Interannual variation in the metazoan parasite communities of bigeye trevally Caranx sexfasciatus (Pisces, Carangidae)

Parasite communities in Caranx sexfasciatus were characterized and analyzed to determine any interannual variations in structure and/or species composition. In total, 422 C. sexfasciatus were collected from Acapulco Bay, Mexico, between May 2016 and March 2019. Thirty-two taxa of metazoan parasites were identified: five Monogenea, thirteen Digenea, one Acanthocephala, one Cestoda, three Nematoda, seven Copepoda, and two Isopoda. Monogeneans were the most frequent and abundant parasite species in all sampling years. Parasite species richness at the component community level varied significantly from 8 (May 2016) to 25 (March 2019) and was similar to previous reports for other species of Carangidae. The component communities and infracommunities in C. sexfasciatus were characterized by low parasite species numbers, low diversity, and dominance of a single species (the monogenean Neomicrocotyle pacifica). Parasite community structure and species composition varied between sampling years and climatic seasons. Seasonal or local fluctuations in some biotic and abiotic environmental factors probably explain these variations.


Introduction
Parasite communities in marine fish consist of ectoparasites (monogeneans, copepods, and isopods) and endoparasites (digeneans, cestodes, nematodes, and acanthocephalans). Both types of parasite exhibit different transmission strategies to infect their hosts; for example, ectoparasites are often transmitted directly between individual hosts through contact, whereas endoparasites often use trophic transmission routes [9,26,37]. Transmission strategies of parasites are linked to host behavior. For instance, ectoparasite populations may be more abundant on species of fish that form large schools than on solitary species [23,29,31], because the probability of a transmission stage (e.g., eggs, larvae) coming into contact with a host increases as host density rises.
Studies of parasite communities are conducted mainly at two different hierarchical levels: component community (between locations or host populations) and infracommunity (between individual hosts) [14,17,51]. In recent years, a central theme in parasite ecology has been the identification of the factors (biotic or abiotic) determining species richness and composition in these communities. However, an understanding is also needed of how consistent parasite community richness and species composition are in space and time. For example, spatial or temporal variation in community structure may indicate how important local environmental factors (abiotic and biotic conditions) are in structuring a parasite community, particularly when disparities are found between different host species [1,24,46,50,53]. There is still no consensus on whether marine parasite communities can exhibit temporal variations in structure and species composition. Some studies suggest that the infection levels of most parasite species do not experience significant changes over time, and that communities tend to be generally stable in species composition for long periods of time [4,12,16,43]. Others indicate that even though parasite species composition may be relatively stable over time, a community can undergo substantial changes in structure due to variations in local environmental factors [14,39,51].
The bigeye trevally, Caranx sexfasciatus Quoy & Gaimard 1825, is an economically important pelagic fish species distributed widely along the eastern coast of the Pacific Ocean from CA, USA to Ecuador and the Galapagos Islands [2]. Bigeye trevally often form large stationary daytime schools, but usually tend to be solitary at night when feeding. Their diet consists mainly of fish, but the species also preys on squids and crustaceans [2]. Despite its ecological and economic importance, no information is available to date on its parasite fauna along the Pacific coast of Mexico. The present study objectives were (1) to characterize the metazoan parasite communities of C. sexfasciatus; and (2) to evaluate possible interannual variability in its parasite communities.

