Ommastrephes caroli (Cephalopoda: Ommastrephidae) from the Adriatic Sea: Morphometry, Age, and Genetic Characterization

Abstract

This study gives the first data on the body and beak morphometric characteristics, age, and genetic structure of neon flying squid, a rarely caught cephalopod in the Adriatic Sea. We identified specimens as recently resurrected Ommastrephes caroli species using two mitochondrial markers, 16S ribosomal RNA gene and cytochrome c oxidase subunit I gene. Overall, 23 juveniles (3 females, 3 males, and 17 unsexed), with a dorsal mantle range of 65–152 mm, were caught in September 2020 in the waters of the Korčula Channel, island of Palagruža, and island of Jabuka, thus providing the most abundant sample of this species in the Mediterranean waters. The length–weight relationship showed an isometric growth. The results of the beak/length regressions suggest hood length is a useful characteristic for biomass estimation studies, as it showed a good linear fit to the dorsal mantle length. Statolith growth increments were easily visible and statolith microstructure analysis was successfully used to determine the age of 22 individuals. The estimated age ranged from 36 to 64 days (mean = 48 days). The back-calculation analysis showed that the squid hatched during July and August 2020, indicating that O. caroli spawns during the warmer, summertime period. Considering the size and age of the caught individuals, the Adriatic Sea could represent a potential feeding ground for this species. The genetic structure analyses indicate the existence of separate Atlantic and Mediterranean/Adriatic subclusters; however, this warrants further investigation.

1. Introduction

Family Ommastrephidae Steenstrup, 1857 includes many economically and ecologically important squid species distributed in all oceans from the sub-Arctic to sub-Antarctic seas [1]. These abundant, muscular, and fastest-growing squids inhabit shelf, continental slope, and open ocean waters, from the surface to depths of 2000 m, and are one of the most exploited invertebrate fishing resources, representing 50% of the total world cephalopod catch [1]. In 2021, the total catch of Ommastrephidae in the North Atlantic and Mediterranean waters was 73,590 tons [2]. Oceanic ommastrephids are considered an abundant underexploited fishery resource, with only Ommastrephes bartramii and Dosidicus gigas commercially exploited in the Pacific waters [1,3].

Despite ommastrephids’ economic significance, the phylogeny of the Ommastrephes genus has only just recently been resolved by Fernández-Álvarez et al. [4]. Ommastrephes d‘Orbigny, 1834 was considered a monotypic genus with only one cosmopolitan species O. bartramii (Lesueur, 1821) with discontinuous distributions between the southern and northern hemispheres and with three geographically different populations: the North Atlantic, the North Pacific, and the Southern Hemisphere (the South Atlantic, Indian Ocean, and Southwestern) [1,5]. However, previously observed morphological (mainly spermatophore morphology) [6,7] and metabolic (cholinesterase activities of optical ganglia) [8,9] differences between these populations inhabiting different geographic regions brought the monotypy of the Ommastrephes genus into question. Fernández-Álvarez et al. [4] performed a molecular analysis of individuals collected throughout the genus distribution area using two mitochondrial markers (COI, subunit I cytochrome c oxidase and 16S rRNA, ribosomal RNA). Their results showed that the genus Ommastrephes is an allopatric cryptic complex of four species with the following proposed nominal names and distributional ranges: Ommastrephes caroli (Furtado, 1887) (Northeast Atlantic), Ommastrephes cylindraceus d‘Orbigny, 1835 (East Tropical and South Atlantic with South Indian), Ommastrephes brevimanus (Gould, 1852) (South Pacific), and Ommastrephes bartramii (Lesueur, 1821) (Northwest and Central North Pacific). Furthermore, the authors found that O. caroli represented a pseudocryptic species, most phylogenetically divergent from others, indicating a longer evolutionary history due to the isolation from the remaining congeneric species.

In the Mediterranean Sea, mesopelagic O. caroli is still poorly studied, most likely because it is rarely caught by fishing gear and the lack of experimental research in the mid-waters [10]. In addition to sporadic findings of moribund females drifting in shallow waters or stranded on shore, these squids are occasionally found in the gastric contents of teuthophagous predators [11,12]. However, since 2004, the frequency of observations of O. caroli in the Mediterranean has been increasing [10], suggesting possible changes in species distribution and abundance. So far, reported records of juvenile individuals [13,14,15] are scarcer than those of larger adults, mostly stranded post-spawning females [10,16,17,18,19,20]. The known range of this species currently covers the entire Mediterranean Sea, i.e., western and central parts, Taranto Bay, Adriatic, Aegean, and Levantine Seas [4,10,15,20]. According to Lefkaditou et al. [10], this recent record increase in the Northwest and Northeast Mediterranean could be a consequence of warmer sea surface temperatures that drive the northward expansion of native warm-water species. More recently, Lefkaditou et al. [10] reported on the body and beak morphometry, diet, and maturity of adult specimens collected from the Aegean and Ionian Sea (Easternmost Mediterranean), while Agus et al. [20] investigated the morphological, biometric, and reproductive characteristics and assessed the age based on the beak and lens analysis of specimens from the Sardinian waters (Western Mediterranean). Furthermore, Agus et al. [20] confirmed the findings of Fernández-Álvarez et al. [4] and additionally suggested potential genetic differentiation between the Mediterranean and Atlantic populations.

