SPATIO-TEMPORAL VARIATION IN THE DIET COMPOSITION OF RED LIONFISH, PTEROIS VOLITANS (ACTINOPTERYGII: SCORPAENIFORMES: SCORPAENIDAE), IN THE MEXICAN CARIBBEAN: INSIGHTS INTO THE ECOLOGICAL EFFECT OF THE ALIEN INVASION

Arredondo-Chávez A.T., Sánchez-Jimenez J.A., Ávila-Morales O.G., Torres-Chávez P., Herrerias-Diego I., Medina-Nava M., Madrigal-Guridi X., Campos-Mendoza A., Domínguez-Domínguez O., CaballeroVázquez J.A. 2016. Spatio-temporal variation in the diet composition of red lionfi sh, Pterois volitans (Actinopterygii: Scorpaeniformes: Scorpaenidae), in the Mexican Caribbean: Insights into the ecological effect of the alien invasion. Acta Ichthyol. Piscat. 46 (3): 185–200.


INTRODUCTION
). This invasion is one of the main threats to the Caribbean coral reef ecosystems (Morris 2012). The fi rst sighting of lionfi sh in the Mexican Caribbean was reported in Cozumel Island in 2009 (Schofi eld 2009). Since that time, lionfi sh rapidly invaded the entire marine and some estuarine systems in the area (Vásquez-Yeomans et al. 2011), including deep (>75 m) waters.
Invasive species are able to produce what is called a cascade effect through several trophic levels. For example, when an "alien and exotic" fi sh enters an ecosystem and preys on the native predator, the loss of this predator often results in an increase in the abundance of its prey. If the proliferation of the latter has a negative impact on the ecosystem, then this outcome becomes an indirect consequence the invasive species occurrence in this ecosystem. Basically, invasive species causes a chain reaction in which each trophic niche in the ecosystem is affected (Green et al. 2012). Although the negative effects of the lionfi sh in the Caribbean ecosystems seem to be obvious, the majority of the studies have focused in areas such as the Bahamas and Florida and little is known about the basic aspects of the invasion in the Mexican Caribbean. Moreover, most of the research work related to the lionfi sh diet composition focused in the fi sh species, but little is known about other groups that also serve as food for the lionfi sh. Therefore, the main aims of the presently reported study were: • To identify the diet composition of the lionfi sh along the Mexican Caribbean, and • To determine the spatio-temporal variation of their diet in different sites along the Mexican Caribbean.

MATERIALS AND METHODS
Sampling. The Caribbean region features some 26 000 km 2 of coral reefs distributed mostly in shallow areas with mean depths of 20 m. From these, the Mexican Caribbean has a coral reef extension that covers an area of approximately 650 km 2 , constituting part of the Mesoamerican Barrier Reef System (MBRS). In this work, the study site was divided in three zones identifi ed as (10 km scale): • North zone; • Central zone; • South zone. For fi sh collection, zone was subdivided into localities (scale of 1 to10 km); one locality from the North zone (Isla Contoy), three locations for the Central zone (Xpu-Ha, Akumal and X'Cacel), and two locations for the South zone (Banco Chinchorro and Xcalak). Finally, at each location different collection sites were established (scale of 0.1 to 1 km) (Fig.1). Of the six localities and 54 collection sites; four were located inside Natural Protected Areas, that are identifi ed as priorities by the National Commission for the Knowledge and Use of Biodiversity (CONABIO). These areas are known as the National Park of Isla Contoy, Sea Turtle Sanctuary X'cacel-X'cacelito, National Park Arrecifes of Xcalak, and Biosphere Reserve of Banco Chinchorro. The above mentioned sites plus Xpu-ha and Akumal are all locations which support a high touristic activity and are part of the well known "Riviera Maya".
