DISCUSSION

Are patterns observed in the structure of a parasite community at one time and place representative of other parasite communities in different populations of the same host species? Our results say no. Instead, they suggest that even when departures from random assembly expressed as positive associations between the most prevalent metazoan parasite species or nested patterns occurred, no repeatability in parasite community structure existed in time or space for any of the data sets. There were, however, several interesting results: the influence of fish size differences within each locality; the presence of positive correlations in most of the pairwise comparisons; the nature of the subgroups exhibiting nestedness, and the influence of distance between localities on the likelihood of finding nestedness.

One of the most influential variables in this study was fish size, since those localities with nestedness in any subgroup also had a significant correlation between fish size and the logarithm of the total number of metazoan parasites per fish. This was true for the red grouper and mojarra spatial data, as well as for the mojarra temporal data in Celestun, and in MITZA for both 1990 and 1999. For both the red grouper and mojarra it is probable that there were differences in parasite intake that produced significant correlations, meaning that the larger the fish, the higher the number of parasites accumulated. This is to be expected since similar patterns have recently been described by Poulin (2000) in a meta-analysis of 76 datasets on the relationship between fish length and intensity of infection, and by Poulin & Valtonen (2001) for the internal parasite communities of 23 populations of fish from Finland. These latter authors also found that nested patterns occur in parasite communities whose fish hosts accumulate parasites with increases in size. In view of these facts, it is possible that perhaps we have been sampling the wrong size of fish, and that we should study only large (=adult) fish within and between localities to obtain meaningful comparisons. However, limiting the analyses to a narrow size range of hosts would also result in small sample sizes and reduced power. Despite the limitations of the nestedness analyses due to variation in host sizes, it is clear that the patterns uncovered by this approach are not repeatable in time or space.

It is quite unusual that we found no anti-nested patterns (sensuPoulin & Guegan, 2000) in this study. This is likely a result of both the red grouper and mojarra being non-specialist feeders, which allows them to sample most of the pool of locally available parasite species. Similar explanations have been suggested by Poulin & Valtonen (2001) for fish from Finland.

By including data on total communities, we found positive pairwise associations between parasites belonging to different subgroups. Thus, even when a monogenean reaches a high number of individuals in red grouper, such as P. yucatanensis in the gills, it caused no interference in the number of individuals of H. serrana, even in the same habitat. The same pattern applies for P. yucatanensis and Ph. salgadoi from the gonads, and for all the adult and larval helminth species parasitizing mojarra in both spatial and temporal data sets. The positive correlations found in red grouper and mojarra were between adult parasites that were host specialists, which suggests that it is only a matter of coincidence that these parasites are using the same kind of definitive host. Clearly, these species are reaching the host through their own individual specific transmission routes, and appeared to exhibit individualistic responses with respect to microhabitat selection, as suggested by Vidal-Martínez et al. (1998). In the case of the larval helminth species of the mojarra, the positive correlations can be explained by several species using the same intermediate host, the widely distributed snail Pyrgophorus coronatus (Scholz & Aguirre-Macedo, 2000). In this sense, our results support the idea that some of the larval helminths can be acquired by fish as packets, when the mojarra eats a snail harbouring several of these species. That is the case of Echinochasmus leopoldinae, Crassicutis cichlasomae, Oligogonotylus manteri, and Phagicola sp. However, the other larval species Contracaecum and Pelaezia loossi seem to reach the host because they are abundant enough in the locality. In this sense they could be seen as a result of a random selection of the locally available species. Clearly, to assess the agreement between our results and those of Poulin & Valtonen (2001) for larval stages of metazoan parasites of fish from Finland, it will be necessary to undertake a similar analysis to that of these authors.

Spurious correlations are the most likely explanation for the negative significant correlations found between Ph. margolisi (from gonads) and H. serrana (from gills) in red grouper, and S. mexicanum (gills) and both O. manteri metacercariae (encysted in anterior intestine) and adults (from rectum) in Mexican mojarra. Therefore, our results suggest that interspecific interactions are not important as a determinant of species composition for the metazoan parasite communities of red grouper and mojarra. Clearly, negative interactions between metazoan parasites from different microhabitats in a host are extremely unlikely. However, it does not mean that there are not interspecific interactions in tropical fish, since Vidal-Martínez & Kennedy (1998) have demonstrated that such interactions can occur among intestinal species of helminths of Cichlasoma synspilum in south-eastern Mexico.