In the Adriatic Sea, the first record of O. caroli dates back to March 1910, when an individual of 165 mm DML was caught near Vodice (Central -Eastern Adriatic) [21]. In March 1986, in the Central Western Adriatic (20 mls off Falconara Marittima), a female of 560 mm DML was caught by midwater trawl [22]. More recently, Franjević et al. [18] reported a large female caught with a spear at a depth of 1.5 m in February 2013 in the waters of the island of Šipan (Central Eastern Adriatic), weighing 9 kg and measuring 1.3 m in length (without tentacles). Franjević et al. [18] identified this specimen using the mitochondrial COI gene.

Apart from the three above-mentioned findings, of which only one was molecularly identified, knowledge about O. caroli in the Adriatic is lacking. Therefore, when we collected a larger sample of this species, it was a rare opportunity to investigate and complement the scarce existing knowledge about this oceanic squid. The aim of this study was to describe the morphological features of the body and beak, estimate the age using statolith microstructure analysis, and genetically identify and characterize the collected individuals using two mitochondrial genes (COI and 16S rRNA). The results of the present study will greatly contribute to future taxonomic and ecological research of the Ommastrephes genus.

2. Materials and Methods

2.1. Sampling

A total of 23 Ommastrephes caroli squid were collected in the central part of the Adriatic Sea during mid-September 2020 (Figure 1). Individuals were caught with commercial purse seine in the area of the Korčula Channel (N = 8; 42.9852 N, 16.8426 E) and with squid jig in the waters of the island of Palagruža (N = 14; 42.3896 N, 16.2632 E) and the island of Jabuka (N = 1; 43.0991 N, 15.4332 E) during the experimental research cruise of the Institute of Oceanography and Fisheries, Split. Squid remained afloat on the sea surface attracted by vessel lights, and at times, even grouped with Loligo vulgaris individuals (D.R., D.U., pers. obs.). In the waters of the island of Jabuka, squid were recorded jumping up to 2 m above the sea surface, while at Palagruža, this behavior was not noticed. All specimens were frozen immediately after capture for further laboratory analysis.

2.2. Morphometry and Age Analysis

After thawing at room temperature in the laboratory, the following morphometric measurements were made according to Petrić et al. [24]: total body length (TL), dorsal mantle length (DML), head length (HL), fin length (FL), fin width (FW), mantle width (MW), head width (HW), width between the eyes (EW), lens diameter (LD), and arm IV length (AL-R: right, AL-L: left). Furthermore, body weight (BW) was recorded. Sex was determined macroscopically, by the appearance of the gonads and accessory reproductive organs, in only 6 individuals, 3 males and 3 females. Squid maturity was assigned according to Jereb and Ragonese [25]. In females, small, thin, and translucent nidamental glands were visible, as well as a translucent and filamentous ovary. In males, testis, spermatophoric organ, and penis were also small and barely perceptible.

The beaks were surgically removed from the buccal cavity and stored in 70% ethanol for further analysis. The following morphometric features of the upper (U) and lower (L) beaks were measured using a digital caliper to an accuracy of 0.01 mm [24,26]: hood length (HL), crest length (CL), rostral length (RL), wing length (WL), jaw angle width (JW), width of the lateral wall (LWa) in the upper (U) beak, and length of the baseline (BL) in the lower (L) beak. According to the previous literature [10,27], linear (y = a + bx), power (y = ax^b), exponential (y = ae^bx), and logarithmic (y = aln(x) + b) models were fitted to explore the relationship between the DML (dependent variable) and beak features (independent variable). The distribution of the data was visually inspected by quantile–quantile plots. Akaike’s information criterion with correction for small sample sizes (AICc) [28] was used to evaluate each model, as implemented in AICcmodavg package [29] for R v4.3.0 [30]. Nonlinear models were fitted in R by the nls() nonlinear least squares method, and linear with lm() function. Bootstrapping was used to evaluate the stability of the best model choice (1000 bootstrap iterations) and to determine 95% confidence intervals for the a and b coefficients of best-rated models (10,000 bootstrap iterations).

