Introduction

Pagrus pagrus (Linnaeus, 1758) (Perciformes: Sparidae), popularly known as red porgy, is a demersal fish that feeds on a wide variety of benthic and other demersal fish, and invertebrates (^{ÁVILA-DA-SILVA; HAIMOVICI, 2004}). This species is widely distributed in the Eastern Atlantic from the British Isles south to the coastal zone of Angola, and also in the Mediterranean and the Adriatic Sea. In the Western Atlantic, P. pagrus is distributed from the coast of New York, USA, to southern Argentina (^{MANOOCH; HASSLER, 1978}; ^{MENEZES; FIGUEIREDO, 1980}).

Pagrus pagrus is a fish of significant commercial value, particularly for export. It is caught using various devices at fishing depths of up to 200 m (^{ÁVILA-DA-SILVA; HAIMOVICI, 2004}). In Brazil, this species is widely distributed along the coast of southeastern to southern regions (^{ÁVILA-DA-SILVA; HAIMOVICI, 2004}). Currently, P. pagrus is included in the list of endangered species (^{IUCN, 2013}).

To date, 29 species of parasites of P. pagrus distributed among Nematoda, Digenea, Monogenea, Cestoda, Acanthocephala and Copepoda have been recorded in Brazil (^{LUQUE; TAVARES, 2007}; ^{KOHN et al., 2007}; ^{MUNIZ-PEREIRA et al., 2009}; ^{LUQUE et al., 2011}; ^{COHEN et al., 2013}). Nematodes are the most numerous parasite group in this host, with nine recorded species (^{LUQUE et al., 2011}). In Brazil, a description of the structure of the parasite communities of P. pagrus was previously published by ^{Paraguassú et al. (2002)}.

Communities of parasites of marine fish are often unstructured and unpredictable because the vagility, behavior, physiology and feeding habits of the hosts, as well as the phylogenetic specificity and possible interactions between parasites (^{LUQUE, 2004}; ^{LUQUE; POULIN, 2008}). A large number of studies have focused on the structure of parasite communities of marine fish. However, many of them do not address spatial-temporal variations or the determination of local processes and those of short duration that may affect the spatial-temporal dynamics of parasite populations and communities (^{POULIN; VALTONEN, 2002}).

The aim of this study was to detect possible changes in the structure of the communities of metazoan parasites of P. pagrus after a long time period (12 years) on the coast of the state of Rio de Janeiro, Brazil.

Materials and Methods

We examined 100 specimens of P. pagrus collected between January and February 2012. Data from this study were compared with data from 90 specimens studied by ^{Paraguassú et al. (2002)}, using the original raw data used in that study. In both studies, the fish were caught in the coastal zone of Cabo Frio, state of Rio de Janeiro (22°52′43″S, 42°1′12″W), Brazil. The host species were identified in accordance with ^{Menezes and Figueiredo (1980)}.

The analysis included only parasite species with prevalence higher than 10% (^{BUSH et al., 1990}). The Spearman rank correlation coefficient (rs) was calculated to determine possible correlations between host length and total abundance of parasites. The Pearson correlation coefficient r was used as an indication of the relationship between the total length of the host and parasite prevalence, with further prior processing of prevalence data (^{ZAR, 1996}). The Student ttest was used to make comparisons with data from ^{Paraguassú et al. (2002)}, in order to ascertain whether the relationships remained the same after 12 years. This test was used for statistical comparisons between the quantitative characteristics of the metazoan infracommunities, using data from ^{Paraguassú et al. (2002)}. The species diversity of the parasite was calculated using the Brillouin index (H), given that each fish analyzed corresponded to a full sample community (^{ZAR, 1996}), so as to ascertain whether the diversity values remained similar after 12 years. The likely variation in sex-related host diversity was analyzed using the Mann-Whitney U test, and the host total length was tested using the Spearman correlation coefficient, and the results were compared with the study of ^{Paraguassú et al. (2002)}. The possible interspecific association between the competing species was determined using the chi-square test. Possible covariance between the abundance of competing species was analyzed using the Spearman correlation coefficient to check whether the same associations were detected 12 years after the study of ^{Paraguassú et al. (2002)}. The ecological terminology was used in accordance with ^{Bush et al. (1997)}. Chi-square analyses were used to test significant differences in the prevalence of parasites between the two samples and the Mann-Whitney test was used for comparison of parasite abundances. The statistical significance level was set at P < 0.05.