Materials and methods
A total of 422 individuals of C. sexfasciatus were obtained from commercial fishermen over a four-year period (May 2016-March 2019) from Acapulco Bay (16°51 0 N; 99°5 2 0 W) in Guerrero, Mexico. The tropical Pacific region experiences two distinct climatic seasons: a rainy period from June to November (total precipitation % 950 mm), and a dry period from December to May (total precipitation < 70 mm). In the years 2017 and 2019, a single sampling was carried out, but in a different climatic season: August (rainy season) and March (dry season), respectively. In 2016 and 2017, two samplings were carried out per year, which included both climatic seasons: May and October in 2016, and May and November in 2018. Records of surface temperatures of the ocean waters and salinity per sampling date were obtained from other studies carried out at the same location but not yet published. Multivariate El Niño index values (MEI) for each sampling date were obtained from the National Oceanic and Atmospheric Administration (NOAA: https://www.esrl.noaa.gov/psd/enso/mei/table.html). Fish were measured and weighed at the time of collection. A complete necropsy was made for each specimen and all parasites were collected from the internal and external organs [52].
Contents of the digestive tract were examined to identify prey items consumed by this species. Dietary items were identified to the family level when possible. Prey item analysis was carried out using the frequency of occurrence method [19,30]. Parasites were identified to the lowest possible taxonomic level and vouchers of the most abundant and best-preserved specimens deposited in the Coleccion Nacional de Helmintos (CNHE), Instituto de Biologia, Universidad Nacional Autonoma de Mexico, Mexico City.
Infection levels for each parasite species were described using prevalence (percent of fish infected with a particular parasite species); mean abundance (mean number of individual parasites of a particular species per examined fish), expressed as the mean ± standard deviation (SD); and intensity (number of a particular parasite species per infected fish), expressed as range (minimum-maximum) [5,6,8]. Possible differences in infection levels between sampling years and climatic seasons were identified using G-tests [44], and a general linear model (GLM) for abundance. The dispersion index (DI = variance to mean abundance ratio) was applied to describe parasite dispersion patterns. The infracommunity index (ICI) [55], which describes the frequency of double and multiple infections by a single species of parasite in a distinct host (affinity level of a species of parasite; i.e. species with great tendency to join the infracommunity), was also calculated. Spearman's correlation coefficient (r s ) was used to determine possible relationships between total host length and abundance or DI values of each parasite species.
Analyses were done at the levels of component community (i.e. total parasite species in all fish collected at a sampling year) and infracommunity (i.e. total parasite species in each individual fish). Component community parameters included total species richness, total number of individuals of each parasite species, the Shannon-Wiener Index (H) as a measure of diversity, species evenness (equitability), and the Berger-Parker Index (BPI) as a measure of numerical dominance [25]. The qualitative Sorensen index and quantitative percentage of similarity (PS) index were used to evaluate similarity and difference in parasite community species composition between sampling dates. Differences between component community parameters were identified with Student t and v 2 tests.
Infracommunities were described in terms of mean number of parasite species per host, mean number of individuals of each species, and the mean Brillouin Diversity Index (H 0 ) value per host. The multivariate general linear model (GLM) was used to identify possible differences in infracommunity parameters (dependent variables) between sampling years and climatic seasons (predictor variables); fish body size (total length) was used as a covariate to control for the influence of host body size. The significance of all statistical analyses was established at p < 0.05, unless stated otherwise. A multivariate analysis (principal component analysis, PCA) was applied to identify factors that influenced parasite infracommunity species richness and diversity. The predictor variables used were: Fulton's condition factor (K n ) [13]; host diet diversity (calculated as the diversity of items consumed by host populations at each sampled date, through the use of the Shannon-Wiener index at the family level); number of endoparasite and ectoparasite species; surface temperature; salinity; MEI values; sampling year; and climatic season. The Kaiser-Meyer-Olkin (KMO) test of sampling adequacy for each variable in the model, as well as the Bartlett sphericity test, which evaluates the possibility that there is redundancy among variables were applied. The variance maximizing rotation method was applied to produce the two ordination axes. Discriminant function analyses based on Mahalanobis distances were used to identify possible differences in parasite community structure between sampling years. The probability of correct classification expected by chance alone of fishes to any of the four sampling years was calculated using the proportional chance criterion, which is a simple method to account for differences in sample sizes between the groups being compared [36]. Only parasite species with a prevalence >10% in at least one of the sampling years (a component species; sensu [7]) were included in this analysis.

Species composition
Thirty-two taxa of metazoan parasites (23 of helminths and nine of Crustacea) were recovered and identified (9866 individual parasites) from 422 individuals of C. sexfasciatus collected from Acapulco Bay, Mexico. Five species of Monogenea (adults), 13 Digenea (12 adults and one metacercaria), one Acanthocephala (adult), three Nematoda (one adult and two larvae), seven Copepoda, and two species of Isopoda were collected (Table 1). Species richness was highest among the digeneans, representing 37.1% of the total species, followed by copepods (20%). Based on infection site, 14 species of parasite were classified as ectoparasites and 18 as endoparasites. The numbers of species of ectoparasite varied significantly from 2.02 ± 0.90 in 2016 to 2.41 ± 1.20 in 2019 (ANOVA F 3,405 = 3.23, p < 0.05) (Fig. 1

Interannual variation in infection levels
For eight of these perennial species (the exceptions were Pr. manteri, S. apharei and Ca. robustus), prevalence varied significantly between sampling years, though no clear pattern was observed in variation. For the parasites B. varicus   (Table 1). Prevalence values were generally positively correlated with mean abundance values, indicating that the most prevalent species also were the most abundant (r s = 0.946, p < 0.01).

