Introduction

Barracudas (Sphyraenidae) are carnivorous predators that affect the behavior and recruitment of the species they prey on and, as such, function as regulators of fish communities in coastal areas (^{Barreiros et al. 2002}, ^{Hooker et al. 2007}, ^{Mohammadizadeh et al. 2010}). There are 27 barracuda species worldwide, all belonging to one genus, Sphyraena (^{Nelson et al. 2016}). There are 5 barracuda species in the Gulf of California: the Pacific barracuda (Sphyraena argentea Girard, 1854), the Mexican barracuda (Sphyraena ensis Jordan and Gilbert, 1882), the pelican barracuda (Sphyraena idiastes Heller and Snodgrass, 1903), the Cortés barracuda (Sphyraena lucasana Gill, 1863), and the blackfin barracuda (Sphyraena qenie Klunzinger, 1870) (^{Robertson and Allen 2015}). The most abundant barracuda species in the southeastern end of the Gulf of California is the Mexican barracuda, S. ensis, a pelagic/neritic species that forms schools comprising several hundred individuals, although solitary individuals can also be found (^{Sommer 1995}). Barracudas inhabit waters between 0 and 25 m deep, sandy, muddy bottoms, and rocky and coralline reefs near the coast (^{Sommer 1995}, ^{Robertson and Allen 2015}).

The Mexican barracuda sustains a permanent fishery that is only interrupted during strong storms or hurricanes. This species is captured with several types of fishing gear (trolling lines, seine nets, gill nets, and handlines). Given the high quality of the flesh and low price, demand for this species is high in regional markets (^{Espino-Barr et al. 2003}, ^{Ulloa-Ramírez et al. 2008}). Despite the ecological and economic importance of S. ensis, there is no published information on its basic ecology. Research on this species could include studies on food intake, diet composition, diet breadth, diet overlap, trophic levels, energetic physiology, and ecosystem-wide management (^{Pauly and Christensen 2000}, ^{Abitía-Cárdenas et al. 2002}). All of these aspects can influence ecological characteristics such as growth, reproduction, and mortality of marine fish (^{Wootton 1998}).

According to ^{Zavala-Leal et al. (2018)}, previous biological studies on barracudas worldwide have mostly focused on feeding habits. Studies have included the trophic aspects of Sphyraena viridensis in the northeastern Atlantic Ocean (^{Barreiros et al. 2002}), Sphyraena barracuda off Colombia (^{Hooker et al. 2007}), Sphyraena putnamae in the Persian Gulf (^{Mohammadizadeh et al. 2010}), and S. viridensis, Sphyraena sphyraena, and Sphyraena cryzotaenia in the Indo-Pacific Ocean (^{Kalogirou et al. 2012}), among others. In Mexico, no studies on the trophic biology of the Mexican barracuda have been carried out. The only previous study was that of ^{López-Peralta and Arcila (2002)}, who analyzed the stomach contents of 5 Mexican barracuda specimens caught in Colombian waters. These authors reported that Mexican barracuda fed on mainly nekton (fishes) and, in smaller portions, on zoobenthos (crustaceans).

In this context, we evaluated the dietary spectrum of the Mexican barracuda to determine possible diet variations with sex, size, and/or season. We provide basic information necessary to characterize the feeding interactions and, by extension, the role of this species in the energy flow of trophic webs in the coastal ecosystems of the southeastern end of the Gulf of California.

Materials and methods

Mexican barracuda specimens were obtained from the artisanal fishery in the San Blas region, Nayarit, Mexico (21º29′42.07″-21º26′47.47″N; 105º17′27.30″-105º13′43.88″W) (Fig. 1). Fish were caught at night (19:00 to 06:00) using 5- to 20-m-long weighted handlines and J-type hooks (numbers 6 and 7). It should be noted that S. ensis can be caught using other types of fishing gear; however, off the coast of Nayarit, barracudas are caught mainly with handlines at night and in the early hours of the morning. According to artisanal fishermen, barracudas are not available during the day.

Figure 1 Map showing the Mexican barracuda, Sphyraena ensis, sampling area (shaded area) off San Blas, Nayarit, Mexico. 