Non-metric multidimensional scaling (NMDS) based on the Bray-Curtis similarity index was used to rank the infracommunity metazoan parasites of the current study and make comparisons with the data of ^{Paraguassú et al. (2002)}, according to their relative similarities in parasite abundances. This analysis was performed using 100 random departures with two and three-dimensional configurations in order to find the optimum ranking. All the methods yielded similar results, but three-dimensional ranking based on the transformation log(x+1) was ultimately chosen because it provided the best quality ranking (assessed using the stress value and inspection of Sheppard plots). One-way analysis of similarity (ANOSIM) was used to determine the significance of possible grouping patterns in the NMDS ranking (^{CLARKE, 1993}). Percentage similarity-contribution of species (SIMPER) was used to define the contribution of the species within and among groups (current study and ^{Paraguassú et al. (2002)}) shown in the NMDS (^{CLARKE; WARWICK, 2001}). A cumulative contribution of 80% was applied as in ^{Boaventura et al. (2002)}.

The differences in infracommunity composition were tested through one-way permutational multivariate analysis of variance (PERMANOVA) of samples from the year 2000 (data from ^{Paraguassú et al. (2002)}) and 2012 (present study). A sequential sum of squares (Type I SS) was applied because host size was introduced as a covariable (ANCOVA model) due to the known effect of fish size-age on parasite burdens. The structures of parasite infracommunities between the samples per year (2012 and 2000) were compared using a model (1 x 2 factorial design, with year as a random factor) by testing for the main effects after 9999 permutations. The same model was tested for differences in dispersion using the PERMDISP routine (^{ANDERSON et al., 2008}). Dispersion distances to centroids were measured, and each term in the analysis was tested using 9,999 permutations, with significant terms investigated using a posteriori pairwise comparisons with the PERMANOVA t-statistic (^{ANDERSON et al., 2008}). Statistical analyses were carried out using the PRIMER^® v.6 and PERMANOVA+ for PRIMER software (^{CLARKE; GORLEY, 2006}; ^{ANDERSON et al., 2008})

Results

Infracommunities

All the fish were parasitized by at least one parasite species. A total of 1,902 individual parasite specimens were collected in 2000 and 5,347 in 2012, with mean abundances of 21.1 ± 17.4 (1-100) and 143.5 ± 53.4 (1-672) respectively. Positive relationships between the total parasite abundance and the total body length of the fish were observed in both studies: current study (rs = 0.334, P = 0.001) and ^{Paraguassú et al. (2002)} (rs = 0.235, P = 0.025). The mean richness of parasite species (3.4 ± 1.4), (3.5 ± 1.2) was not correlated with total host body length (rs = 0.078, P =0.437), (rs = 0.102, P=0.335). The mean diversity of parasite species (H) was 0.537 ± 0.569 and 0.306 ± 0119. The maximum diversity value was 1.136 and 0.644 in the current study and ^{Paraguassú et al. (2002)}, respectively. The diversity of parasites was not correlated with total host body length (rs = 0.051, P = 0.610), (rs = 0.120, P = 0.260) and no significant difference in parasite diversity was observed between the sexes, in either study (t = 0.514, P = 0.773), (t = 0.873, P = 0.384) (Table 3).

Table 3. Characteristics of the parasite infracommunities of Pagrus pagrus in the coastal region of the state of Rio de Janeiro, Brazil. t = Student test value for comparison between the hosts, in the study by ^{Paraguassú et al. (2002)}(P1) and in the present study (P2). 

Characteristics/ species	P1	P2	t
Species richness	22	19	-----
Total number of specimens	1902	5347	-----
Mean parasite richness	3.5 ± 1.2	3.4 ± 1.4	–0.8034
Total mean abundance	21.1 ± 17.4	143.5 ± 53.4	–3.7727*
Brillouin index (mean value)	0.306 ± 0.119	0.537 ± 0.569	–0.6541
Dominant species	Hysterothylacium sp.	Hysterothylacium sp.	-----
Berger-Parker Index (mean value)	0.787 ± 0.172	0.786 ± 0.185	–5.9873*

^*Significant values.