The length–weight relationship between the dorsal mantle length (DML) and body weight (BW) was calculated using the equation BW = aDML^b. The parameters were estimated by the nonlinear least squares method in R [30], followed by the estimation of 95% confidence intervals for a and b by bootstrapping (10,000 iterations) as implemented in the nlstools package for R [31].

The individual age of squids was estimated using statolith growth increment analysis. Total statolith length (TSL) was measured from the end of the dorsal dome to the top of the rostrum (accuracy of 0.01 mm) (Figure 2). Preparation and processing of statoliths were made according to Ceriola and Milone [32]. Briefly, the concave side of the statolith was mounted onto a microscopic slide with the thermoplastic resin (Crystal BOND^TM), then ground and polished using lapping film sheets (30, 12, 5, and 0.05 μm grades). Statolith microstructure was examined under an Olympus BX51 light microscope at 400× magnification. Statolith growth increments were counted from the natal ring to the edge of the dorsal dome using the Olympus CellˆA Image Analysis Software v2.4.106, following the assumption of daily increment deposition [33,34,35]. For each statolith, increments were counted independently by two readers and the mean number of growth increments between these two readings was expressed as total age in days. The hatching date was back calculated from the date of capture and increment counts. The linear relationship between the statolith length and dorsal mantle length was estimated by lm() function in R [30].

2.3. DNA Extraction, PCR Amplification, and Sequencing

In the laboratory, a small piece of mantle tissue was preserved in 96% ethanol and stored at +4 °C for later DNA analysis. Total genomic DNA was extracted from the mantle tissue by proteinase K digestion, following the saline extraction protocol of Martínez et al. [36]. Two mitochondrial genes were amplified: 16S ribosomal RNA gene (16S; 480 bp) using primers 16S ar (5′CGCCTGTTTATCAAAAACAT3′) and 16S br (5′CCGGTCTGAACTCTGATCAT3′) [37] and cytochrome c oxidase subunit I gene (COI; 650 bp) using primers LCO1490 (5′GGTCAACAAATCATAAAGATATTGG3′) and HCO2198 (5′TAAACTTCAGGGTGACCAAAAAATCA3′) [38]. PCR reactions were carried out in a final volume of 25 µL containing the following: 1 × PCR buffer, 1.5 mM MgCl2, 0.2 mM dNTP, 0.4 µM of each primer, 1 U Taq Polymerase (Sigma Taq Polymerase (MiliporeSigma, Burlington, MA, USA) for 16S gene and GoTaq^® G2 Hot Start Polymerase (Promega Corporation, Madison, WI, USA) for COI), and DNAse/RNAse free water. The DNA template was set to 75 and 50 ng for 16S and COI, respectively. For both genes, PCR was performed as follows: 3 min at 94 °C, 5 cycles of 30 s at 94 °C, 1.5 min at 45 °C, 1 min at 72 °C plus 35 cycles of 30 s at 94 °C, 1.5 min at 50 °C, and 1 min at 72 °C, followed by a final extension of 7 min at 72 °C. Successful PCR products were verified using 1% agarose gel electrophoresis and sent to Macrogen Europe (Amsterdam, The Netherlands) for sequencing service.

2.4. Genetic Data Analysis

DNA fragments of 16S and COI genes were successfully amplified from 22 and 21 individuals, respectively. Obtained sequences were checked, trimmed, and aligned using Geneious Prime v2024.0.2 (https://www.geneious.com (accessed on 15 March 2023); Auckland, New Zealand) [39] and then deposited to GenBank (National Center for Biotechnology Information, Bethesda, MD, USA) under the Accession numbers PP839008-PP839028 and PP839030-PP839051 for 16S and COI, respectively. Publicly available Ommastrephes genus 16S and COI gene sequences were retrieved from the GenBank database and used for further phylogenetic analyses and molecular identification (Accession numbers in Supplementary Tables S1 and S2). The best-fit evolutionary models were selected based on the Bayesian information criterion (BIC) calculated using IQ-TREE web server with auto model selection [40,41,42,43]. Geneious Prime v2024.0.2. was used to reconstruct phylogenetic trees using Bayesian inference (BI) and maximum likelihood (ML) methods, with HKY+F+I evolutionary model for 16S and HKY+F+G4 for COI. Dosidicus gigas AB270959 and AB270944 were used as the outgroup species for 16S and COI sequences, respectively. Phylogenetic trees were subsequently visualized and graphically processed using FigTree v1.4.4. (http://tree.bio.ed.ac.uk/software/figtree (accessed on 15 March 2023); Edinburgh, UK). Pairwise genetic distances (p-uncorrected) within and between Ommastrephes species were calculated with MEGA 11 for both molecular markers [44,45].