Surveys of red lionfi sh, Pterois volitans, were conducted during the dry-(February-May), rainy-(June-September) and nortes-* seasons (October-January), in 2011 and 2012. The fi sh were collected using SCUBA diving and a Hawaiian harpoon, and the survey covered different habitats (coral reefs, rocky reef, coral patch, Thalassia fi eld, gorgonids fi eld, and laja bottom). Collected specimens were frozen, packed in labelled nylon bags, and transported to the laboratory where they were taxonomically identifi ed following Schultz (1986). Specimens were measured for total length (TL) and standard length (SL) to the nearest centimetre and weighed to the nearest gram. Sample processing. Prey items present in the stomach and part of the intestine were recovered from each collected lionfi sh and were then identifi ed using a dissection microscope and taxonomic keys (Abele and Kim 1986, McEachran and Fechhelm 1998, 2005, Carpenter 2002, Humann and DeLoach 2002, Nizinski 2003, Robertson and Van Tassell 2015. After identifi cation, prey items were separated, counted, and individually weighted. In some cases, the level of digestion of prey did not allow the separation of the samples in groups, therefore, such samples were classifi ed as unidentifi ed prey. The classifi cation used here for stomach contents (full, half full, half empty, and empty) was done by volume, therefore, half empty refers to the apparent volume observed between half full and half empty stomachs. Data analyses. A sample base rarefaction curve with Estimates (Version 9.1.0.) was made using 500 sample randomization without replacement (Coldwell 2013). To identify if our sample effort fully characterized the lionfi sh diets, richness was assessed by using 2 nonparametric estimators: Chao2 and Mao Tau. In order to separate the fi sh collected by size the standard length of all the surveyed fi shes and the percentile were used in JMP version 6.0 ** . Then, the 11 size classes obtained, were assigned to three sizes: small (6-15 cm), medium (15-25 cm), and large (25-39 cm). The repletion index was obtained for all the samples to determine the amount of food found in each stomach (Hyslop 1980). The percentage by number (%N) and the percentage frequency of occurrence (%F) of each prey was calculated following Hyslop (1980). The percentage by weight (%W) of each prey was calculated following Pinkas et al. (1971). The data obtained was used to calculate the relative importance index (%IRI) following Stevens et al. 1982. The niche amplitude was obtained using the Levin standardized index (Krebs 1999).
Due to the differences in the sample size per habitat, the diet composition by habitat could not be compared, therefore, sample localities were separated in North-, Central-, and South zones (Fig. 1). The diet composition was compared between seasons using the Simpson Index, the Shannon Index, and the Student' t-test. The Sørensen similarity index, was obtained by a simple league cluster analysis (Moreno 2001). Data were processed using the Bio-DAP software (Gordon and Douglas 1988).

RESULTS
Lionfi sh diet composition by size. A total of 1482 specimens of Pterois volitans were collected. Standard length (SL) of the collected fi sh was between 60 mm and 390 mm with a mean value (±SD) of 194 ± 56 mm. The fi sh weight was between 9.8 g and 997.3 g with a mean value (±ED) of 256 ± 200 g. Out of the total number of the fi sh collected, 962 specimens (64.9%) were found with prey in their stomachs. The total number of prey items was 1609. Specimens were collected in different habitats, including high profi le coral reefs (72.13%), rocky reefs (9.31%), coral patch (6.61%), laja bottom (4.99%), gorgonian fi elds (3.98%), and Thalassia fi eld (2.98%).
The rarefaction analyses indicated that sample size was not suffi cient to reach the asymptote, indicating that more sampling would be necessary to fully represent the lionfi sh diet composition. The 2.23% of the stomach analysed were categorized by volume as full, 12.21% as half full, 50.47% as half empty, and 35.09% as empty. Out of the total number of the prey items recovered from the fi sh studied only 76 were identifi ed to the species level. In this number there were 47 fi shes and 29 crustaceans. A total of 57 prey items were identifi ed to the genus level, including 32 fi shes, 22 crustaceans, and 3 molluscs. As many as 38 preys were assigned only to the family level, with 18 families representing fi shes, 17-crustaceans, and 3molluscs (Table 1). Eleven size classes of fi shes showed an ontogenetic variation in the groups consumed, where small lionfi sh specimens consumed more crustaceans, and larger lionfi sh consumed more fi shes (Fig. 2) (Table 1). As clearly shown in Fig. 2, the diet composition of small size lionfi sh is based mainly on crustaceans, and the diet composition of the large size lionfi sh is based mainly on fi sh.       Fig. 3). Whereas the fi sh families with more genera represented in the diet of lionfi sh were: Serranidae (4), Pomacentridae (3), Labridae (3), Scaridae (3), Labrisomidae (3), and Gobiidae (3). The families with more species represented as part of the diet components were: Pomacentridae (7), Scaridae (6), Serranidae (6), Gobiidae (5), and Labrisomidae (4). For the crustaceans the families with more genera were: Penaeidae (5), Palaemonidae (3), Gonodactylidae (2), and Portunidae (2). The families with more species were: Palaemonidae (7) and Penaeidae (6) The lionfi sh predation on teleost fi sh was found to be greater as compared to the predation of crustaceans and molluscs. Therefore, from these results it is clear that the lionfi sh presence in the Mexican Caribbean is affecting mainly the local population of reef fi sh. Diet composition by season. A total of 201 lionfi sh specimens were collected during the dry season. A minimum length of 60 mm SL and maximum of 300 mm SL; and minimum weight of 10 g and maximum of 835 g were found. Eighteen families, 26 genera (one mollusc, eight crustaceans, and 17 fi shes), and 24 species were identifi ed, of which 20 were fi shes and four were crustaceans.