The infracommunities of ectoparasites (monogeneans and copepods) were separated to determine possible interspecific associations. Larval stages of nematodes and acanthocephalans, and adult digeneans were not included in this analysis because only one species had prevalence exceeding 10% (Table 1). Among ectoparasites, three pairs of species shared significant positive covariation (Table 4) and two species pairs shared a significant positive association (Table 4). Prevalence and abundance were compared between studies (Table 5), and this showed that the prevalence and abundance were higher in the previous study.

Table 4. Spearman correlation coefficient values (rs) and chi-square values (χ2) for co-occurrences of ectoparasite species pairs in Pagrus pagrus from the coastal region of the state of Rio de Janeiro, Brazil, in the study by ^{Paraguassú et al. (2002)} (P1) and in the present study (P2). 

Species pairs	P1				P2
	rs	p	χ2	p	rs	p	χ2	p
Anoplodiscus longivaginatus - Echinopelma brasiliensis	0.233	0.026	5.240	0.022	–0.062	0.535	19215*	0.004
A. longivaginatus - Encotyllabe spari	–0.202	0.055	2.330	0.127	0.081	0.42	46.702	0.185
A. longivaginatus - Lamellodiscus baeri	0.191	0.070	3.501	0.061	0.001	0.995	6.459	0.971
A. longivaginatus - Lernanthropus caudatus	0.002	0.981	0.050	0.921	–0.022	0.822	1.444	0.998
E. brasiliensis - E. spari	–0.126	0.235	0.201	0.654	0.073	0.468	18.116	0.871
E. brasiliensis - Lamellodiscus baeri	0.277*	0.008	5.770*	0.016	0.353*	0.0003	34.68*	0.0001
E. brasiliensis - L. caudatus	0.022	0.832	0.020	0.885	–0.134	0.183	2.520	0.866
Encotyllabe spari - Lamellodiscus baeri	–0.065	0.542	0.780	0.377	–0.059	0.558	32.312	0.999
E. spari - L. caudatus	0.239*	0.022	1.120	0.291	–0.109	0.278	152.713	0.999
Lamellodiscus baeri - L. caudatus	–0.130	0.220	1.250	0.263	–0.058	0.563	1.757	1

^*significant values.

Table 5. Comparison of prevalence values (χ2) and abundance values (Mann-Whitney U test) for helminth community component species in Pagrus pagrus from the state of Rio de Janeiro, in the study by ^{Paraguassú et al. (2002)}(P1) and in the present study (P2) (significance: P ≤ 0.05). 

Parasites	Prevalence			Abundance
	χ2	P		U	P
Encotyllabe spari	1.063	0.302	---	3653.5	0.0151	P1>P2
Anoplodiscus longivaginatus	2.573	0.109	---	4504	0.870	---
Echinopelma brasiliensis	0.87	0.351	---	5006	0.105	---
Lamellodiscus baeri	1.943	0.163	---	4535.5	0.967	---
Parahemiurus merus	55.841	<0.001	P1>P2	3443	<0.001	P1>P2
Lernanthropus caudatus	18.251	<0.001	P1>P2	120	0.0005	P1>P2
Hysterothylacium sp.	164.476	<0.001	P1>P2	3572.5	0.0142	P1<P2

The analysis of similarity (ANOSIM) was statistically significant (P < 0.001). The NMDS ranking of 190 helminth infracommunities was based on abundance data that had previously been [log (x + 1)] transformed for the parasite species with prevalence ≥ 10% in both studies (Figure 1). The results show a clear similarity in the distribution of metazoan parasite infracommunities between the study by ^{Paraguassú et al. (2002)} and the current study, i.e. the same pattern held after 12 years. Correlation analysis between the abundance of each species of helminths and the axes of the NMDS showed that Encotyllabe spari and Hysterothylacium sp. were responsible for the major differences. This result was also found from the percentage contributions of the species (SIMPER) in which Hysterothylacium sp. accounted for 36.53% and Encotyllabe spari 23.77%, thus contributing to the differences between the studies. The results from the PERMANOVA analysis showed that there was no interaction between year and host length between the study of ^{Paraguassú et al. (2002)} and the current study (Pseudo F 1.0062, P [perm] 0.5377) (Table 6). However the year factor levels differed between the years 2000 and 2012 (Pseudo F 6.0852, P [perm] 0.0003). PERMDISP was used to test the null hypothesis of no differences in dispersion of significant effect in the groups formed in the PERMANOVA analysis. The results from the PERMDISP analysis were not significant (Pseudo F = 0.51618), which confirmed the results from NMDS (Figure 1).