The habitat where organisms were captured is characterized by soft, sandy, muddy bottoms, with depths of 8.5-15.0 m. Sampling was carried out over an annual cycle (February 2014 to January 2015), catching approximately 30 fish per month. Specimens were frozen and transported to the Laboratorio de Ecología Trófica at the Escuela Nacional de Ingeniería Pesquera (Nayarit, Mexico), where biometric measurements (total length and weight with precision of 1 mm and 0.1 g, respectively) were taken and stomachs collected.

Stomach contents were separated by taxonomic group, and each item was identified to the lowest taxonomic level possible. Depending on the digestion state of prey, fish taxonomic identification was carried out using the keys by ^{Clothier (1950)}, ^{Miller and Jorgensen (1973)}, and ^{Fischer et al. (1995a}, ^b). Crustaceans and mollusks were identified using the keys by ^{Brusca (1980)} and ^{Morris et al. (1980)}.

A species accumulation curve was created with the program EstimateS Swin820 (^{Colwell 2009}) using the Shannon-Wiener diversity index (H′) value obtained for each stomach to determine the representativeness of the number of analyzed stomachs. The coefficient of variation (CV) was calculated to obtain a quantitative estimate of the number of stomachs representative of the diet. If the CV was equal to or less than 5% (0.05), the number of examined stomachs was considered adequate to represent the diet (^{Jiménez-Valverde and Hortal 2003}, ^{Moreno-Sánchez et al. 2015}).

The frequency of occurrence (%FO), numerical (%N), and gravimetric (%W) percentages were calculated (^{Hyslop 1980}) and used to estimate the index of relative importance (IRI) proposed by ^{Pinkas et al. (1971)} and modified by ^{Hacunda (1981)}: IRI = (%N + %W) × %FO. To facilitate comparisons with previous studies, the standardized version of the index (%IRI) was also used (^{Cortés 1997}): %IRIi=100IRIi∑i=1 IRIIn.

In addition to the characterization of the general trophic spectrum, diets by sex (male or female), size, and season were also determined. Because size at maturity for the Mexican barracuda has not yet been determined and with the purpose of detecting a possible change in food intake (biomass) with size of organisms (total length), a cluster analysis was performed using the program Primer v.6.0. The results obtained when we “let the data speak” consisted of a series of groups without a defined pattern. These groups comprised different size records with no logical trend; that is, there was no increase in prey biomass intake with the increase in predator size. Considering the above and to simplify the ontogenetic analysis, 3 groups based on total length were created (Sturges’ rule, ^{Daniel 1997}): small (30-39 cm), medium (40-49 cm), and large (50-60 cm). In the case of the seasonal analysis, monthly precipitation over the study area was taken into account, and 2 seasons were identified: a rainy season from June to October and a dry season from November to May (^{CONAGUA 2013}).

The breadth of the Mexican barracuda trophic spectrum was calculated with the absolute values of the numerical index using Levins’ standardized index of niche breadth (B _i ) (^{Hurlbert 1978}). Values for this index range from 0 to 1; values <0.6 indicate a diet that includes few prey items and corresponds to a specialist predator, and values >0.6 indicate that the diet includes several prey items and corresponds to a generalist predator (^{Krebs 1999}). This index was calculated as Bi= 1n-11∑jPij2-1, where ∑jPij2 is the proportion of the jth prey species in the diet of predator i, and n is the total number of prey species.

A scatter plot was constructed using the graphical method of ^{Costello (1990)} as modified by ^{Amundsen et al. (1996)} to interpret the Mexican barracuda’s feeding strategy. These authors distinguished 4 strategies: (1) specialization on different types of prey, (2) a more generalized diet with some individual variation in diet breadth, (3) specialization on one type of prey while occasionally consuming other species, and (4) a mixed foraging strategy in which some individuals consume a specialized diet and others employ a more generalized foraging strategy. The Costello method was used as a complementary technique to corroborate trophic niche breadth because it allows identifying patterns of population specialization or individual diet, and testing for significant differences in the diet.