To characterize the diversity of the O. caroli Adriatic sample in respect to other findings from different geographical locations, a global set for the O. caroli was constructed by using sequences from the Northeast Atlantic, Central Western Mediterranean, and Adriatic. Two 16S sequences were removed due to uncertain geographic origin (KY793561 and KY793587). In total, this consisted of 35 sequences for 16S gene: 11 from the Northeast Atlantic, 3 from the Central Western Mediterranean, and 21 from the Adriatic, while the COI global dataset included a total of 30 sequences: 4 from the Northeast Atlantic, 3 from the Central Western Mediterranean, and 23 from the Adriatic. Genetic variability was measured using the DnaSP program 6.12.03 [46] by calculating the number of polymorphic sites (S), haplotypes (H), haplotype diversity (Hd), nucleotide diversity (π) [47], and the mean number of differences between sequence pairs (k) [48]. In DnaSP, Raggedness (r) [49] and Ramos-Onsins and Rozas’s (R2) [50] were calculated. Haplotype network maps were constructed by the median-joining distance algorithm in PopART v1.7 software [51]. ARLEQUIN v3.5.2.2. [52] was used to estimate genetic differentiation indexes: F_ST (based on haplotype frequences alone) and Φ_ST (taking into account genetic distances: Tamura and Nei distance matrix with no gamma correction was used). The significance of the pairwise F_ST and Φ_ST values was tested with the Exact test of population differentiation (No. of steps in Markov chain = 100,000), and the corresponding p-values were adjusted using Bonferroni correction. Demographic history changes were calculated using Tajima’s D [48] and Fu’s Fs [53] statistics implemented in ARLEQUIN v3.5.2.2. [52].

3. Results

3.1. Morphometry and Age Analysis

All collected Ommastrephes caroli specimens were immature juveniles (Figure 3). Dorsal mantle length (DML) ranged from 65 to 152 mm (mean 103.27 ± 26.55 SD) and body weight (BW) from 7.57 to 89.80 g (mean 27.79 ± 25.26 SD). Fin length (FL) varied from 18 to 62 mm (mean 36.13 ± SD) and fin width (FW) from 44 to 109 mm (mean 68.20 ± 20.19 SD). The measurements of the body morphometric characters of all specimens are given in

Table 1. Body morphometric characteristics (mm) of Ommastrephes caroli collected in the Adriatic Sea (F—female, M—male, N—sex not determined, for abbreviations see M and M section).

ID	Sex	BW	DML	TL	HL	FL	FW	MW	HW	EW	LD	AL-R	AL-L
1	N	75.20	149	287	26	59	104	37	29	24	9	57	59
2	F	89.80	152	298	33	62	109	40	28	24	8	63	61
3	M	74.60	145	288	26	56	96	33	27	26	8	58	56
4	F	84.70	149	266	29	59	105	35	28	25	6	64	65
5	N	34.20	119	224	22	43	75	25	19	14	4	44	44
6	M	39.70	112	241	25	42	81	30	20	14	6	44	44
7	M	35.70	115	205	23	41	77	29	20	16	7	41	38
8	N	31.20	103	202	22	37	73	28	19	15	6	37	36
9	N	34.40	111	229	22	41	76	28	18	14	6	46	44
10	N	28.50	109	199	21	37	72	26	17	14	6	35	38
11	N	16.00	90	168	16	29	56	21	17	14	4	25	27
12	N	23.60	102	191	20	37	68	27	19	17	6	38	38
13	N	16.20	90	172	13	30	59	24	13	11	3	26	27
14	F	53.20	131	257	24	48	89	34	25	21	6	55	56
15	N	20.70	96	170	16	35	59	24	14	10	4	29	30
16	N	40.84	116	249	30	43	81	34	14	11	-	49	47
17	N	7.57	65	133	19	19	44	24	12	9	5	26	27
18	N	10.76	72	152	21	23	49	22	13	7	6	24	28
19	N	16.15	89	179	19	28	56	24	11	9	4	33	35
20	N	12.39	82	162	19	26	47	24	14	9	6	28	29
21	N	14.06	84	168	18	25	52	26	14	6	5	32	31
22	N	10.02	71	137	18	18	45	26	10	6	4	27	28
23	N	9.68	73	144	12	23	49	21	11	8	4	29	28

.

The length–weight relationship for all 23 individuals was described by the following equation: BW = 0.00001893 DML^3.052. The value of coefficient b did not significantly differ from 3 (95% bootstrap confidence interval: 2.86–3.23), suggesting the growth of the analyzed juvenile individuals is isometric, i.e., proportional to the weight and length.