During the rainy season a total of 286 lionfi sh specimens were collected. Their length (SL) ranged from 60 mm through 290 mm, while their weight was within 9.8-642 g. Twenty-four families, 38 genera (two molluscs, 17 crustaceans, and 19 fi shes), and 36 species (19 fi shes and 17 crustaceans) were found.
During the nortes season a total of 995 lionfi sh specimens were collected, with a standard length ranging from 60 to 390 mm, and the weight-from 10 to 977 g. Thirty three families, 48 genera (28 fi shes and 20 crustaceans), and 57 species were identifi ed (38 fi shes and 19 crustaceans) ( Table 2).
Results showed that there is an evident variation in the composition of the diet which seems to be associated with the climatic season. Even though that the numbers of lionfi sh collected during the three seasons were not the same, it is evident that the most abundant prey, in spite of the season, were the fi shes. The presently reported study demonstrated that the increase in the number of lionfi sh captured was directly associated with the number of families, genera, and species found in their diet. It should also be emphasized that the composition of prey species in general, was different in each season (Figs. 6-8). Overall, species of the genus Cinetorhynchus seems to be the most available and preferred food component of the lionfi sh diet in the three seasons. However, the total number of consumed fi sh was higher and the diversity of fi sh prey was higher compared to the diversity of crustacean preys. It is important to mention that the sampling intensity was partially lower during the dry and rainy season, due to a function of time/visibility during these seasons, which limited the total catch by period. Diversity measurements. The Shannon diversity index (H′) (bits · individual -1 ) showed that the highest diversity of the diet composition was found during the nortes season (H′ = 2.66 bits · ind -1 ), followed by the dry season (H′ = 2.38), and the rainy season (H′ = 2.37). The diet composition during the dry season showed high evenness with a value of 0.73, and the rainy season showed the high dominance with a value of 0.65. The Sørensen test showed that the most similar values found in the rainy season were diversity and abundance (0.558 and 0.379 qualitative and quantitative, respectively). The less similarity was found during the dry season and the nortes season (0. 486 and 0.248 qualitative and quantitative, respectively) ( Table 3).
The highest diversity of the diet composition was found during the nortes season, and the lowest value during the rainy season. The lowest value of species similitude was found also in the rainy season, being understandable to fi nd the value of higher dominance over the same period.  F = number of prey species representing fi shes, C = number of prey species representing crustaceans; Nortes season = October-January.

DISCUSSION
Red lionfi sh, Pterois volitans, demonstrates a high competitive capacity, high reproduction, and high growth rates (Albins and Hixon 2013), all of which makes this species as one of the most effi cient predators and a dangerous invasive species that could, in a very short time, affect the ecology relations of coral reefs in the Mexican Caribbean, as well as in other parts of the world. The invasion in the Mexican Caribbean is relatively recent, the fi rst lionfi sh sighting was reported in 2009 in Cozumel island (Schofi eld 2009), and although in other areas such as Florida and Bahamas this species was systematically reported as early as 1992 (Albins and Lyons 2012), few studies have been done to understand the impact of the lionfi sh in the ecology and local populations, by analysing its diet composition in natural and invaded areas.
Although Hamner et al. (2007) showed that two species of Pterois are present in the Atlantic, P. volitans represented around 93% of the population, and P. miles only 7%. In the presently reported study the 1482 specimens collected were identifi ed as P. volitans, based in traditional measurements and identifi cation analyses, however, 57 specimens were classifi ed as taxonomically uncertain. Overall results presented in this paper are consistent with the results reported by Valdez- Moreno et al. (2012), where they used a barcode molecular method for identifi cation of lionfi sh in the Caribbean and found only the species P. volitans. This is also supported by unpublished data obtained from our collaborators, that used DNA barcode (COI sequences) and obtained at least 15 sequences from different specimens that shared a 99% of nucleotide identity with the sequence published at GenBank for P. volitans (Hernandez-Zepeda et al. unpublished data). Therefore, to date the only species of lionfi sh identifi ed in the Mexican Caribbean is P. volitans.