In order to assess potential differences in the diet of the barracuda as a function of sex (male or female), size (small, medium, or large), season (dry or rainy), and possible interactions between factors, we applied a permutation analysis of variance (PERMANOVA, 1,000 permutations) to the obtained numerical data (N). The dissimilarity matrix used for the PERMANOVA test was constructed by calculating the Bray-Curtis dissimilarity distances matrix using the original data (no data transformation was used). Because some likelihood-based tests of differences in multivariate data can be sensitive to the heterogeneity of multivariate dispersion, we performed a PERMDISP analysis (^{Anderson 2006}) to assess potential differences in dispersion among factors (sex, size, and season); PERMDISP can be used as a multivariate analogue of Levene’s test for homogeneity of variances (^{Anderson 2006}, ^{Oksanen et al. 2016}). We used the betadisp and adonis functions for the PERMDISP and PERMANOVA tests, respectively, using a significance level of 95%; both functions belong to the Vegan package (^{v.2.4-1, Oksanen et al. 2016}) in the R environment (^{R Core Team 2016}).

The equation proposed by ^{Cortés (1999)} was used to determine the trophic level of the Mexican barracuda (TL _k ), which considers the type of prey found in stomach contents: TLk=1+(∑j=111Pj × TLj, where TL _j is the trophic level of the jth prey and P _j is the proportion of each prey item in the diet. The trophic level of each prey item was obtained from ^{Cortés (1999)}, ^{López-García et al. (2012)}, and ^{Froese and Pauly (2016)}. Numerical data were used for these analyses.

Results

A total of 308 Mexican barracuda were obtained, ranging between 30.6 and 58.7 cm total length and between 405 and 836 g. Of the analyzed stomachs, 264 (86%) contained food and 44 (14%) were empty. According to the CV (≤0.05), the prey species accumulation curve reached its asymptote at 85 stomachs, indicating that the total number of analyzed stomachs was adequate for the characterization of the feeding habits of the Mexican barracuda and the characterization of the diet by sex, size, and season (Fig. 2, Table 1).

Figure 2 Species accumulation curve of Sphyraena ensis prey. The upward arrow indicates the asymptote of the curve. H′, Shannon-Wiener diversity; H′_Max, maximum diversity; H′_Mean, mean diversity; H′_Min, minimum diversity; and CV, coefficient of variation.  

The diet of the Mexican barracuda comprised 13 prey items, including 9 fish, 2 squid, and 2 shrimp species, and unidentified fish remains. Stomach contents weighed 951.35 g in total and included 273 prey items, of which fishes comprised 885.00 g and 235 prey items. The most frequent species in the diet were the fishes Sardinops spp. (Hubbs, 1929) (26.8%, n = 73), Hemiramphus saltator (Gilbert and Starks, 1904) (25.7%, n = 71), and Opisthopterus dovii (Günther, 1868) (12.8%, n = 37). According to the %IRI, the most important prey species in the Mexican barracuda diet were the fishes Sardinops spp. (40.36%), H. saltator (40.24%), O. dovii (10.83%), Anchoa spp. (4.17%), and Mugil cephalus (3.05%) (Fig. 3a, Table 2).

Figure 3 3-D graphs showing trophic indices for prey consumed by Sphyraena ensis (number [abundance], weight [biomass], and frequency of occurrence). Data: general (a), males (b), females (c), small size (d), medium size (e), large size (f), dry season (g), rainy season (h). Abbreviations: Sar, Sardinops spp.; Hem, Hemiramphus saltator; Opd, Opisthopterus dovii; Anc, Anchoa spp.; Muc, Mugil cephalus; Lop, Lolliguncula panamensis; Hyp, Hyaloteuthis pelagica; Cac, Caranx caballus; Laa, Larimus argenteus; Lut, Lutjanus spp.; Otp, Other prey.  

Of the 264 stomachs containing food, 91 were extracted from males and 173 from females. The male diet comprised 10 prey items, whereas the female diet included 13 items. According to the %IRI, the most important prey for both sexes were H. saltator and Sardinops spp. (Fig. 3b, c). There were no significant differences in trophic spectra between males and females (PERMANOVA F = 13.514, P > 0.05) (Table 3).