Figure 1. Non-metric multidimensional scaling (NMDS) plot on the first two axes based on the Bray-Curtis similarities of the log (x+1) transformed parasite abundance data (stress = 0.13). Vectors are Spearman correlations of helminth species abundances with the NMDS axes. (The circle represents a correlation of 1). Symbols denote sample membership as follows: □ = year 2000 (data from ^{Paraguassú et al., 2002}); ○ = year 2012 (data from current study). 

Table 6. Effects explaining variation in the abundance of seven parasite species, revealed by permutational multivariate analyses of variance (PERMANOVA) based on Bray-Curtis similarities to the log(x+1) transformed abundance matrix. Only significant effects are listed. Permutational test for homogeneity of multivariate analysis (PERMDISP) was applied to test for differences in dispersion among groups formed by the factors that had significant effects in PERMANOVA. 

	PERMANOVA		PERMDISP
Source	2000-2012		2012-2000
	Pseudo-F	P	Pseudo-F	P(Disp)
Year	6.0852	0.0001	0.51618	NS
Total length	1.0062	0.5297	----	-----
Year x Total length	1.0559	0.3804	----	----

Discussion

In comparing the data from the current study with that of the previous study by ^{Paraguassú et al. (2002)}, we can affirm that the composition and structure of the community of metazoan parasites in this host species has shown temporal stability over the last 12 years. This can be seen from the fact that the same structure was reported in both studies and the dominance of larval stages of anisakid nematodes remained unchanged; from confirmation of the correlation between parasite abundance and total host body length, at the infracommunity level; from the absence of correlation between parasite infrapopulations size and host sex; and from the scarcity of interspecific associations in the parasite infracommunities. The only significant difference was in the abundance of larvae of anisakid nematodes, in relation to the study by ^{Paraguassú et al. (2002)}. In the present study the abundance was doubled for anisakid nematodes.

Pagrus pagrus is a marine teleost fish, with demersal habits, which feeds mainly on crustaceans, fish and mollusks. This condition may explain the dominance of larval stages of anisakid nematodes, since the intermediate hosts in this life cycle are marine invertebrates (crustaceans or mollusks) (^{ANDERSON, 2000}). Anisakid nematodes use fish as second intermediate or paratenic hosts, and the interrelationship between the components of the food chain enables transmission of parasites that have a complex life cycle. The difference in abundance between the two studies (current data and ^{Paraguassú et al. (2002)}) is reflected in the short sampling period in the present study, which focused on the summer. According to ^{Costa et al. (1997)}, the spawning period for red porgy caught in the Cabo Frio region extends from November to January which would explain the differences in prevalence and abundance that coincide with the peak reproductive period of the host, since during this period the fish are more susceptible to parasitism caused, among other factors, by the stress of reproduction (^{LIZAMA et al., 2006}). At this same time of year, there is the effect of the South Atlantic Central Water (SACW) phenomenon, which means that rearing takes place in a nutrient-rich mass, thus resulting in increased food availability throughout the food chain. Changes to the likelihood of infection could be due to the effect of changes in environmental conditions, particularly water temperature, on the life cycle of the parasites, but also to changes in host diet in cases of endoparasites that live in the digestive tract (^{TIMI et al., 2009}).

The higher abundance of larvae of anisakid nematodes in the community of metazoan parasites of P. pagrus may suggest that the habits of this host favors its participation as an intermediate host in the parasite life cycle of this nematode group. This situation has also been recorded in the parasite infracommunities of some benthic marine fish from Rio de Janeiro (^{SILVA et al., 2000}).

