INTRODUCTION
Spatial ecology is one of the great advances of modern population and community ecology, which has highlighted the importance of the spatial scale for understanding a wide range of ecological phenomena (Holt, Reference Holt, Bell, Brylinsky and Johnson-Green2000). Currently, some parasite ecologists have attempted to explain the parasite distributional patterns (abundances and species richness) on various spatial and temporal scales through different ecological models derived from those developed for free-living organisms (e.g. Rohde et al. Reference Rohde, Worthen, Heap, Hugueny and Guégan1998; Gotelli and Rohde, Reference Gotelli and Rohde2002; Poulin, Reference Poulin2004). However, there is no consensus regarding whether there are general patterns in parasite communities (Poulin, Reference Poulin2007a). To date, most efforts in parasite community studies have been focused on determining distributional patterns in the abundances or species richness (references in Poulin, Reference Poulin2007a, Reference Poulinb). Early predictions were based on the theory of island biogeography (Kuris et al. Reference Kuris, Blaustein and Alió1980) and, more recently, empirical tests show that epidemiological processes may be important as a determinant of local parasite species richness (Morand and Poulin, Reference Morand and Poulin1998). In contrast, patterns in community composition (taxonomic identity) have been by far less analysed. Presence/absence matrices of sites versus species records, with species being present or absent at each site, are commonly recorded and potentially give valuable distributional information about species, communities, and environments (Wright et al. Reference Wright, Patterson, Mikkelson, Cutler and Atmar1998).
Nested structure is a pattern originally described in island biogeography to characterize how a set of species is distributed among a set of islands (Patterson and Atmar, Reference Patterson and Atmar1986). A nested pattern has been defined as a departure from a random association of species in which species that compose a depauperate island community constitute a proper subset of those species inhabiting richer islands (Atmar and Patterson, Reference Atmar and Patterson1993). In order to detect community structure, this pattern has also been investigated in other ecological systems such as mountains, ponds, streams, and fragmented forest patches (references in Fischer and Lindermayer, Reference Fischer and Lindenmayer2005). Nestedness has also been recorded for parasite species among conspecific host individuals (=infracommunities) from a locality, but these results differ according to the host species studied (Rohde et al. Reference Rohde, Worthen, Heap, Hugueny and Guégan1998; Poulin, Reference Poulin2007a), which could be a consequence of the different methods applied to evaluate nestedness in those host-parasite systems (Timi and Poulin, Reference Timi and Poulin2008). Individual hosts represent replicated habitats in time and space, allowing consideration of them as a unit of study in the context of island biogeography theory (Kuris, Reference Kuris, Blaustein and Alió1980). However, at this local scale (individual hosts), there are several other non-biogeographical factors (i.e. fish size range, differential food consumption rate, differential susceptibility, etc) than can cause nestedness (Morand et al. Reference Morand, Rohde and Hayward2002). The geographical distributional patterns of parasites are directly associated with host distributional range, which are influenced at the same time by biogeographical processes such as dispersal abilities of parasites and hosts, and by prey geographical distributions or prey availabilities (intermediate hosts), which is in agreement with the original idea of the nested subset patterns developed by Patterson and Atmar (Reference Patterson and Atmar1986). Thus, nested patterns in parasite communities should be sought on a larger geographical scale, that is, across different localities or host populations (=component communities sensuBush et al. Reference Bush, Lafferty, Lotz and Shostak1997). The nestedness patterns associated with distributional range in parasites of rodents have been investigated (Goüy de Bellocq et al. Reference Goüy de Bellocq, Sarà, Casanova, Feliu and Morand2003; Krasnov et al. Reference Krasnov, Shenbrot, Khokhlova and Poulin2005). However, to date, there is only a single study that analyses the nestedness in parasite communities of the same marine fish host across its geographical range (González and Poulin, Reference González and Poulin2005). Therefore, more investigations are necessary to determine if the nestedness pattern of parasite assemblages of the same host species could be a pattern associated with the geographical distributional range of the host.