There were 62 small-sized specimens, which fed on 10 prey items. According to the %IRI, the most important prey in this group’s diet were the fishes H. saltator (48%), Sardinops spp. (24%), O. dovii (14%), Mugil cephalus (Linnaeus, 1758) (6%), Anchoa spp. (Jordan and Evermann, 1927) (5%), and other prey (0.84%) (Fig. 3d). There were 158 medium-sized fish, which fed on 12 prey items. The diet of this group comprised mainly fish, of which the most important, according to the %IRI, were H. saltator (51%), Sardinops spp. (35%), O. dovii (6%), Anchoa spp. (3%), M. cephalus (3%), and other prey (1.61%) (Fig. 3e). There were 44 large barracuda, which fed on 8 prey items. The most important prey in this group’s diet were Sardinops spp. (67%), O. dovii (21%), H. saltator (6%), Anchoa spp. (3%), and other prey (0.04%) (Fig. 3f). There were no significant differences in diets between size categories (PERMANOVA F = 14.468, P > 0.05) (Table 3).

A total of 158 stomachs collected in the dry season and 106 stomachs collected in the rainy season were analyzed. During the dry season, the diet comprised 10 prey items; according to the %IRI the most important prey were Sardinops spp. (71%), O. dovii (11%), Anchoa spp. (10%), and H. saltator (4%), and the remaining items comprised only 4% of the diet (Fig. 3g). During the rainy season, the diet comprised 8 items; the main prey, according to the %IRI, were H. saltator (84%), Sardinops spp. (5%), O. dovii (4%), M. cephalus (3%), and Lutjanus spp. (Bloch 1790) (1%) (Fig. 3h). Significant differences were observed between seasons (PERMANOVA F = 184.66, P < 0.05) and in the interactions between sex and season (PERMANOVA F = 14.488, P < 0.05) (Table 3). The dispersions of the Bray-Curtis dissimilarities were not statistically different for any of the analyzed factors (F: sex = 0.6537, size = 0.7452, season = 1.4383; P: sex = 0.4195, size = 0.4756, season = 0.2315).

The Mexican barracuda can be categorized as a specialist predator (B _i = 0.34). This feeding behavior was consistent for males and females (males: B _i = 0.37; females: B _i = 0.36), for the 3 size intervals (small: B _i = 0.38; medium: B _i = 0.36; large: B _i = 0.48), and for the 2 analyzed seasons (dry: B _i = 0.41; rainy: B _i = 0.39). The feeding strategy confirmed that the Mexican barracuda is a specialist predator (ichthyophagous) that consumes mainly the fishes H. saltator, Sardinops spp., Anchoa spp., O. dovii, M. cephalus, and Lutjanus spp. (Fig. 4). The trophic level calculated for S. ensis and for each sex was 4.1. Small-sized individuals occupied a trophic level of 4.0, medium-sized individuals a trophic level of 4.1, and large-sized individuals a trophic level of 4.0. The trophic level during the rainy season was 4.2, and during the dry season, it was 3.9.

Figure 4 Feeding strategy graph obtained with the Costello method. Prey-specific abundance (%N) and percentage frequency of occurrence (%F) in the general diet of Sphyraena ensis. Abbreviations: Sar, Sardinops spp.; Hem, Hemiramphus saltator; Opd, Opisthopterus dovii; Anc, Anchoa spp.; Muc, Mugil cephalus; Lop, Lolliguncula panamensis; Hyp, Hyaloteuthis pelagica; Pen, Penaeus spp.; Xyp, Xyphopenaeus spp.; Cac, Caranx caballus; Laa, Larimus argenteus; Lug, Lutjanus guttatus; Lut = Lutjanus spp. 

Discussion

Mexican barracuda fed mainly on the fishes Sardinops spp., H. saltator, O. dovii, and Anchoa spp. These are all small pelagic/neritic species that gather in schools and are distributed along the eastern Pacific Ocean, from Baja California, including the Gulf of California, to northern Peru and even Chile (^{Sommer 1995}, ^{Robertson and Allen 2015}).

In general, evidence indicates that the 5 barracuda species found along the Pacific coast of Mexico play a similar ecological role within the limits of their geographical distribution, because the 5 species (S. ensis, S. argentea, S. lucasana, S. idiastes, and S. qenie) are carnivorous predators that feed mainly on small school-forming fish in the pelagic-neritic zone but can also consume benthic crustaceans and cephalopods (^{Sommer 1995}, ^{López-Peralta and Arcila 2002}, ^{Robertson and Allen 2015}).

