey sharing (and competition) appeared to affect intermediate and markedly large individuals, as they were associated to the sharing dimension as well (or very weakly associated to the exclusivity dimension, such as size 2). In C. caelorinchus, the sharing of prey by large and medium-sized individuals has already been observed in the work of [93]. Competition appeared stronger for more abundant groups, as observed for polychaetes [145].

Ecosystem productivity and resource diversity play an important role in intraspecific competition and regulate dietary variation between individuals in a population [146,147,148,149]. Despite the dominant or rare status of the prey in question, the results indicated that the stability of food resources and the diversity of the prey between periods are the main factors shaping intraspecific competition between predators of different sizes [150,151,152,153,154,155,156,157]. In fact, competition for food resources promotes niche variation among individuals within a population, including an important role of intraspecific competition for individual specialization [125,152,153,154,158].

Information on the bathymetrical and/or temporal distribution of these invertebrate groups would be necessary, but the available data do not provide a clear picture of their variation in abundance within the study area [145]. Whatever the case, it is very likely that such a generalist predator switches its feeding mode depending on where and when a given prey is available and based on stability of the prey’s abundance. For example, young C. caelorhincus, which is known to be distributed in the upper slope [159], are likely to feed on the prey groups available in the environment they patrol. This could be the case of the cephalopods, which are likely small-sized, suprabenthic sepiolid species [160] targeted by C. caelorhincus. Some species of this group are known to be more abundant in epi-bathyal grounds (200–350 m depth) than at deeper stands of the northern Tyrrhenian Sea [161]. Similar considerations could also apply to onuphids [145], which are considered among the most abundant and diversified polychaetes in deep waters [162].

Unfortunately, no hypothesis can be formulated on intraspecific trophic niche exclusion based on prey size or prey species. Food crushing or grinding that occurs during predation/digestion makes both the identification and the body measurement of the prey difficult or unfeasible [70,163].

4.4. Temporal Variation in the Diet: The Exclusivity Gradient

Seasonality was an essential factor in determining species distribution both in coastal and deep-water environments [115,164,165,166,167,168].

Variation of food resources within the slope is closely linked to seasonal and recurring phenomena occurring in the surface layers, such as peak primary production peaks and advective flows from continental platform [24]. Therefore, the effect of seasonality is smoothed and time-buffered in the slope compared to coastal waters [52]. The former enriches slope environment of organic matter in summer-autumn period through a cascade effect, whereas the latter supplements macro-zoobenthic community of new terrigenous material in spring, as found in the middle slope of Catalan Sea [127]. Despite the local circulation of water masses and physical profile of deep grounds have to be taken into account, such a temporal pattern generally applies to Mediterranean slope environment [24]. For instance, high densities of suprabenthic and meso-pelagic crustaceans were observed in spring and summer, respectively, for the bathyal eastern Ionian Sea [116]. Temporal variations in trophic habit for similar deep-water benthic/subrabenthic predators relate first to abundance of available prey in such an environment [84,116]. An indirect proxy of the environmental abundance of prey is generally represented by the relative abundance index (N%) of stomach contents [70].

Bilaterally calculating the percent change in relative abundance provided a finer descriptor for the variation in prey abundance between periods. In fact, the observed percent variations in prey abundance were different when prey abundance in a period was set as the initial or final value relative to the opposite period, in Formulas (1) and (2). This allowed representing the circular effect of seasonal recurrent phenomena that influence prey abundance in the deep environment [24,52,84,116,127] inhabited by the species.

For instance, cephalopods were the group showing the largest relative variation between periods at a macro taxa level, being more than twice as abundant in the warm period compared to the cold one. The increase in abundance of cephalopods in the warm period was also observed by [30], who analyzed the commercial catches and discards composition in the same area. This study showed an increase in the abundance of many species of cephalopods (eg Sepiola spp., Sepia orvigiana, Todaropsis eblanae, etc.) in summer and spring and especially between 240 m and 500 m.

Polychaetes were more than once more abundant in cold with respect to the warm period. At micro-taxa level, isopods were even more than five times more abundant in the cold period compared to warm one, tanaids and euphausids were about more than twice as abundant in warm and cold periods compared to opposite periods, respectively. Regarding benthic detritivores (e.g., some polychaetes, Apseudes sp. and the amphipods of the genus Ampelisca), their variation in abundance could be linked to the variation of organic matter in sediments [169,170,171,172,173], which in turn depends on seasonality [174,175,176,177]. In fact, the organic matter depositions play a decisive role in the seasonal variation of seafloor biogeochemical cycles and food-web dynamics, and—at the same time—it acts on the structure of marine benthic communities [176,178,179,180,181].

Overall, high abundances related to the cold period could be linked to advective flows, such in the case of polychaetes [145] and benthic crustaceans, which could benefit from the new terrigenous input of the vernal. Similarly, summer input of organic matter could benefit cephalopods due to an increase in their prey, such as mesopelagic and migrating crustaceans, which are also hollow-snout grenadier’s prey as well. According to our data, the cold period appeared as the time-span favoring a trophic niche expansion for the macrourid species considered.