The distributional patterns of marine parasites may be determined by oceanographic characteristics (temperature, depth, specific water masses) and also by factors associated with the hosts (density, feeding habits and migratory patterns) (Bush et al. Reference Bush, Fernández, Esch and Seed2001). Along the Chilean coast, 2 faunistic provinces are generally recognized: the Peruvian faunistic province, extending from Peru to the northern Chilean coast (ca. 30°S), and the Magellanic faunistic province, extending southward of 42°S along the southern Chilean coast. Between both areas lies a transitional zone where species of northern and southern origin overlap (Briggs, Reference Briggs1974; Lancellotti and Vásquez, Reference Lancellotti and Vásquez1999). Therefore, it is expected that parasite assemblages show nested structure across host populations of fish species with an extensive geographical distributional range. Nestedness in those parasite communities could be produced by environmental factors (which could limit the survival of ectoparasites), and by changes in the prey availability (intermediate hosts) along their distributional range. Thus, the aim of this study was to determine whether assemblages of ectoparasites and endoparasites – analysed separately – of several marine fishes with an extended geographical distribution in the south-eastern Pacific show a nested structure, which could be associated with extrinsic factors associated to host species distributions.
MATERIALS AND METHODS
Specimens of Pinguipes chilensis, Prolatilus jugularis, Scomber japonicus, Nezumia pulchella and Hippoglossina macrops were captured from different latitudes along the southeastern Pacific coast from March to June 2006 and from February to May 2007 (Fig. 1). The first 3 fish species were collected either by hand line, speared by divers, or they were acquired from local fishermen. The 2 deep-sea fish species were captured as by-catch from the shrimp fishery. The samples were captured in the same period of year to avoid seasonal influence on the composition of parasites in the fish hosts. Additionally, we used our own data base for parasites of Sebastes capensis, Trachurus symmetricus, Engraulis ringens and Merluccius gayi. Sampled latitudes, sample sizes, and host habitat are given in Table 1.
Fig. 1. Approximate position of sampled localities. Code for locality: 1=Paita, 2=Callao, 3=Arica, 4=Iquique, 5=Antofagasta, 6=Caldera, 7=Huasco, 8=Coquimbo, 9=El Teniente, 10=Valparaíso, 11=Bucanero, 12=Talcahuano, 13=Valdivia, 14=Puerto Montt, 15=Aysén, 16=Punta Arenas.
Table 1. Ectoparasite (left value) and endoparasite (right value) species richness of the component communities from the different fish species
(In parenthesis, sample size. Habitat/habits: a Benthic/non-migratory; b demersal/non-migratory; c demersal/migratory; d pelagic/migratory; e pelagic/non-migratory; * no data.)
The total length (TL) of each fish was measured (to ±1 cm) prior to dissections. Ectoparasites and endoparasites were collected using standard parasitological techniques outlined by González and Acuña (Reference González and Acuña1998). The collected parasites were sorted, counted, and preserved in 70% ethanol for identification. The specialized literature was used to identify each parasite species.
For each parasite species, prevalence (=number of fish infected with a particular parasite species divided by the number of fish examined) was estimated according to Bush et al. (Reference Bush, Lafferty, Lotz and Shostak1997). Parasite richness (number of ectoparasite or endoparasite species present in each component community sensuBush et al. Reference Bush, Lafferty, Lotz and Shostak1997) was calculated for each locality. Because sampling effort and host sizes may exert a strong bias in parasite species richness estimation across different localities (Poulin, Reference Poulin2007b), Spearman correlation matrices were used to evaluate the relationships among sample size, mean host size, latitude and species richness for each fish host species (Zar, Reference Zar1999). Subsequently, meta-analyses for correlation data using fixed and randomized effects models were performed by Comprehensive Meta-Analysis (CMA) program (www.Meta-Analysis.com).