Lionfi sh is characterized by a slow movement, a camoufl aged coloration, and elongated fi n ray projections that results in a low detectability by predators (Albins and Hixon 2008). As a consequence of these features, lionfi sh may be escaping from signifi cant top-down control predators, with only few occasional predators (Pimiento et al. 2013). Another possible reason for the lionfi sh success is their effi cient reproduction and the movement of the larvae with the currents where there are not natural predators . It has been proposed that lionfi sh preferentially (but not exclusively) settle in shallow habitats before moving to deep reefs when they reach larger sizes (Claydon et al. 2012). This pattern is often a consequence of fi sh ontogeny (Mumby et al. 2011). In the presently reported work, results showed that lionfi sh is more abundant in areas resembling cracks and caves of shallow reef areas (2 to 35 m).
The cumulative curve for the 1482 stomachs with prey analysed indicated that the sample size was insuffi cient to reach asymptote, therefore more sample effort is required to be able to fully describe the lionfi sh diet. However, it is important to notice that this work represents the highest collection effort compared to other similar studies conducted in the area. This could be a consequence of the opportunistic feeding behaviour, since lionfi sh can eat at almost any species that it can gulp (Mumby et al. 2006(Mumby et al. , 2011 and, since in this work specimens were collected in six different habitats, the variety of prey that the lionfi sh may consume increased. The lionfi sh diet in the Atlantic is composed by fi shes and crustaceans as the most representative groups, and accidentally some molluscs (due to the low frequency found in the stomachs revised) , Arias-González et al. 2011, Muñoz et al. 2011, Valdez-Moreno et al. 2012. In the presently reported study the largest number of prey items in the lionfi sh diet were fi sh and crustacean (76 species), compared with 43 species reported by , 18 species in Muñoz et al. (2011), 34 in Valdez- Moreno et al. (2012), and 42 species in Green et al. (2012). The differences observed in the lionfi sh diet composition could be explained because in previous published studies, few stomachs were analysed , and other studies used different techniques to identify preys such as molecular tools (Valdez-Moreno et al. 2012). Also, the number of crustaceans previously reported as lionfi sh diet components was low, compared with the 29 species identifi ed in the present study. The total of 48 species of fi sh found as main components of lionfi sh diet in this study can be considered higher than those obtained in studies that included a larger sampling area , where a total of 41 fi sh species were reported. In this work an ontogenetic lionfi sh diet composition variation was found, with crustaceans as the more abundant/important prey for small specimens, whereas as lionfi sh grows the more abundant/important prey were found to be fi sh. These results are similar to other studies that also reported a relation between sizediet in lionfi sh (Cure et al. 2012). Many reef fi sh species use seagrass and mangrove as juvenile habitat (Mumby et al. 2006). Lionfi sh in a juvenile nursery may reduce the recruitment pool available to colonize reefs through predation or competition (Barbour et al. 2010) acting in concert with lionfi sh predation on coral reefs (Albins and Hixon 2008, Barbour et al. 2010) to further stress reef fi sh populations. Additionally, lionfi sh may differentially use habitats throughout their ontogeny. Lionfi sh in mangrove habitat, for example, may be smaller than in reef habitat (Barbour et al. 2010) suggesting mangroves may function as lionfi sh nurseries.