The beaks of the sampled individuals had a dark brown pigmentation limited to the rostrum and front part of the hood. The lower beak’s crest displayed light brown coloring, whereas the lateral wall of the upper beaks displayed no color (Figure 4).

Summary statistics of the beak morphological variables are presented in

Table 2. Beak morphological variables (mm) of Ommastrephes caroli (N = 23) collected in the Adriatic Sea (for abbreviations see M and M section).

Variable	Min	Max	Mean	SD
UCL	4.61	12.03	7.79	2.10
UHL	3.65	10.12	6.29	1.99
UJW	0.71	2.05	1.16	0.44
ULWa	1.67	5.43	3.53	1.07
URL	0.87	2.85	1.56	0.64
UWL	0.95	3.01	1.81	0.58
LBL	0.72	6.70	3.70	2.18
LCL	1.89	6.31	3.97	1.34
LHL	0.96	3.83	2.28	0.75
LJW	0.78	3.42	1.79	0.70
LRL	0.92	2.57	1.50	0.47
LWL	2.49	4.90	3.50	0.69

. The upper (UHL) and lower hood length (LHL) ranged from 3.65 mm to 10.12 mm (mean 6.29 ± 1.99 SD) and from 0.96 mm to 3.83 mm (mean 2.28 ± 0.75 SD), respectively. All measures displayed a relative increase with the increase in the body size (DML), except for the LJW, which showed a wider spread of data, indicating possible difficulties with taking this measure. According to the AICc criterion and bootstrap procedure, it was determined that the increase in the DML with the LHL, UHL, ULWa, and UWL was best described as linear, exponential for the LBL, LCL, and LWL, logarithmic for the LRL, LJW, and UJW, and power for the UCL and URL (Figure 5). However, the AICc values were very close for all models, suggesting that they all equally well describe the relationship between the beak measures and DML (Supplementary Table S5). This was also evident in the instability of the best model choice observed during bootstrapping for the UJW and URL, where the model initially chosen as the best was not chosen in the majority of bootstrap subsamples. The small sample size in the current dataset is an important contributing factor in this, also evident in wide 95% confidence intervals for the a and b parameters for all models (Supplementary Table S5).

Out of 23 processed Ommastrephes caroli individuals, age could not be determined in only 1 squid (ID 23) because the statolith pair was broken. The age of the youngest individual was 36 days (82 mm DML), and the oldest was a female at 64 days old (152 mm DML). For the total sample, the mean value was 47.57 days (SD = 8.59).

The total statolith length (TSL) ranged from 0.5 to 1.2 mm, with a mean value of 0.91 mm (SD = 0.18). There was a good linear relationship between the DML and TSL (DML = 133.7 × TSL − 15.87, R² = 0.88). The AICc criterion indicated that the power model might better describe this relationship; however, due to the small sample size in this study and limited size and age range, further growth models were not investigated.

The back-calculation analysis showed that the individuals hatched during July and August 2020, indicating that O. caroli spawns during the warmer, summertime period.

3.2. Phylogenetic Analysis and Genetic Characterisation

In the present study, sequencing of mitochondrial DNA fragments yielded 21 16S and 22 COI sequences of good quality. Bayesian inference (BI) and maximum likelihood (ML) tree construction methods displayed similar topology for both investigated gene fragments. Two main clades were resolved with both genes: one clustering all Adriatic sequences from this study with Mediterranean and North Atlantic sequences from the literature, corresponding to nominal species O. caroli [4] and the other composed of the other three species: O. bartramii, O. brevimanus, and O. cylindraceus (Figure 6 and Figure 7). Main clades were resolved with high support (BI posterior probability of 100% for both genes; ML bootstrap values of 50.7% for 16S and 96.5% for COI). Within the O. caroli clade for COI, but not for 16S, a slight substructuring of the Atlantic samples is visible, suggesting possible genetic differentiation. The COI phylogeny resolved the O. bartramii, O. brevimanus, and O. cylindraceus clade into three separate groups, with O. bartramii a sister species to the other two (BI posterior probability of 62.28%; ML bootstrap value of 47.3%). This is not clearly resolved in the 16S tree, indicating a closer relationship of O. bartramii and O. cylindraceus (Figure 6).

The reconstructed phylogenetic topology was also consistent with the intra- and interspecies genetic distances. The overall mean uncorrected p-distances were 1.38% (s.e. = 0.00377) and 4.68% (s.e. = 0.00552) for the 16S and COI genes, respectively. The interspecific uncorrected p-distances between Ommastrephes species ranged from 0.29 to 5.22% for 16S and 2.44 to 9.40% for COI. The interspecific mean p-distances are shown in

Table 3. Interspecies mean uncorrected p-distance (%) of the Ommastrephes genus for 16S ribosomal RNA (16S rRNA) and cytochrome c oxidase subunit I (COI) gene regions marked in bold.