Nested subset analyses were carried out separately for ecto- and endoparasite component communities of each host species. First, the analyses were performed considering all recorded parasites, and then including only the parasite species with prevalence >5% in at least 1 locality. Patterns of nestedness using the software Nestedness were evaluated (Ulrich, Reference Ulrich2006). This program includes 6 null models, and different nestedness indices. The FF (fixed row and fixed column) null model has been demonstrated to be the most conservative; and among the indices, the Brualdi and Sanderson discrepancy index (BR) of unexpected presences was the best in the performance, according to Ulrich and Gotelli (Reference Ulrich and Gotelli2007a, Reference Ulrich and Gotellib). Then the BR index was used because data-base matrix properties did not influence it. The FF algorithm is not recommended for matrices that are ‘evidently’ nested. In these matrices, the Z score value is zero because there is no analysis performed. So, the FF algorithm cannot be used in a perfectly nested matrix because there are no possible matrix rearrangements to maintain fixed row and column addition (Ulrich and Gotelli, Reference Ulrich and Gotelli2007a). Since different null models may give different results, 5 null models for presence-absence data: FF (fixed row and fixed column totals), FE (fixed row totals, equiprobable column totals), EF (equiprobable row totals, fixed column totals), EE (equiprobable row totals and equiprobable column totals), PE (proportional row totals, equiprobable column) were used. To compute the null models, we used default values for randomizations (=100), cell minimum distance to the border line (=0·5), and matrices were randomized by species richness. Additionally, we evaluated nestedness using the option ‘unsorted matrix’, which is helpful in studies of gradients that might influence the degree of nestedness (Ulrich, Reference Ulrich2006). However, we did not find significant differences using either option. Nestedness significance levels were obtained from Z-scores, and lower and upper 95% confidence limits of the respective null model distributions (Ulrich, Reference Ulrich2006; Ulrich and Gotelli, Reference Ulrich and Gotelli2007a).
RESULTS
The fish species studied harboured between 3 and 12 ectoparasites and between 3 and 29 endoparasite species (Table 1). Engraulis ringens harboured only 2 ecto- and 2 endoparasite species with prevalences higher than 5%. Similarly, Merluccius gayi harboured only 3 ectoparasite species with prevalence >5%, and Trachurus symmetricus harboured only 4 ectoparasite species, 2 of them present in all localities and the other 2 present only in 1 locality (Table 1). Therefore, as those hosts harboured few parasites, nestedness was not evaluated for ectoparasites of E. ringens, M. gayi and T. symmetricus, and for endoparasites of E. ringens. In the remaining fish hosts, the ectoparasite richness did not show common latitudinal gradients (Table 1; Figs 2 and 3). Among conspecific populations, ectoparasite richness decreased with latitude in Nezumia pulchella (Fig. 2C), but richness increased with latitude in Hippoglossina macrops (Fig. 2D). Similarly, the endoparasite richness increased with latitude in Sebastes capensis (Fig. 3B), but it decreased with latitude in N. pulchella (Fig. 3C). Meta-analyses of correlations showed that both ecto- and endoparasite species richness were not significantly correlated with sample size by locality, mean fish sizes or latitude (P>0·05 for all correlations; Table 2). However, the heterogeneity tests indicated that the studied hosts do not share a common effect size (Table 2).
Fig. 2. Matrix presence (dark square) and absence (white square) of the ectoparasites recovered across the host latitudinal range. (A) Pinguipes chilensis. a: Lepeophtheirus mugiloides; b: Caligus cheilodactylus, c: Paramicrocotyle sp. d: Neobenedenia sp.; e: Chalguacotyle sp., f: Gnathia sp., g: Piscicolidae gen sp., h: Cirolana sp.; i: Cimothoa sp.; j: Rocinela sp., k: Udonella australis. (B) Sebastes capensis. a: Caligus cheilodactylus; b: Microcotyle sp.1; c: Lepeophtheirus chilensis; d: Gnathia sp.; e: Interniloculus chilensis; f: Neobenedenia sp.; g: Udonella australis; h: Piscicolidae gen. sp.; i: Cirolana sp.; j:Microcotyle sp.2; k: Rocinela sp. (C) Nezumia pulchella. a: Jusheyhoea macroura; b: Diclidophora sp.; c: Clavella sp1; d: Clavella sp2; e: Lophoura sp. (D) Hippoglossina macrops. a: Holobomolochus chilensis; b: Protochondria longicauda; c: Neoheterobothrium chilensis; d: Glyptonobdella sp.; e: Entobdella sp. (E) Scomber japonicus. a: Kuhnia sprostonae: b: K. scombri; c: Ceratothoa sp.; d: Rocinela sp.; e: Clavella sp.; f: Caligus bonito.