The Mexican barracuda has a specialist ichthyophagous feeding strategy. This type of strategy has been reported for S. viridensis, S. sphyraena, and Sphyraena chrysotaenia in the Mediterranean Sea (^{Kalogirou et al. 2012}), for S. barracuda in the San Andrés Archipelago in Colombia (^{Hooker et al. 2007}), and for Sphyraena guachancho in the Gulf of Mexico (^{Bedia-Sánchez et al. 2011}).

No significant differences were detected between males and females or between sizes (small, medium, and large) in the diet of the Mexican barracuda due to the predominant consumption of the same fish species (Sardinops spp., H. saltator, O. dovii, Anchoa spp., and M. cephalus) in similar proportions. This trophic behavior indicated no feeding segregation by sex or size. However, large individuals fed on a larger proportion of Sardinops spp. These larger individuals probably have better search, attack, and capture abilities than smaller barracuda. Sardinops spp. are extremely abundant in the southern Gulf of California and in waters along the Pacific coast of Mexico (^{Fischer et al. 1995a}).

According to the optimal foraging theory, feeding on small abundant fish would allow barracudas to obtain greater energetic benefits than they would feeding on large, less available prey that would represent greater energy expenditure during the search, capture, and consumption (^{Gerking 1994}). It should be noted that the diet breadth of the Mexican barracuda increased with size, as larger barracuda also included demerso-pelagic prey in their diet (Carangidae, Lutjanidae, and Mugilidae). ^{Kalogirou et al. (2012)} also reported that the diet breadth of S. viridensis, S. sphyraena, and S. chrysotaenia increased with size, which could be related to improved predatory abilities as they grow.

Although results obtained in this study indicated a specialist feeding strategy, differences found in the diet of the Mexican barracuda between seasons reflected changes in the consumption of Sardinops spp. (84%) during the dry season and of H. saltator (71%) during the rainy season. This alternation of the main prey species could be related to the natural fluctuations in the populations of the prey species. Although no specific data on variations in prey abundance in the study area are available, it is probable that Mexican barracuda prey on the most abundant species in the pelagic-neritic zone, improving their chances of feeding success throughout the year. This would indicate that instead of being a specialist predator, the Mexican barracuda is an opportunistic ichthyophagous predator, with trophic plasticity that allows it to feed on available and abundant fish species. ^{Bedia-Sánchez et al. (2011)} and ^{De Sylva (1963)} reported that the barracudas S. guachancho and S. barracuda alternated the main food components in their diet according to environmental conditions, which affected the availability and abundance of prey species.

The trophic level calculated for the Mexican barracuda was 4.1. This is similar to what has been reported for other ichthyophagous predators with specialist tendencies, such as Xiphias gladius (Linnaeus, 1758) (4.5), Coryphaena hippurus (Linnaeus, 1758) (4.3), Fistularia commersonii (Rüppell, 1838) (4.3), Scomberomorus sierra (Jordan and Starks, 1895) (4.2), Thunnus albacares (Bonnaterre, 1788) (4.1), and Alectis ciliaris (Bloch, 1787) (4.0) (^{López-Peralta and Arcila 2002}, ^{Stergiou and Karpouzi 2002}, ^{Moreno-Sánchez et al. 2011}, ^{Tripp-Valdez et al. 2015}, ^{Froese and Pauly 2016}, ^{Alatorre-Ramirez et al. 2017}), which also feed on school-forming pelagic fishes and on cephalopods and crustaceans. It should be noted that the Mexican barracuda is part of the same trophic guild and interacts with these species in the ecosystems of the southern Gulf of California, which suggests possible interspecific competition. However, it seems that the availability and abundance of small pelagic fishes that form large schools (families Engraulidae and Clupeidae) (^{Tripp-Valdez et al. 2015}, ^{Varela et al. 2017}, ^{Zambrano-Zambrano et al. 2019}), in addition to the morphological differences of predators, allow these species to coexist without affecting their population densities (^{Cruz-Escalona et al. 2000}, ^{Moreno-Sánchez et al. 2015}).

We conclude that the Mexican barracuda is an opportunistic ichthyophagous predator that does not present differences in its diet by sex or size, but it can change its preferred prey intake seasonally, depending the availability and abundance of its prey in the environment. This behavior, as stated in the optimal foraging theory, ensures optimization of the trade-off between consumption and energy use and, therefore, better bioenergetic performance.