Gene Region	N	O. bartramii	O. cylindraceus	O. brevimanus
16S rRNA
O. bartramii	12	-
O. cylindraceus	16	1.05	-
O. brevimanus	3	0.87	1.20	-
O. caroli	37	2.19	2.51	1.39
COI
O. bartramii	6	-
O. cylindraceus	10	3.70	-
O. brevimanus	3	3.43	2.51	-
O. caroli	31	8.48	9.13	9.24

. The intraspecific distances were lower and ranged from 0 to 3.77% for 16S and 0 to 1.75% for COI (

Table 4. Intraspecies mean uncorrected p-distance (%) of the Ommastrephes genus for 16S ribosomal RNA (16S rRNA) and cytochrome c oxidase subunit I (COI) gene regions marked in bold.

Gene Region	N	Mean	Range
16S rRNA
O. bartramii	12	0.84	0–3.77
O. cylindraceus	16	0.15	0–0.58
O. brevimanus	3	0	0
O. caroli	37	0.25	0–1.16
COI
O. bartramii	6	0.64	0–1.75
O. cylindraceus	10	0.08	0–0.35
O. brevimanus	3	0	0
O. caroli	31	0.20	0–0.70

). Tables with all the calculated p-distances for 16S and COI are available in Supplementary Tables S3 and S4, respectively.

Genetic diversity estimates for each fragment gene (16S and COI) and geographical location are presented in

Table 5. Genetic diversity estimates, neutrality tests, and population size change statistics for 16S ribosomal RNA (16S rRNA) and cytochrome c oxidase subunit I (COI) gene regions (marked in bold) of Ommastrephes caroli from the Adriatic, Central Western Mediterranean, and Northeast Atlantic and pooled global dataset.

Gene Region	N	H	S	Hd	π	k	D	FS	r	R2
16S rRNA
Adriatic	21	5	3	0.352	0.0013	0.457	−1.1867	−3.188 *	0.1891	0.0932
Central Western Mediterranean	3	1	0	0.000	0.0000	0.000	0.0000	0.000	-	-
Northeast Atlantic	11	3	2	0.345	0.0011	0.364	−1.4296	−1.246	0.2030	0.1928
Global dataset	35	7	5	0.605	0.0022	0.763	−0.9755	−3.502	0.1201	0.0815
COI
Adriatic	23	5	5	0.565	0.0014	0.857	−1.1533	−1.447	0.0570	0.0872
Central Western Mediterranean	3	2	1	0.667	0.0012	0.667	0.0000	0.201	0.5556	0.4714
Northeast Atlantic	4	2	1	0.667	0.0012	0.667	1.6329	0.540	0.5560	0.3333
Global dataset	30	8	8	0.671	0.0019	1.074	−1.4228	−3.661	0.0468	0.0665

N—number of individuals, H—number of haplotypes, S—number of polymorphic sites, Hd—haplotype diversity, π—nucleotide diversity, k—average number of nucleotide differences, D—Tajima’s statistic, Fs—Fu’s statistic, r—Raggedness statistic, R₂—Ramos-Onsins and Rozas statistic, * statistical significance (p < 0.01).

. Due to the small sample size for the Central Western Mediterranean (both genes) and Northeast Atlantic for COI, these should be interpreted with caution and are here for exploratory purposes only. A total of seven and eight haplotypes were identified in the global O. caroli 16S and COI datasets, respectively, indicating moderate haplotype diversity (global 16S dataset Hd = 0.605, global COI dataset Hd = 0.671) and low nucleotide diversity (global 16S dataset π = 0.0022, global COI dataset π = 0.0019) in the analyzed groups.

The Adriatic 16S dataset (463 bp) revealed three variable polymorphic sites (one singleton and two parsimony informative) and five haplotypes. Both the haplotype diversity (Hd = 0.352) and nucleotide diversity (π = 0.0013) were low. A somewhat higher haplotype diversity (Hd = 0.565) and low nucleotide diversity (π = 0.0014) were shown in the Adriatic COI dataset (612 bp), with five polymorphic sites (one singleton and four parsimony informative) and five haplotypes. Demographic estimators showed non-significant negative Tajima’s D and Fu’s Fs values for the global set; however, there was a significant negative value for Fu’s Fs for the Adriatic sample for the 16S gene. This generally indicates there are no departures from neutral evolution; however, there might be signs of demographic expansion of the Adriatic sample.