Fig. 3. Matrix presence (dark square) and absence (white square) of the endoparasites recovered across the host latitudinal range. * Species with prevalence <5%. (A) Prolatilus jugularis. a: C. australe; b: Neoleburia georgenascimentoi; c) Phyllodistomum sp.; d: Anisakis sp.; e: Cucullanus sp.; f) Dichelyne sp.; g: Echinorhynchidae gen. sp., h: Phillometra; i: Aporocotyle sp.; j: Lecithastheridae gen. sp.; k: Ascarophis sp., l: Nybelinea sp.; m: Contracaecum sp.; n: Lecithochirium sp.; o: Hysterothylacium aduncum. (B) Sebastes capensis. a: Ascarophis cf. sebastodis; b: Anisakis sp., c: Corynosoma australe; d: Pseudopecoelus sp.; e: Psettarium sp.; f: Hysherothylacium sp.; g: Cucullanus sp.; h: Lecithochirium genypteri; i: Helicometrina nimia; j: Zoogonidae gen. sp; k: Hemiuridae gen. sp.; l: Scolex pleuronectis; m, n, o: Lecithastheridae spp.; p, q: Digenea spp. (C) Nezumia pulchella. a: Lepidapedon sp.; b: Anisakis sp.; c: Bucephalidae gen. sp.; d: C. australe; e: Hemiuridae gen. sp.; f: Contracaecum sp.; g: Capillaria sp.; h: H. aduncum; i: Proleptus sp.; j: Cystidicolidae gen sp. (D) Hippoglossina macrops. a: C. australe; b:Neobothriocephalus aspinosus; c: Floridosentis sp.; d: Anisakis sp.; e: Nybelinea sp.; f: Scolex pleuronectis; g*: Bolbosoma sp.; h*: Philometra sp.; i*: Lecithochirium sp1; j*: Hemiuridae gen. sp.; k*: Arhythmorhynchus sp.; l*: Lecithochirium sp2; m*: Lecitophyllum sp. (E) Trachurus symmetricus- a: Radinorhynchus trichiuri; b: A. simplex; c: A. physeteri; d: Hysterothylacium sp.; e: Larva Anisakidae; f: Tentacularia coryphaenae; g: Nybelinea sp.; h: Anisakidae sp.; i: Scolex pleuronectis; j: Eutetrarhynchus sp.
Table 2. Results of meta-analyses for correlations data
(Test for effect size and 95% confidence intervals; Z value (=Fisher' Z/Standard Error) and P value associated to respective null hypotheses. Also given are test of heterogeneity, Q-values, degree of freedom (d.f.), and P values.)
Nested pattern detection was influenced by the null model. According to the FF null model, none of the analysed ectoparasite assemblages showed a nested structure across the latitudinal range of their hosts (Table 3), but the FF algorithm was not useful to evaluate nestedness for ectoparasites of N. pulchella and H. macrops. In the other extreme, according to the EF and EE null models only Prolatilus jugularis and Scomber japonicus ectoparasites were not nested (Table 3; Fig. 1). However, according to the PE null model only the ectoparasites of these two last hosts were nested. For ectoparasites, nested analyses results were not influenced by parasite inclusion or exclusion with a prevalence of less than 5%.
Table 3. Summary of nestedness analyses for ectoparasite assemblages of analysed marine fishes from the Southeastern Pacific Coast, using five null models and BR index (Ulrich, Reference Ulrich2006)
(Given are Z-scores of BR-index for each null model, number of parasite species (spp), sites, and matrix fill (Fill) Significances are marked in bold.)