Spatial dispersion
The dispersion index (DI) values indicated that at least 10 species (31%) exhibited an aggregated dispersion pattern in one or more sampling years (Table 1). Higher mean aggregation values (DI > 8) occurred in several years for the monogeneans N. pacifica and Pr. manteri ( Table 1). The DI values of each parasite species correlated positively with its prevalence (r s = 0.824, p < 0.01), total number of individuals (r s = 0.878, p < 0.01), and mean abundance (r s = 0.871, p < 0.01) values, but not with host body size or sample size (p > 0.05).

Host diet composition
The alimentary spectrum of the fish included twelve prey items, of which smaller fish (36%), crab larvae (24.5%) and penaeid larvae (15.6%) represented the largest proportions. Additonal prey items accounted for 24% of the diet, and included amphipods, polychaetes, mollusk larvae, isopods, stomatopods, pistol crabs, and copepods. Diet composition varied between sampling years (t = 3.10, p < 0.05). For example, in May 2016 diet composition was less varied (H = 0.68, Table 2), with crab larvae being the main prey item. Feeding habits also changed with fish body size; larger individuals consumed higher percentages of fish while smaller ones fed on higher percentages of crustaceans and other prey.

Component community
Species richness of parasite by sampling year (Table 2) varied widely from eight (May 2016) to 25 (March 2019) (v 2 = 11.6, p < 0.05). No correlation was observed between sample size and species richness at this level, indicating that the different sample sizes used in the analyses had no effect on the results. The total number of individual parasites ranged from 670 (October 2016) to 3465 (March 2019) (v 2 = 3,997.2, p < 0.05). The only dominant species was the monogenean N. pacifica (Table 2), although its degree of dominance varied significantly between sampling years (t = 4.70, p < 0.01). Shannon-Wiener diversity index values were generally low, ranging from 0.90 (October 2016) to 1.40 (March 2019), but varied between sampling years (t = 3.68, p < 0.01). Similarity between component communities was slightly low overall (<65%, Fig. 2) and was higher at the qualitative level (mean = 64.05%) than at the quantitative level (mean = 56.15%).

Infracommunity
The body size of individuals of C. sexfasciatus varied from 19.0 ± 1.3 cm (October 2016) to 28.8 ± 4.5 cm (August 2017) ( Table 2), and it differed significantly between sampling years (ANOVA F 3,402 = 32.3, p < 0.01). Mean species richness of parasites ranged from 2.22 ± 1.30 to 2.96 ± 1.54, and the mean number of individual parasites per fish from 12.71 ± 11.5 to 31.50 ± 30.35 ( Table 2). The Brillouin diversity index (H 0 ) values varied from 0.53 ± 0.48 to 0.71 ± 0.58. The mean number of species (GLM; F 3,406 = 3.95, p < 0.01) and the mean number of individuals (GLM; F 3,406 = 10.72, p < 0.01) were highest in March 2019 (Table 2), although mean diversity (H 0 ) did not exhibit significant variation between sampling years (p > 0.05). Dominance of the monogenean N. pacifica at the infracommunity level was very similar to that at the component level (Table 2). Considering all samples, the mean number of species (r s = À0.368, p < 0.01), and the mean diversity (r s = À0.098, p < 0.05) exhibited negative correlations with host body size of fish, while the mean number of individual parasites was positively correlated (r s = 0.313, p < 0.01). The infracommunity index values (ICI) indicated that N. pacifica, Pr. manteri, and Ca. robustus had the highest number of double or multiple co-occurrences with other parasite species (ICI mean > 0.15) in all the sampling years (Table 1).

