FIRST RESULTS ON THE DIET OF THE INVASIVE PTEROIS MILES (ACTINOPTERYGII: SCORPAENIFORMES: SCORPAENIDAE) IN THE HELLENIC WATERS

The diet composition of devil lionfish, Pterois miles (Bennet, 1828), from the Aegean Sea was investigated. The stomach contents of 42 samples (10 of which were empty) collected between May 2016 and November 2017 from Rhodes Island, Greece, were studied. The diet was composed predominantly of bony fish (78.5% in number, approximately 94.7% in biomass) belonging to a large variety of families, and invertebrates (15.4% in number, 5.1% in biomass), thus indicating an opportunistic feeding behaviour of the species in the study area. The length–length and length–weight relations were also determined.


INTRODUCTION
Alien species are considered one of the greatest threats to global biodiversity, because of their potential to directly consume or outcompete native species, alter habitats, affecting detrimentally ecosystem structure and function , Costello et al. 2010, Bianchi et al. 2012. More than 1000 alien species are listed in the Mediterranean Sea (Giakoumi et al. 2016) and the majority of them are thermophilic species introduced into the eastern Mediterranean basin through the Suez Canal (Katsanevakis et al. 2014).
The devil lionfish, Pterois miles (Bennett, 1828), and its congeneric red lionfish, Pterois volitans (Linnaeus, 1758), are considered among the most successful invasive alien species in the history of aquatic invasions (Albins andHixon 2008, Rocha et al. 2015). The devil lionfish, P. miles, is a tropical reef-associated species native to the Indian Ocean from south Africa to the Red Sea and east to Sumatra, commercially exploited for the aquarium trade, like the red lionfish, P. volitans, native to the western Pacific from southern Japan to western Australia and east to the Pitcairn Group in the south Pacific (Côté et al. 2013, Froese andPauly 2019). Outside its native range, the species is reported along the south-eastern United States coast from Florida to North Carolina, from Bermuda, Bahamas, and the Caribbean (Schofield 2009, 2010, Johnson and Swenarton 2016, sharing invaded western Atlantic regions with the sister species P. volitans. Although genetically distinct, these two species are difficult to be distinguished visually and frequently they are collectively referred to as invasive Indo-Pacific lionfish, invasive lionfish (Pterois volitans/miles) or invasive Pterois spp. (Schofield 2010, Côté et al. 2013, Johnson and Swenarton 2016, Peake et al. 2018). In the Mediterranean Sea, a single individual of P. miles was recorded in 1991 from the coast of Israel (Golani and Sonin 1992). No further observations were made until 2012, when a wave of the devil lionfish invasion started and it is currently ongoing in the eastern Mediterranean, while records are reported up to the central Mediterranean, from Tunisia and South Italy (Azzurro et al. 2017). The lionfish sightings, that followed the common invasion pattern of Lessepsian migrants, and the results of recent genetic studies, are consistent with an invasion of the Mediterranean Sea through the Suez Canal ). In the Hellenic waters, P. miles was firstly recorded from the south-eastern Aegean Sea, in 2009, at Kalymnos Island ) and later in 2015 at Rhodes and Crete islands (Crocetta et al. 2015, Dailianis et al. 2016. DOI: 10.3750/AIEP/02616 The population of P. miles is impressively growing in the Levantine Sea (Azzurro et al. 2017). Due to its abundance in fishing nets and sightings in shallow waters, the species is considered invasive in the south Aegean waters . The purpose of this study was to describe the feeding habits of this Lessepsian migrant in order to give first information on its potential impact on the native food web. Furthermore, data on the lengthweight and length-length relations of the species in the area are briefly discussed.

MATERIAL AND METHODS
A total of 42 Pterois miles specimens were randomly collected from ten sites, rocky rich of vegetation or sandy-rocky near sparse meadows, located along the east coast of Rhodes Island, from May 2016 to November 2017, at day time (Fig. 1). The majority of the samples were collected in summer 2017 (17 individuals) and autumn 2017 (12 individuals). A total of 36 samples were captured by spear gun at depths ranging from 1 to10 m, while 5 specimens caught by the boat-seining method and one by long line, were captured as deep as 50 m. Specimens were identified following Schultz (1986) and preserved frozen.
Each specimen was thawed and measured. The following parameters were determined, total length (TL), standard length (SL), both with accuracy of 0.01 cm, and wet weight/total weight (TW) with accuracy of 0.01 g. After the removal of internal organs, the net weight (NW, g) was determined. The stomach-intestine was separated and its contents removed. Prey items were counted by number and identified to the lowest taxonomical level possible, depending on the extent of digestion. Fish prey species were identified according to Bauchot (1987). When necessary, prey items were identified from otoliths found in the stomach, following Anàlisi de Formes d'Otòlits (AFORO) * . After identification, each prey item was wet weighed with an accuracy of 0.01 g and the length was measured, with 0.01 cm accuracy, by use of a calliper. Given the occurrence of one or more specific items in the stomachs (F), the frequency of occurrence (%F) of each prey taxa was estimated as the percentage of all stomachs (Hyslop 1980).
The TL values were tested for normality using the Shapiro-Wilk test (when P < 0.05 the hypothesis that data follow the normal distribution is rejected). The Pearson's correlation coefficient (r) was calculated in order to determine if the SL-TL and NW-TW relations of the lionfish samples were significantly linearly related. Furthermore, for the length-weight relations of the lionfish, regression analysis was used to test the power curve where W is TW or NW in g, L is TL or SL, in cm. The analyses were carried out using the statistical program SPSS.