According to the FF null model, among endoparasite assemblages only T. symmetricus showed a nested structure (Table 4). However, nested analyses based on EF and EE null models (including all endoparasites) showed that only Pinguipes chilensis and P. jugularis were not nested through their host latitudinal range (Table 4; Fig. 3). In P. jugularis, H. macrops and S. japonicus, parasite exclusion with a prevalence <5% influenced the results of nestedness in different ways (Table 4).
Table 4. Summary of nestedness analyses for endoparasite assemblages of analysed marine fishes from the Southeastern Pacific Coast, using five null models and BR index (Ulrich, Reference Ulrich2006)
(Given Z-scores of BR-index for each null model, number of parasite species (spp), sites, matrix fill (Fill). Significances are marked in bold (−2·0>Z-scores >2·0).)
DISCUSSION
González and Poulin (Reference González and Poulin2005) have demonstrated that parasite communities of a marine fish (S. capensis) extensively distributed along the Southeastern Pacific coast (more than 44 degrees of latitude) showed significant nested patterns. However, the type of nestedness pattern differed between ectoparasites and endoparasites. In our study, migratory fish species (M. gayi, T. symmetricus, and E. ringens) harboured few ectoparasite species, which were present in the different sampled latitudes, whereas non-migratory fish species (S. capensis, P. chilensis, P. jugularis, H. macrops and N. pulchella) showed ectoparasite richness gradients along the distributional range of the hosts. However, gradients in ectoparasite richness were not correlated with latitude. On the other hand, the endoparasite richness of S. capensis increased with latitude, but the parasite richness of N. pulchella decreased. Therefore, ecto- and endoparasite assemblages of marine fishes from the Southeastern Pacific coast do not show general latitudinal patterns in parasite richness as has been previously reported (Rohde, Reference Rohde and Rohde2005).
The detection of nestedness in binary presence-absence matrices is affected by both the metric used to quantify nestedness and the reference null model distribution. According to Ulrich and Gotelli (Reference Ulrich and Gotelli2007a, Reference Ulrich and Gotellib), the best performing algorithm maintains fixed row and fixed column totals, but it is conservative and may not always detect nestedness when it is present. Therefore, when one matrix shows an evident nested pattern it is better to use an alternative null model such as EE, FE or EF (see Ulrich and Gotelli, Reference Ulrich and Gotelli2007a for explanations about algorithms). Likewise, among the metrics of nestedness, the Brualdi and Sanderson discrepancy index (BR) performs better because it is less affected by matrix properties (that is, shape, size, fill and richness differences), which was confirmed in our data base. Taking into account the analysed matrices characteristics of this study and null model election, ectoparasite assemblage nestedness frequency in analysed marine fishes could be between 33 and 66%, whereas nested structure detection in endoparasite assemblages through the distributional range of their host could vary between 25 and 75%.
Several studies have emphasized the importance of host geographical distribution on the patterns of parasite species richness (Poulin, Reference Poulin2007b). Thus, the sampled extension of the distributional range of the hosts as well as the lack of specificity of some parasites could influence their parasite species richness and the nestedness structure (González and Poulin, Reference González and Poulin2005; González and Oliva, Reference González and Oliva2006). Three fish species, S. capensis, P. chilensis and P. jugularis, inhabit shallow waters, and are distributed approximately between 11°S and 50°S in the Southeastern Pacific coast (Pequeño, Reference Pequeño1989). In these 3 host species the ectoparasite assemblages are richest at latitudes associated to the transitional area (30°S–40°S). These fish species share some parasites such as the cirolanid isopods, and Gnathia sp., the copepod Caligus cheilodactylus, the monogeneans Udonella australe and Neobenedenia sp. Thus, it is possible that the presence of these parasite species in intermediate latitudes could be influenced by the presence (or abundance) of some of these 3 host fish species (González and Poulin, Reference González and Poulin2005). On the other hand, parasites of the deep-water fishes N. pulchella and H. macrops, are most host-specific, and there are no shared parasite species amongst them. N. pulchella is distributed between 7°S to 34°S (Pequeño, Reference Pequeño1989), but we sampled approximately from the centre toward the southern limit of its distributional range only (24°S–33°S). In this host species, many parasite species were lost from their central distributional range toward southern latitudes. The distribution of H. macrops is not well known, but may extend from Perú to 47°S on the Chilean coast (Ojeda et al. Reference Ojeda, Labra and Muñoz2000). Although we sampled a limited portion of their geographical range (25°S to 37°S), the loss of 2 ectoparasite species was observed northward of 28°S. Therefore, those non-migratory host fishes share a similar characteristic, that is, their ectoparasite assemblages are richest in host populations located in the central geographical distribution of the host species. This pattern might be concordant with the ‘abundant centre’ distribution rule (Sagarin and Gaines, Reference Sagarin and Gaines2002) because the higher ectoparasite species richness in central populations of marine fishes from the Southeastern-Pacific could reflect the optimal environmental conditions at the distributional centre of these host species. High abundances (or densities) of host species would facilitate the transmission rates of parasite species, which could cause nested patterns. Additionally, host specificity might determine whether a parasite is able to colonize a host and may be a factor that could influence nestedness structure in parasite communities (Matejusová et al. Reference Matejusová, Morand and Gelnar2000). However, it is possible that generalist parasites could be distributed evenly among ranges of host species too, producing predictable parasite communities within and between these hosts (González and Oliva, Reference González and Oliva2006).