Multivariate analyses
The KMO (0.476) and the Bartlett's tests results (v 2 = 2,944.2; df = 45, p < 0.001) for the PCA, applied to identify the influence of biotic and abiotic factors on parasite infracommunity structure (Fig. 3), indicated that there is a sufficient relationship between the number of variables studied and the sample sizes, confirming the relevance of the PCA. The first three component variables generated by the model explained 61.9% of total variance, contributing 24.9% (eigenvalue = 3.74), 19.8% (eigenvalue = 2.97), and 17.2% (eigenvalue = 2.58). The first variable suggests that the richest and most diverse infracommunities among all the sampled years were characterized by a significant number of ectoparasite and endoparasite component species, with more homogeneous species abundances. The second variable indicated that both the total number of ectoparasites and the parasite load per infracommunity were highest in larger hosts with a more varied diet. The third variable associated host diet with the environmental factors that were studied (Table 3). This variable indicated that host diet was more varied during the dry season when water surface temperature was slightly cooler (as suggested by the negative correlations), which occurred mainly during the final sampling year (Tables 2 and 3).  In the model constructed to identify possible interannual differences in parasite community structure, the first two discriminant variables explained 86.9% of the total variance, contributing 56.2% (eigenvalue = 0.164) and 30.6% (eigenvalue = 0.090). A significant overall group effect was observed (Wilks' lambda = 0.759, p < 0.001). Individual fish were mainly distributed along the first axis (Fig. 4). Dimensionality tests showed that the four sampling years were significantly different in both dimensions (v 2 = 110.37, df = 15, p < 0.001). Each fish was correctly assigned to one of the four sampling years with a 41.1% accuracy, more than double that achieved by chance alone (16%).
Of the nine component species (i.e. prevalence > 10%), only five were accepted by the model (Table 4) based on their lower Wilks' lambda values. High abundance of the monogenean Pr. manteri and lower abundance of the digenean E. virgulus were characteristic of hosts collected in 2016. The monogenean N. pacifica and the copepod L. ilishae effectively functioned to identify hosts that were collected in 2019, while the copepods Ca. alalongae and L. ilishae effectively distinguished between hosts collected in 2017 and 2018.

Parasite community species composition
All the present parasite species records (23 helminths and nine crustaceans) are new geographical records for C. sexfasciatus on the Pacific coast of Mexico (Table 1). Ectoparasites (five monogeneans, seven copepods, and two isopods) numerically dominated the parasite communities, representing 92.71% of the total number of individual parasites recovered. This predominance is probably due to the monogeneans' transmission strategies (direct contact), increasing their transmission probabilities especially in gregarious fish such as C. sexfasciatus [27,42]. Eight of the reported monogenean species (including N. pacifica, Pr. manteri, and Ps. selene) have been recorded in Carangidae from the Pacific coast of Mexico [32]. Copepods are a highly diverse group that mainly parasitize marine fish [22,28,45]; however, this crustacean group has received limited attention in Mexico [14]; in the present study, they were the second most diverse group and represented 22% of the total taxa recovered (Table 1).

Interannual variation in parasite infection levels
At least five species of parasite (two endoparasites -B. varicus and a tetraphyllidean cestode, and three ectoparasites -N. pacifica, Ps. selene and Ca. alalongae) exhibited significant interannual differences in infection levels (Table 1). Endoparasites often are trophically transmitted, so variations in infection levels can be explained by interannual changes in host diet and availability of prey harboring infective stages [16,18,23,38]. The population density and gregarious behavior (schooling behavior) of C. sexfasciatus [2] may be a better explanation for the annual variation observed in the three species of ectoparasites. Fish that form large schools facilitate ectoparasite transmission, particularly parasites with a direct transmission cycle such as monogeneans and copepods [14,23,39,41,51].

Dispersion pattern
The DI values indicated that at least 31% of the species of parasites exhibited aggregate dispersion in one or more of the sampling years (Table 1). Aggregation is considered a typical  dispersion pattern of parasites in marine fish [3,18,37]. However, the widely variable aggregation levels of five helminth species (the monogeneans N. pacifica and Pr. manteri; the digeneans B. varicus and Ectenurus virgulus; and the nematode Anisakis sp.) suggest the existence of differences in host exposure rates to parasites. Differing exposure rates could be due to the influence of local environmental factors [35], or the differences in host feeding behavior revealed in the diet analysis. In contrast, some contact-transmitted parasites such as monogeneans, exhibit high host specificity and can be more abundant in larger fish that form schools [23,29], like C. sexfasciatus. Some studies also suggest that DI values can be influenced by sample size [33,35], but there was no positive correlation between DI values and sample sizes in the present analysis.