The nature of the metazoan parasite subgroups which exhibited nestedness was not relevant to the presence of this departure from randomness. We might expect that parasite subgroups, such as that of internal parasites formed by digenean metacercarial stages and intestinal parasites, will tend to accumulate as the fish age, thus leading to a nested pattern. This was not true, especially in mojarra, where there was no constancy in the subgroups producing nestedness. The most likely explanation for the spatial data is that parasite species composition is slightly different among localities, which in turn produces a random assignment of the species forming the regional pool of metazoan parasites to specific localities. This seems to be true, even for larval stages of digeneans. This view seems to be opposite to what has been found in southeastern Mexico with respect to larval digeneans, since they are quite frequent and abundant between and within localities (Salgado-Maldonado & Kennedy, 1997; Vidal-Martínez et al. 2001). However, the lack of nestedness indicates that the same species of larval digeneans were not present in all the localities or all the individual hosts in one locality, which in turn produces differences in species composition. An additional factor here is the fish size as indicated above. In the case of luminal intestinal parasites, adult metazoan parasites may not be available at a specific locality, which in turn also produces differences in species composition and then the lack of repeatability. Overall, these results suggest that parasite community structure is fickle, influenced by the presence or absence of one or very few species, and that it may be hopeless to search for repeatable patterns.

Distance among localities for spatial analyses of repeatability and sampling date for temporal analyses seem to be an additionally important factor affecting species composition and thus nestedness and repeatability. In the specific case of the red grouper, there was nestedness in 3 of the localities examined: Chelem, Chuburna, and Progreso. This departure from randomness in the last two localities was between the subgroups including all the metazoan parasite species, while nestedness for Chelem was only for intestinal parasites. Nevertheless, the closeness of the localities (<40 km from each other) seems to be influencing the presence of nestedness. Clearly, in the case of the red grouper, the continuity of the sea provides a constant connection between the localities. As the localities are close to each other, fish are exposed to the same local pool of parasites, and therefore similar nested patterns (i.e. repeatability) are possible. In the case of mojarra, nestedness was present also among close localities (MITZA, Rio Lagartos) for internal parasites, and Celestun for intestinal ones, all of them in the Peninsula of Yucatan. There was also nestedness in intestinal worms from Atasta, a locality far away from those of the Peninsula. However, there is a large geographical distance between the closest locality of the Yucatan Peninsula (i.e. Celestun) and Atasta (362 km). Therefore, it is possible that the reason there was nestedness in the 4 localities is because they have a different biogeographical history, a condition sufficient and necessary to have nestedness (Kodric-Brown & Brown, 1993). However, the point is that geographical distance seems to be an important factor determining the repeatability of nestedness patterns. Similar conclusions regarding the influence of geographical distance upon the similarity of parasite species composition between localities have been reached by Poulin & Morand (1999). The extent to which geographical distance is affecting repeatability can only be determined by producing databases taking into account this variable, and determining whether or not there exists spatial structure in the data (sensuBurrough & McDonnell, 1997).

In the case of the temporal data, there was a rough tendency for nestedness to be present from October to February, even when this departure from randomness was present for different subgroups in Celestun. Still, the presence of nestedness seems to be associated with the post-reproductive season of C. urophthalmus, i.e. the time during which the mojarra increases dramatically its food intake after a long nesting time (Martínez-Palacios et al. 1993). A similar explanation for the presence of maximum values of average abundance of O. manteri, C. cichlasomae and S. mexicanum, has recently been suggested by Jiménez-García et al. (2002). In the case of MITZA, there was no clear pattern either in the presence of nestedness or in the subgroups in which it was present. The main reason for this non-repeatability seems to be related to the fact that MITZA is an artificial, and relatively young (25 years) water body, which has been colonized only recently by both mojarra and their parasites (Vidal-Martínez et al. 1998). Therefore, we might expect that, due to its recent origin, the transmission dynamics of the parasites are not fully established, a factor which affects nestedness and its repeatability. However, as in the case of the spatial data, it would be extremely useful to determine whether or not there is temporal structure in the data.

In conclusion, nested patterns in community structure and their repeatability for the spatial data on red grouper and mojarra seem to be deeply affected by geographical distance between localities, and by fish size within each locality. In the case of the temporal data, at least in one locality the presence of nestedness and its repeatability appear related to the reproductive biology of the host. Therefore, there is a need for investigations into whether or not there is spatial and temporal structure in parasite community structure. Our results make it clear that one cannot extrapolate the results from one community to those of other parasite communities in different populations of the same host species.