The large array of prey consumed indicated that lionfi sh is a top predator, as the same level as sharks and groupers (Arias-González et al. 2011). All previous studies remark the negative impacts of lionfi sh in native fi sh groups, as a predator or competitor with the native fauna Lyons 2012, Albins andHixon 2013). In this work, as in most of the previous reports, the most important fi sh species that are components of the lionfi sh diet are the families Labridae, Pomacentridae, Gobiidae, Serranidae, and Scaridae. Also, the parrotfi shes (Scaridae) were one of the most important families in the lionfi sh diet; this group of fi shes are mainly herbivorous, and feed on algae that grow in the coral reefs, therefore, they have a very important ecological role maintaining the algae population, acting as a "gardeners" of the reef, preventing the invasion of algae and the subsequent coral damage. The ecological role of parrotfi sh is more relevant in areas where there is an increase in nutrients in the water as a result of human pollution, which can be the case for the Mexican Caribbean, where one of the main economic activities is tourism. Results presented in this paper suggest that lionfi sh predation on parrotfi sh could decrease their numbers in coral reefs, resulting in the subsequent affectation to the coral reef community health, however, more studies regarding fi sh abundance are necessary to corroborate this hypothesis. As an example striped parrotfi sh, Scarus iseri (Bloch, 1789), which was found as the most important parrot fi sh species in the lionfi sh diet, is recognized as one of the most important species associated to the health of coral reefs, because they have the highest consumption rate of algae, making more important their presence in disturbed areas, were the algae production increases considerably (Mumby et al. 2006 Durán andClaro 2009). Muñoz et al. (2011) discussed that although the pomacentrids are the most abundant prey in the environment, this group show a low importance in the diet of lionfi sh in locations of Carolina, USA, explaining that this results were related with the high substrate association in the Pomacentridae, making this group less vulnerable to predation. Also, in the Bahamas ) and Florida (Jud et al. 2011) Pomacentridae does not appear to be an important item. In contrast, the results presented in our paper showed the Pomacentridae as the most important (IRI) fi sh family in the lionfi sh diet. This can be explained by a lionfi sh "learning behaviour" where they adapted to eat new available/abundant preys. This possibility is based in the well investigated learning process that territorial fi sh showed in other areas, for example pomacentrids, can decide if to attack or avoid an invader according with the level of the threat that the new invader possess (Helfman andWinkelman 1997, McCormick and. The predation of all grouper species (Family Serranidae) found in this work, suggests that lionfi sh might decrease the recruitment of economically important species affecting the already stressed fi sheries in the area. For example "cherna", Cephalopholis cruentata (Lacepède, 1802), was found in this work as component of the lionfi sh diet; cherna has a high economic value due to the quality of its meat. Also chernas feed mainly on Chromis multilineata (Guichenot, 1853) (Family Pomacentridae), which is a species with an important ecological role in the area, and is also a recurrent food for lionfi sh. This work showed that economically important crustaceans (being a possible competitor) such as shrimps and lobsters and other groups (molluscs) are present in the diet of lionfi sh, highlighting a nested effect that directly or indirectly affect species and regional biodiversity.
The data analysis conducted by seasons, determined that during the dry period crustaceans of the genus Cinetorhynchus were abundant preys where Cinetorhynchus rigens was the more abundant prey. This species is widespread in the region, it has nocturnal habits and its usually located in the vicinity of the cracks and caves that serve as shelter during the day. It is well known that lionfi sh also prefers to refugee in cracks and caves where there is a greater chance of fi nding crustaceans for feeding. Another important genus found as component of the lionfi sh diet was Pleoticus. The highest record of this crustacean as part of the diet in the period coincided with the peak of the crustacean reproduction (Fernández et al. 2012). For the rainy season, again the most important prey was Cinetorhynchus, followed by the species of the genus Periclimenes. During the nortes season also the genus Cinetorhynchus was the most representative component of the diet, followed by genus Stegastes (Pomacentridae), with Stegastes partitus (Poey, 1868) as the most frequent species.
Previous studies reported the diet composition of the red lionfi sh, Pterois volitans, in the Caribbean, demonstrating that this invader ate mainly fi shes, which represented between 78% and 99% of its food volume. The most representative prey groups were the fi sh families: Gobiidae, Labridae, Grammatidae, Apogonidae, and Pomacentridae followed by a low number of crustaceans It is evident from all these previous works, that the crustaceans did not represent an important component of the lionfi sh diet. Some authors did not even consider them as part of the lionfi sh diet. Therefore, this work is the fi rst to acknowledge the importance of crustacean and molluscs in the diet composition of lionfi sh.
The Mexican Caribbean is part of the second largest coral reef barrier in the world, and supports the most important touristic area in Mexico, where this is the main economic activity that relies on a very signifi cant demand of coastal and marine resources, including fi shes and crustaceans. These resources are highly vulnerable to natural and anthropogenic changes, such as habitat destruction, pollution, overfi shing, and introduction of non-native species (Albins and Hixon 2008). Therefore, it is very important to monitor and generate new data related to the effect of the lionfi sh invasions in the area. Results from this work and many others have pointed out the lionfi sh as an economic risk because its diet habits not only include juveniles of commercially important species such as lobsters, but because it also competes with snappers (Lutjanidae) and grouper (Serranidae) for food and habitat. This factor is aggravated by the strong fi sheries exploitation that exists for these and other groups of marine organisms. It is a threat to the tourism industry that revolves around the reef.