Component communities
The parasite communities in C. sexfasciatus exhibited similar patterns at the component and infracommunity levels: low species numbers (2.22-2.96 species on average per host), low species diversity, and dominance by a single species (the monogenean Pr. manteri, Table 2). Species richness at the component level (8-25 species, Table 2) is similar to that reported previously in other Caranx species in the Americas, such as C. hippos (19 species), C. latus (17 species), and C. caballus (18 species) [15,21]. However, overall parasite fauna of C. sexfasciatus included 32 species across the four sampling years. The broad geographical distribution of C. sexfasciatus may explain this greater overall richness since  hosts with a broad geographical distribution are exposed to large numbers of parasite species by interaction with myriad intermediate host species throughout their distribution range [40]. Species composition similarity between the parasite communities ( Fig. 2) was considered generally low at both levels (qualitative and quantitative levels: <65%), compared to the similarity recorded for parasite communities of other marine fish examined on the Mexican Pacific coast, such as Caranx caballus 78.5% [14], Oligoplites altus 85% [39], or Parapsettus panamensis 76% [51]. This low similarity can be attributed to the fact that 47% (15 species: six ectoparasites and nine endoparasites) of the identified parasite species occurred only in one or two sampling years (Table 1). This suggests that annual variations in biotic (e.g. diet, body size, availability of larva-infected prey) and/or abiotic environmental factors [14,23,39,49,51], may have been responsible for the low similarity between the parasite communities of C. sexfasciatus observed in the present study.

Infracommunities
Several biotic and abiotic factors are known to strongly influence species richness and diversity in the parasite communities of marine fish over time [14,23,34,39,48,51]. The PCA results (Fig. 3) indicated that infracommunity species richness and diversity depended heavily on the occurrence of a distinctive set of component species (both ectoand endoparasite species). However, these component species' frequency or abundance varied between sampling years (Table 1). Temporal variations in parasite infracommunity structure and species composition have been attributed to variations in the prevalence and/or abundance of some dominant taxa [4,14,16,38,39,51]. As the ICI results suggest, the ectoparasites N. pacifica, Pr. manteri, and Ca. robustus play an important role in structuring the parasite infracommunities of C. sexfasciatus. Due to ease of transmission through direct contact, ectoparasite populations contribute substantially to the structuring of infracommunities in gregarious fish such as carangids [23,29,31]. Fluctuations in the infection levels of some endoparasite species (e.g. B. varicus and the Tetraphyllidean cestode) can be attributed mainly to changes in C. sexfasciatus diet or foraging habitat, both between sampling years and climatic seasons (dry/rainy). Host body size was an important predictor of the total parasite load, as suggest by the global positive correlation (r = 0.633) recorded for this variable in the PCA (Table 3). In marine fish, body size has proved to be the main predictor of total parasite abundance [23,39,40]. A large body size can facilitate parasite colonization, since larger individuals tend to ingest greater quantities of food and are older and thus have had more time to accumulate parasites than smaller individuals [39,40]. However, the negative correlations registered between host body size and species richness and diversity parameters suggest that the richest and most diverse parasite infracommunities were recorded in smaller rather than larger fish. Ontogenic changes in the feeding habits [10,11,20,51] of C. sexfasciatus can explain the broad dietary diversity in smaller fish (for example, October 2016 and November 2018 samples, Table 2). This would in turn raise infracommunity species richness and diversity in these individuals (Table 2), as indicated in the diet analysis. Parasites are known to be useful as biological tags in distinguishing between fish stocks of the same species [47,54], but the parasite fauna method has rarely been used to quantify possible variations in community structure and species composition over time. The discriminant analysis results (Table 4) indicated that the high variation in the infection levels of at least five parasite species (the monogeneans N. pacifica and Pr. manteri; the digenean E. virgulus; and the copepods Ca. alalongae and L. ilishae) may generate substantial changes in the parasite community structure of C. sexfasciatus over time. In other words, even though community species composition remains relatively stable, its structure may be less predictable.
Overall, the parasite communities of C. sexfasciatus were characterized by high numerical dominance of ectoparasites, mainly monogenean species. Community structure and species composition varied between sampling years and climatic seasons. Despite occurrence of a distinctive set of host-specialist parasites (monogenean species), similarity between the component parasite communities was generally low. Seasonal or local variations in some biotic (e.g. feeding behavior, body size, and infected prey availability) and abiotic environmental factors are possible sources of the observed interannual variations in C. sexfasciatus parasite community structure and species composition.
Acknowledgements. The authors thank the fishermen who caught the fish at each sampling date, as well as the students of the Marine Ecology Academic Unit (UAGro) for their assistance with field and laboratory work. Four anonymous reviewers provided useful comments that substantially improved this manuscript.