Median-joining haplotype networks based on the global 16S dataset showed that haplotype 4 was most frequent, shared by 17 Adriatic and all 3 Central Western Mediterranean individuals. Two individuals each from the Northeast Atlantic represented two haplotypes (H2 and H3), while the remaining nine shared haplotype 1 with only one Adriatic individual (Figure 8). For the global COI dataset, haplotype 1 showed the highest frequency in the whole sample, shared by 17 individuals, of which 15 were from the Adriatic and 2 from the Central Western Mediterranean. Two haplotypes were exclusively comprised of two North Atlantic individuals each (H2 and H3), and one haplotype included only one Central Western Mediterranean individual (Figure 8). All haplotypes differ from one another by only one mutation, except haplotype 6 for COI. Generally, the haplotypes are not shared between Atlantic individuals and others.

The observed haplotype network configuration was also reflected in the differentiation indexes. The global pairwise F_ST for 16S and COI genes was 0.6055 and 0.2731, respectively, while the Φ_ST for 16S and COI genes was 0.6251 and 0.4717 (

Table 6. Pairwise Φ_ST (below the diagonal) and F_ST (above the diagonal) values based on 16S ribosomal RNA (16S rRNA) and cytochrome c oxidase subunit I (COI) gene regions (marked in bold) for Ommastrephes caroli populations from the Adriatic, Central Western Mediterranean, and Northeast Atlantic.

Gene Region	Central Western Mediterranean	Adriatic	Northeast Atlantic
16S rRNA
Central Western Mediterranean	-	−0.1359	0.7309
Adriatic	−0.1650	-	0.6361 *
Northeast Atlantic	0.7591	0.6587 *	-
COI
Central Western Mediterranean	-	−0.0658	0.3333
Adriatic	0.0008	-	0.4062 *
Northeast Atlantic	0.6380	0.5948 *	-

* statistical significance (p < 0.05).

). Statistically significant genetic differentiation after Bonferroni correction was observed only between the Adriatic and Northeast Atlantic populations for both genes. Some separation is also observed between the Atlantic and CW Mediterranean samples; however, this was not statistically supported and warrants further investigation due to the small sample size.

4. Discussion

This is the first comprehensive study on the morphometry, age, and genetic structure of Ommastrephes caroli in the Adriatic Sea. All twenty-three investigated individuals were caught in September 2020, providing the most abundant sample of this species in the Mediterranean waters. Up to date, only three individuals were seen in the Adriatic, one juvenile in 1910 of 165 mm DML [21] and two adult females in 1986 [22] and 2013 [18] of 560 mm DML and 1300 mm in length without tentacles, respectively. Interestingly, these mature females were reported in February and March, and the juveniles in this study were caught in September. Similarly, in the Eastern Mediterranean (Aegean Sea), Lefkaditou et al. [10] reported juveniles (DML < 143 mm) mainly in August and September, while the largest specimens (560–660 mm DML) were caught from April to June, coinciding with the significant warming of the Tyrrhenian Sea surface. In the Western Mediterranean (Sardinian waters), Agus et al. [20] also recorded large mature O. caroli specimens in the spring–summer period (March–July). These observations reveal the species’ reproductive activity in the Mediterranean, with spawning occurring during the summer months, which is congruent with our aging results showing that the squids hatched during July and August. According to Lefkaditou et al. [10], the spawning ground of O. caroli could be in the warmer Eastern basin due to the more frequent presence of large moribund females in the Sea of Crete, while the findings of predominately smaller squid from the coast of western Italy indicate the colder Western basin as an important feeding ground. The O. bartramii migrations between the spawning and feeding grounds in the North Pacific have been connected with a suitable surface sea temperature (SST) and chlorophyll a concentration (chl. a) [54,55], with its distribution affected by the Kuroshio Current [27]. The sea surface temperature is significantly correlated with the monthly latitudinal gravity center of CPUE (catch per unit effort), strongly suggesting that O. bartramii distribution is controlled by an optimal thermal habitat [3]. From the above-mentioned, it can be assumed that the Adriatic Sea could also represent a feeding ground for Mediterranean O. caroli individuals.

The length–weight relationship of O. caroli juveniles inferred in this study showed an isometric growth, proportional to the weight and length, which is a general pattern for oceanic Ommastrephes species that in comparison with coastal squid, have an isometric to positively allometric growth [55,56,57]. Considering our small sample size, this should be taken cautiously, especially since short-lived squids, characterized by high plasticity in growth and maturation, can easily adapt their morphology to environmental or ecosystem change, leading to interannual differences in size and growth rates [58]. Available data regarding the O. caroli length–weight relationship are provided by Lefkaditou et al.’s [10] study on 30 mature individuals using an exponential model, which showed a better fit than the traditional power model. Of course, there is more available information on length–weight relationships and the growth of Ommastrephes species inhabiting the Pacific [54,57,59,60,61,62] and South Indian waters [56] reporting on higher growth rates in females than males and significant interannual variation in length–weight relationships from isometric to positive allometric growth. Unfortunately, due to our small sample size, we could not test any variation between the sexes or seasons; therefore, future studies with a larger number of individuals are very much needed to describe the growth pattern of Mediterranean O. caroli as this information is critical to understand the species’ life-history traits and population dynamic, and essential for assessment, management, and conservation [63].

