Sampling design

Five stations located along the coast of Veracruz, south of the VRSNP, were sampled in March 2018 (Fig. 1), covering latitudes from 20º38 N to 20º48 N, longitudes from 96º51 W to 96º 56 W, and depths from 45 to 71 m. All stations had a fine and thick silty substratum. Sampling was designed to test for different times in a day, with 2 samplings in the morning (station 4, 6:47 AM; station 2, 8:46 AM), 2 in the afternoon (station 1, 12:48 PM; station 3, 1:58 PM), and 1 at night (station 5, 8:00 PM). All samples were collected using a shrimp trawl net (18.3 m in length, 3.4 cm mesh size), which was towed for 20 to 30 min at a constant speed of 2.5 knots. The collected specimens were frozen and identified to species level (^{Schultz 1986}, ^{Labastida-Estrada et al. 2019}).

Figure 1 Map showing the 5 oceanographic sampling stations (filled circles) along the southern coast of Veracruz, Gulf of Mexico. Dashed isolines indicate depth in meters. VRSNP: Veracruz Reef System National Park; TL: Tampamachoco Lagoon. 

Stomach content analyses

The feeding habits of specimens were determined by analyzing stomach contents. All specimens were measured (standard length, SL) and weighed (wet weight). The digestive tract of each specimen was removed and weighed, and stomach contents were all extracted and identified to the lowest possible taxonomic unit. The contribution of each prey taxon to the overall diet was assessed using 3 relative metrics. Of the three, the percent composition by weight takes into account the wet weight of each component, and the total weight of a food category is then expressed as a percentage of the overall weight of stomach contents (^{Vega-Cendejas et al. 1994}). The percent composition by area was used when stomach contents were constituted of very small prey whose weight was not possible to record with an analytical balance; this method provides a more representative measure of biomass and can be applied to all food components (^{Canto-Maza and Vega-Cendejas 2007}). The percent frequency of occurrence takes into account the number of times a prey appears in the stomach contents of a group of fish and is expressed as a percentage (^{Canto-Maza and Vega-Cendejas 2007}).

Size, weight, and diet

Since the size and weight variables did not meet the normality criteria, we used non-parametric methods. The relationship between the SL and weight of specimens and the weight of the digestive tract was analyzed by Spearman’s D correlation (^{Press et al. 1992}). Then, to make a size/weight group (SWG) classification, a cluster analysis based on the size and weight of specimens was performed by the unweighted pair group method with arithmetic mean (UPGMA) using the Euclidian distance. UPGMA is an agglomerative hierarchical clustering method that constructs a tree that reflects the structure present in a pairwise similarity matrix (^{Sokal and Michener 1958}). The obtained grouping was analyzed via a permutational multivariate analysis of variance (PERMANOVA) to test for significant differences between 2 or more groups (^{Anderson 2001}). PERMANOVA was also used to determine if there were differences in the diet across SWGs using the information gathered from prey items, such as occurrence, area, and weight. All statistical analyses were done using PAST 3.20 (^{Hammer et al. 2001}).

Diet

Three main taxonomic groups were identified in the diet of P. volitans, the phylum Mollusca, the subphylum Crustacean, and the superclass Osteichthyes, being the bony fishes the dominant items (Table 2). Of the 32 classified items, 21 were bony fishes, 9 crustaceans, and 2 mollusks. The values for percent composition of prey ítems by area and by weight were highly similar; approximately 95.40% of the total composition by area and weight was made up of bony fishes, whereas crustaceans made up approximately 4.20% and mollusks approximately 0.36% of total composition (Table 2).

Size, weight, and diet

Correlation was strongest between the SL and weight of specimens (r = 76.5, P < 0.000), followed by the correlation between the weight of specimens and the weight of the digestive tracts (r = 146.5, P < 0.001) and the correlation between the SL of specimens and the weight of the digestive tracts (r = 238, P < 0.005). With data on specimen SL and weight, the UPGMA produced 3 statistically independent groups (F = 120.6, P < 0.000): SWG1, which grouped 8 of the smallest and lightest specimens (μ SL = 17.74 cm, weight = 112.4 g); SWG2, which grouped 6 specimens with mean size/weight (μ SL = 26.55 cm, weight = 328.55 g); and SWG3, which grouped 3 of the largest and heaviest specimens (μ SL = 39.47 cm, weight = 842.17 g). The comparison of frequency of occurrence, area, and weight between SWGs showed no significant differences (F = 1.275, P = 0.285; Fig. 2).

Figure 2 Main food categories in the diet of the lionfish Pterois volitans sampled along the southern coast of Veracruz by size/weight groups (SWG). Mean size and weight values are given in parentheses for each group. FO: frequency of occurrence; PA: percent composition by area; PW: percentage composition by weight. 

DISCUSSION

Our study corroborates the results of ^{Morris and Akins (2009)}, ^{Layman and Allgeier (2012)}, and ^{Arredondo-Chávez et al. (2016)}, which indicate that the diet of lionfish in an invaded range is largely composed of bony fishes, crustaceans, and mollusks, although we found mollusks in the diet of only SWG1. In the diet contents of P. volitans, we found fishes belonging to the genus Serranus, which in this area comprises 6 reef-associated species, the fish species Scyacium gunteri, which is a bottom dweller unassociated with reefs (^{Robertson and van Tassell 2019}), and hard-shelled organisms belonging to the infraorder Brachyura and the classes Bivalvia and Gastropoda. We did not find differences in the diets between SWGs; however, it was clear that SWG1 had the more diverse diet.

According to our study, the highest P. volitans density was found at station 5, the one closest to the VRSNP, which could indicate that the closer the area to the reef the higher the density. Station 5, however, was only sampled at night and the high density there could also just be a corroboration of the nocturnal habits of the lionfish (^{Myers 1991}). Our results indicate low lionfish density in the study area, as we captured 1 to 10 specimens per station in each trawl, which covered an area of approximately 22,715 m2. In our study we tried testing the use of a shrimp trawl net with a small mesh size to capture lionfish specimens in a wide range of sizes. Although we achieved the objective, this methodology is useful only for flat bottoms, away from the reefs.

The experience during these years of the invasion of P. volitans in the Gulf of Mexico and the Caribbean made it clear that efforts must be directed to control lionfish populations, as eradication seems unfeasible (^{Albins and Hixon 2013}). It is important to produce information that reveals how this species interacts within an ecosystem in order to propose effective plans to buffer its negative effects. Trophic data can help us anticipate its arrival by directing conservation efforts to protect the key species found in its diet. Trophic analyses can help determine vulnerable life stages given the specific dietary needs, and control efforts can therefore be more specific by focusing on those life stages. The present study was a first attempt to understand the relation between the size, weight, and diet of the invasive lionfish P. volitans on the southern coast of Veracruz, Mexico.