Unfortunately, there was not the possibility of a comparison with Mediterranean previous works [44,45,46] as they had not found any size and temporal-related variation in the diet of the species. The dominance at population level in the diet seems to be linked both to high values of relative abundance and frequency of occurrence and to small temporal fluctuations of these dietary indexes. Otherwise, large variations of the latter were observed for prey classified as more or less rare in the diet of C. caelorhincus. Therefore, the species takes advantage in feeding on prey which are always abundant—such as crustaceans—as a population-level-related specialist, and on more rare prey depending on their temporal availability as a generalist-opportunistic feeder.

CA results showed that this type of prey was associated with the prey exclusivity dimension according to the period. Specifically, the prevalence of a prey in a given predator size class occurred in those periods in which the prey had the highest percent variation in its abundance relative to the opposite period. This may indicate that the prevalence of a given food resource for a given size class is enhanced in periods having large fluctuations in abundance of that resource.

Prey exclusivity was noted particularly in small sized individuals, as they showed the strongest association to such a dimension, and in intermediate individuals, yet weakly. Young specimens, distributed in the epi-bathyal grounds [29,32,33], feed in particular on prey whose temporal fluctuations in abundance are wide. This is coupled with the effect of seasonality, which may be stronger in the upper strata compared to the deeper strata of the bathyal plane [52]. Predators feeding on prey groups when they are more available optimally exploit these food resources [10] according to period.

Fanelli and Cartes [84] give strength to actual observations, demonstrating by stable isotope analysis that similar macrourids (N. aequalis and H. italicus) displayed an intra-annual shift of stable isotopic signature δ15 N between September (lowest values, warm period in the present work) and April to June (highest values, cold period).

The weak effect of both gradients observed in the intermediate-sized individuals introduces the next chapter. Sharing and exclusivity are not independent, rather they interact such that intermediate individuals are in between the two forces in action.

4.5. Sharing vs. Exclusivity

The predator sharing and exclusivity gradients showed a similar importance in determining the variation in the diet of hollow-snout grenadier. As shown by CA results, such an equivalence is summarized ultimately in what is observed for the intermediate individuals, which suffer from effects of both gradients. Therefore, they are likely to experiment mixed ecological conditions, such as prey availability that changes with increasing depth and shifting period.

Such specimens move from epi-bathyal, typical of younger specimens, to deeper stands, typical of adults [29,32,33]. On one side, these predators patrol a benthic environment that progressively varies by depth because of ontogenetic migration of a very slowly growing species [34,35,36,37]. On the other hand, the temporal variation in prey abundance, and abundance in itself, decrease with increasing depth. Intermediate-sized individuals obviate these limitations by simultaneously showing exclusivity and sharing for prey with a low and a high abundance stability, respectively. Furthermore, in this case, such a feeding habit could be functional to maximize food intake for these individuals [10]. In fact, individuals who experiment slow-changing trophic conditions are likely to manifest the highest diversity in their feeding habits. Generalizing the results, this has some links with the theory of intermediate perturbation [182]. According to it, ecological systems that experiment intermediate perturbations along ecological gradients are in a dynamic stable equilibrium [183] and exhibit the largest ecological heterogeneity [184]. Actually, intermediate-sized individuals showed the highest feeding intensity on prey compared to largest and in particular to smaller individuals. According to [11], the dietary niche expansion under intraspecific competition is driven by a given group of individuals that stably shifts their feeding preference toward new prey when preferred prey at the species level is less available. In this context, trophic relations that can alter the population density and structure of the hollow-snout grenadier should also be considered. Predation is one of these relationships, since macrourid and mictophyd fish are included in the diet of deep-sea dominant predators, such as several species of deep sharks [185,186,187].

Parallel environmental data on spatio-temporal variation in the density of deep macro-benthic and supra-benthic prey will be necessary to test the present observations. Stable isotope analysis will also be needed, as it is extremely sensitive to temporal-related trophic variations, as demonstrated by [84] in similar macrourids species. The frontier is the meta-barcoding analysis applied to the composition of the prey in stomach contents [188]. Providing the identity of single species or groups in fish diet [189] will be a giant step toward a thorough comprehension of the ecological factors ruling fish feeding.

5. Conclusions

In conclusion, C. caelorhincus revealed more complex size- and period-related patterns in its diet than expected for a generalist zoo-benthic feeder. The marked interchangeability of species’ trophic habits can be the result of the combined effect of both contest intraspecific competition between different predators (sharing gradient) and the stability of food resources (exclusivity gradient). These appeared as forces harmonizing the need for adequate food intake with living in a difficult environment for well-adapted deep-water species, such as C. caelorhincus. As indicated by corollary biological data, a negative allometric growth, a size-related depth distribution, and a likely asynchronous reproductive strategy are also emblematic of life history traits that have been modeled under the depth-related decrease in food resources.