Direct methods for assessing the age and growth of cephalopods use hard body parts, i.e., gladius, cuttlebone, beak, lens, and statolith because these structures remember ontogenetic events by forming periodic growth increments. For squid, the analysis of the statolith microstructure is the most widespread method of investigating age and growth. Size-at-age data, based on statolith readings, have been shown as a very useful tool in the identification of seasonal cohorts and growth rate estimation for commercially exploited ommastrephids [24,64,65,66,67,68]. For O. bartramii in the North Pacific waters, this method has been used to determine growth rates and lifespan [69], examine the somatic and statolith growth of wild and artificially reared paralarvae and wild juveniles of the autumn cohort [59], analyze the growth parameters of two allopatric stocks, and propose a tempo-spatial migration model [60]. By reading statolith growth increments on 237 individuals, Yatsu et al. [69] concluded a 1-year lifespan of O. bartramii reaching maturity at the age of about 7–10 months, with almost year-round spawning activity. In the present study, the estimated age of the caught O. caroli juveniles (65 to 152 mm DML) ranged from 36 to 64 days, with a mean of 47.6 days. Considering that our sample consisted of only 23 individuals, and that squids can exhibit considerable plasticity in respect to environmental conditions, our data were insufficient to develop an appropriate growth model representative of the population. Based on the beaks and eye lenses, Agus et al. [20] estimated the mean lifespan of Mediterranean O. caroli at around 12 to 13 months.

The beak represents a resistant chitinous structure in squids used for age determination, species and population identification, or investigation of trophic ecology [70,71,72]. Cephalopods are key players in oceanic food webs and understanding their trophic interactions is essential for understanding marine ecosystems. In this study, we have inferred relationships between the dorsal mantle length and various beak measures of juvenile O. caroli in the Adriatic. Four measures showed a linear relationship with the DML: LHL, UHL, ULWa, and UWL, of which the LHL and UHL seem to represent the best fit to the data. Due to the small sample size, all the tested models (linear, power, exponential, and logarithmic) showed a relatively equal fit to the data. Lefkaditou et al. [10] described the fit of the DML to LRL with the power equation, which was second-ranked in our study after the logarithmic with just a small increase in the AICc value. A larger sample size and long-term studies are needed to provide more reliable estimates, especially as it has recently been demonstrated that Pacific O. bartramii changed the body size and growth pattern of the beak shape in response to environmental change [61]. Nevertheless, the models provide a useful starting point in understanding the relationship between beak morphology and body size in the Adriatic Sea.

In line with recent molecular phylogenetic findings and the resurrection of four nominal species within the genus Ommastrephes, the Adriatic samples clustered with other COI and 16S sequences from the North Atlantic [4], the Western Mediterranean [20], and the Eastern Adriatic [18] into a separate clade specific for O. caroli, and separately from other Ommastrephes species. The species identification was further corroborated by a range of intra- and interspecific p-distances: the intraspecific distances did not exceed 1.75% for COI and 3.77% for 16S, while the interspecific distances ranged from 2–9.40% for COI and 0.29–5.22% for 16S. The barcoding gap was evident in the genetic distances for COI [73], and overlap in 16S distances, as observed by Fernández-Álvarez et al. [4]. Haplotype analyses and differentiation indexes point to the existence of genetic structuring between the Atlantic and Mediterranean/Adriatic samples, as first suggested by Agus et al. [20]. This analysis is, however, hampered by a small number of COI sequences available for O. caroli of Atlantic origin in public databases. A better sample is available for 16S, which further supports the separation between the Atlantic and Mediterranean/Adriatic basins. However, a much smaller fragment of 16S was analyzed in this study, and its taxonomic resolution for this species as well as suitability for population studies need to be further investigated [4,20]. The distributional range of the Ommastrephes species is largely shaped by oceanic gyres, currents, and surface water temperatures affecting the dispersion of paralarvae [1]. Population structuring between the Atlantic and the Mediterranean has been observed in other marine species, cephalopods included, and attributed to the existence of historical and contemporary hydrographic barriers to the gene flow: the Gibraltar Strait and the Almería–Oran front, in addition to isolation by distance [74,75,76,77]. This might also be a significant contributing factor shaping O. caroli structure. For further research, a larger number of individuals should be sampled throughout the Mediterranean and Atlantic using additional genetic markers to clarify these processes.