RESULTS
In all 42 specimens of the devil lionfish, Pterois miles (12 females, 21 males, 9 unsexed), collected from Rhodes, the dorsal fin count was XIII spines and 10 soft rays, the anal fin count was III spines and 6 soft rays.
The TL values of lionfish specimens were normally distributed (P = 0.094) and ranged from 14.9 to 31.5 cm, the majority (38 specimens) from 15.0 to 25.0 cm, while the TW ranged from 38.2 to 352.4 g ( Table 1). The full stomach weight ranged from 1.1 to 11.7 g, while the empty stomach weight ranged from 0.7 to 8.7 g.
Both the relations between TL and SL and between TW and NW of all the lionfish samples resulted significantly linearly related (P < 0.001 in both cases) ( Table 2). The power curve was a good fit for the length-weight values of all lionfish specimens (P < 0.001), approaching an isometric growth pattern, both for TW-TL (b = 2.894) and TW-SL (b = 3.068) relations (Table 2).
Among the 42 red lionfish specimens examined, 10 had empty stomachs. A total of 65 items was found in the remaining 32 stomachs. The majority were bony fishes (51 items, 78.5%), partially or well digested, and this rendered difficult their identification and approximate their measurements. Only ten specimens were identified to species, of which five from otoliths. Ten invertebrate items were also detected, 15.4% of total number of prey items, including beaks of Sepiidae, pieces of shrimp exoskeletons, and also the isopod Nerocila orbignyi, a common ectoparasite on many of the prey fishes identified (Table 3). Finally, pieces of phytobenthic taxa were observed. Fish identified to family, genus, or species were 72.5% of total number of fish prey items, and they belonged to 13 families. The best represented family was Gobiidae, followed by Tripterygiidae, Sparidae, and Labridae and many of the prey items identified are common inhabitants of rocky bottoms or dwell near rocks on sand and seagrasses beds ( Table 3). The length of fish preys was measured in 39 items obtained from the stomach of 25 lionfish (Table 4). Fish prey length ranged from 8.9 mm to 70 mm and it appeared that both small and larger fish were consumed by lionfish of different sizes (Tables 3, 4). The TW of food items was approximate, due to the advanced state of digestion of many fish and of the few decapods found. Also, the weight of otoliths found in six cases (total 0.08 g) was not considered. The TW of the remaining 45 fish items was 49.9 g, and contributed approximately to the 94.7% of total food biomass. The higher contribution to the stomach content in weight was observed for Apogonidae, Pomacentridae, and Gobiidae. The weight of invertebrates and phytobenthos was negligible, compared to fish. Among the identified fishes, Gobiidae showed the higher frequency of occurrence in the stomachs (14.3%F), followed by Sparidae (9.5%F) and Labridae, Pomacentridae, and Tripterygiidae (7.1%F) ( Table 3). Bony fishes were present in 73.8% of total number of stomachs and in about all the stomachs containing food (96.9%F).