In the study of ^{Paraguassú et al. (2002)}, the authors erroneously identified the larvae of Hysterothylacium as larvae of Contracaecum(^{PARAGUASSÚ et al., 2002}). In the current study, we examined specimens collected by ^{Paraguassú et al. (2002)} that were deposited in the Helminthological Collection of Oswaldo Cruz Institute (CHIOC). Larvae identified as Contracaecum by ^{Paraguassú et al. (2002)} (CHIOC No 34437) possess the excretory pore in the region of the nerve ring, which is a characteristic of larvae of Hysterothylacium sp. Thus, the diagnosis of larvae of Hysterothylacium was confused with larvae of Contracaecum sp. by ^{Paraguassú et al. (2002)}. In species of Hysterothylacium, the excretory pore is located in the region of the nerve ring, thus differing from species of Contracaecum, in which it is located near the ventral interlabia (^{DEARDORFF; OVERSTREET, 1981}; KLIMPEL; RUKERT., 2005; ^{KLIMPEL et al., 2007}; ^{KNOFF et al., 2012}; ^{BORGES et al., 2012}). According to Smith and Wooten (1978), although it is difficult to identify species based on morphological characters alone, it is even more difficult in relation to the larval stages. Species identification is especially important for larvae, because they have been implicated as causative agents of human anisakiasis. Anisakis simplex and Pseudoterranova decipiens are the principal species involved in human infection by nematodes of the family Anisakidae (^{ISHIKURA et al., 1993}; ^{ADAMS et al., 1997}).

According to ^{Poulin (1995)}, the potential diversity of parasites may also be associated with a variety of intermediate and definitive hosts. Furthermore, host size is most frequently correlated with the species richness of the community of parasites. Large-sized hosts can consume greater quantities of food than can small-sized hosts, and thus may be exposed to a greater range of infective stages of parasites. These larger hosts can also provide more space for the parasites, and thus harbor more species and offer greater variety of niches for occupation by the parasites. This also allows simultaneous occurrence of more species of parasites.

A positive correlation between parasite abundance and size of host was detected in the populations of red porgy studied by ^{Paraguassú et al. (2002)} and in the current study. As pointed out in the study by ^{Polyanski (1961)}, quantitative and qualitative changes in parasitism are expected with fish growth. The significant positive correlation of mean number of parasites with total host body length indicates that larger hosts harbored greater numbers of parasite individuals and parasite species than did smaller ones. This can be attributed to the facts that larger (i.e. older) fish offer larger target areas for parasites transmitted by cercariae, ingest larger quantities of food, and have had more time to accumulate parasites than have smaller (i.e. younger) fish (^{ZANDER; KESTING, 1998}; ^{FIORILLO; FONT, 1999}; ^{POULIN, 2000}; ^{BUSH et al., 2003}; ^{FELLIS; ESCH, 2004}; ^{ZANDER, 2004}).

According to ^{Saad-Fares and Combes (1992)}, in the case of the digeneans, this correlation might be influenced by changes to the fish diet. In addition, as mentioned by ^{Cezar and Luque (1999)}, the digenean Parahemiurus merus showed a negative correlation between parasite abundance and host body size. This is in agreement with what was observed by ^{Paraguassú (2002)}, who suggested that there was heterogeneity of dietary components and feeding behavior of P. pagrus between different age classes.

Absence of correlation between host sex and the prevalence and abundance of parasite community components of marine fish is common. In P. pagrus, the lack of such correlation might be attributed to similarity in ecological relationships (behavior, habitat and diet) between males and females, as stated by ^{Luque et al. (1996)}. According to ^{Poulin (1996)}, the influence of host sex on the prevalence and abundance of parasites is a topic hardly touched upon in community analysis discussions, and it is necessary to conduct experiments to show the influence of other factors, particularly fish physiology and behavior.

Comparison of the prevalence and abundance of common helminth species in P. pagrus, in the study by ^{Paraguassú et al. (2002)} and in the current study, revealed significant differences relating to the following species: Parahemiurus merus, Hysterothylacium sp., Lernanthropus caudatus and Encotyllabe spari. The qualitative similarity of the infracommunities of ectoparasites in P. pagrus from the municipality of Cabo Frio was shown at the genus level, with a complex of ectoparasite species (diplectanids, microcotylids, anoplodiscids, lernanthropids and caligids) that is a characteristic of other species of sparid fish. Lamellodiscus (Monogenea: Diplectanidae) has mainly been studied in sparids. Sparid fish are widely distributed in tropical and temperate coastal waters (^{NELSON, 2006}). Diplectanids often show strict host-specificity (^{OLIVER, 1992}).

All the analyses confirmed and maintained the data of the study of ^{Paraguassú et al. (2002)}, confirming the homogeneity of the structure and composition of the parasite community of P. pagrus. Variations in environmental conditions and their influence on the distribution of organisms that are part of the diet of red Porgy, involving the life cycle of parasites, can be considered to be important for observing the differences in abundance of some parasites between the current study and the study of ^{Paraguassú et al. (2002)}.