Different oceanographic conditions are present along the southeastern Pacific coast (Escribano and Hidalgo, Reference Escribano and Hidalgo2001; Silva and Calvete, Reference Silva and Calvete2002). These differences do not interrupt the distributional range of the 3 benthic fish species (S. capensis, P. jugularis and P. chilensis); however, the distributions of a few ectoparasite species (i.e., Piscicolidae spp., U. australe, Cirolana sp.) seem to be restricted at some localities from the transitional area. The dispersion of those host-specific ectoparasites can be considered to be entirely dependent on their particular hosts, because they have direct life cycles and a minimal opportunity to disperse to new regions during their typically brief free-living stage (Hayward, Reference Hayward1997). Then, the loss of these parasite species from their hosts could be explained by environmental characteristics associated with biogeograpical areas northward and southward of the transitional area, which could limit whether the dispersion of the infested host fish populations or survival of different stages of the life cycle of some ectoparasites. A similar pattern was suggested by Krasnov et al. (Reference Krasnov, Shenbrot, Khokhlova and Poulin2005), but in other host-parasite systems (Rodentia and Insectivora- fleas). In brief, the possible gradual loss of ectoparasites in small mammals is explained by the expansion of the host distributional range.
Several mechanisms have been suggested to cause nestedness in assemblages of free-living organisms, the most frequently cited are: selective extinction, selective immigration, nested habitats and passive sampling (Lomolino, Reference Lomolino1996). Recently, other mechanisms have been tested through ecologically explicit null models of nestedness (Moore and Swihart, Reference Moore and Swihart2007). According to Rohde et al. (Reference Rohde, Worthen, Heap, Hugueny and Guégan1998) differential colonization probabilities are the most likely cause of nestedness in parasite assemblages of marine fishes. Since the life cycle and the mechanisms of transmission of ecto- and endoparasites are different, the nested patterns (and/or richness latitudinal gradient) for those two groups of parasites could result from different processes affecting the parasite's colonization (González and Poulin, Reference González and Poulin2005; Guégan et al. Reference Guégan, Morand, Poulin, Thomas, Renaud and Guégan2005). In the present study, we observed latitudinal changes in the endoparasite richness of populations of marine fishes distributed along the southeastern Pacific coast. For instance, the endoparasite richness of N. pulchella, P. chilensis and P. jugularis tend to be higher in northern latitudes, whereas in H. macrops and S. capensis the endoparasite richness increased in southern latitudes. These latitudinal gradients in the endoparasite richness of the hosts might be explained by zoogeographical breaks of prey and/or by changes in prey availabilities (intermediate hosts) along host latitudinal range (González et al. Reference González, Barrientos and Moreno2006).
This study was funded by a Postdoctoral FONDECYT Project N°3060054 granted to the principal author. The authors thank Professor R. Poulin of the University of Otago and two anonymous referees for their critical comments on an early version of this manuscript.