DISCUSSION
The devil lionfish, Pterois miles, inhabits rocks and coral in its original habitat down to the depth of 50 m, reaching commonly a size between 15 cm and 35 cm (maximum 50 cm) (Golani et al. 2006, Darling et al. 2011, Froese and Pauly 2019. Rocky habitat, range of length and depths of capture of the presently reported study agree with the above-mentioned studies, while mean length and weight obtained appear lower than the majority of those reported in the west Atlantic areas invaded by lionfish (Sabido-Itzá et al. 2016).
The length-weight relation is important in fisheries biology and highly useful in fishery assessment studies (Stergiou and Moutopoulos 2001). In the case of established invasive species, it is essential to document the lengthweight relations since they provide information on the Both native and invasive lionfish are primarily crepuscular hunters, especially at dawn or at both dawn and dusk (Fishelson 1975, Jud and Layman 2012, Benkwitt 2016. In the native range juvenile lionfish live in small groups, but as adults they typically occur alone (Fishelson 1975). Recent studies in invaded regions revealed that older lionfish also frequently form groups and furthermore they hunt during the day. Grouping during movement as well as resting in refuges could provide a better defence from predators, while hunting in groups may improve lionfish catch efficiency and increase their feeding success Small invasive lionfish show high site fidelity (Jud and Layman 2012), while larger individuals are able to travel from their shelter to surrounding habitats (Benkwitt 2016). High site fidelity may be due to several factors. For example, invaders may not be recognized as predators by native preys, consequently they do not need to move from a specific location searching for food, or lionfish are not recognized as prey by native predators, consequently  they do not need to move to escape from them (Jud and Layman 2012). Also, lionfish of small size feed primarily on invertebrates and they do not need to forage far from their site, while they feed mainly on fish when they reach approximately 20 cm TL (Benkwitt 2016).
The majority of the lionfish examined in the presently reported study were caught by spear gun on rocky substrate. Specimens observed and photographed by local divers at Rhodes during the daytime, were solitary and relatively motionless, typically on rocks, sheltering in reef crevices, more often with a head-down posture with open pectoral fins, but also in a sideways or head-up posture and furthermore in horizontal position on sand near rocks. The head-down position is one of the lionfish postures   associated mainly with hunting, but also the horizontal and head-up postures are used to catch prey Maljković 2010, García-Rivas et al. 2018).
Lionfish are considered opportunistic generalist carnivorous with a broad dietary niche including fishes and invertebrates in the western Atlantic Maljković 2010, Peake et al. 2018). This is perhaps one of the reasons they have been such successful invaders in the Caribbean region, combined with other factors such as the long venomous spines that deter predation, their rapid growth, the early maturation and year-round reproduction, the dispersal over long distance during egg and larval stages, the variety of predatory strategies and ecological versatility, as well as prey naivety, weak competitors, and native predators that are overfished and may not recognize lionfish as prey (Côté et al. 2013, Johnson and Swenarton 2016, García-Rivas et al. 2018). In the original habitat, such as the Red Sea and the Gulf of Aqaba, lionfish feed up to a big variety of benthic fishes and also shrimps and crabs (Fishelson 1975, Khalaf andDisi 1997). The results of the presently reported study support the assumption that lionfish in their new Mediterranean environment are not changing their feeding habits, compared both to their native and introduced western Atlantic range. In fact, they are carnivorous and, for the sizes examined in this study (15-31 cm in TL), they were predominantly piscivorous (Benkwitt 2016). The diet was in fact composed by a large variety of small fish species, not only inhabiting rocky bottoms, in caves and crevices, but also dwelling nearby rocks, on seagrasses meadows and also on sandy bottom. These results may indicate that the lionfish are able to forage moving from the rocky habitats where they are usually observed, to nearby different habitats.
In conclusion, lionfish are generalist opportunistic ambush feeders, which makes them especially successful predators and can confer high ecological impact on native Mediterranean communities (Galanidi et al. 2018), as dramatically ascertained in the invaded Caribbean ecosystem (Albins andHixon 2008, Côté et al. 2013).
In their initial stage of invasion, lionfish probably share habitats and food with native Scorpaenidae, such as Scorpaena scrofa Linnaeus, 1758 and Scorpaena porcus Linnaeus, 1758, which are common in areas similar to those occupied by the lionfish and feed on fish (gobies, blennids) and benthic invertebrates (Bauchot 1987, Golani et al. 2006. The today's knowledge on the diet of P. miles, a newcomer to the Mediterranean Sea, is restricted (Özgür Özbek et al. 2017) and the results of this work add information on this biological aspect that may be useful to evaluate its potential impact on the native biodiversity and food web composition, along which the lionfish appears to be already well integrated.

AKNOWLEDGEMENTS
The first author would like to thank Dr Andreas Sioulas, Director of the Hydrobiological Station of Rhodes-Hellenic Centre for Marine Research for giving the opportunity to perform at the Hydrobiological Station of Rhodes the laboratory work related to her Marine Environmental Science thesis. She is also grateful to Dr Ernesto Azzurro and Dr Emanuela Fanelli for providing unique support for species identification from otoliths during her Erasmus placement at the CNR-ISMAR Laboratories of Lerici, Italy. Authors are grateful to Gerasimos Kondylatos for technical assistance during the laboratory work and to Konstantinos Perdikaris for providing support in data elaboration. Furthermore, they warmly thank all the professional and amateur fishers for providing the lionfish specimens studied in this work.