DISCUSSION

Microphallid trematodes are a troublesome system for a study of patterns of host-parasite interactions since they often co-occur in the intermediate host and are transferred to the final hosts (birds and mammals) in packets. Furthermore, geographical variation in crustacean-microphallid associations apparently exists (e.g. Helluy, 1983; Mouritsen and Jensen, 1997; Zander et al. 2002) and our new data (see Kostadinova and Mavrodieva, 2005) provide yet another example of the effect of environmental conditions on microphallid infections in amphipod hosts. Our study on the host defensive melanization response to multispecies infections is the first to supply data on virtually the same host-parasite system as that studied by Thomas et al. (2000), and thus provides a unique opportunity for a comparative assessment of the interactions (if any) between local conditions and the manifestation of the amphipod's capacity to defend itself in a natural setting.

Immune defence is regarded as a subcomponent of survival, since its primary function is to ensure survival in an environment rich in parasites (Rigby and Moret, 2000). Immune responses in arthropods have been measured typically by means of intensity and rates of melanotic encapsulation (König and Schmid-Hempel, 1995), activity of the phenoloxidase cascade (Siva-Jothy et al. 2001; Rolff and Siva-Jothy, 2002; Rigaud and Moret, 2003) or the inducible production of antibacterial peptides (Moret and Schmid-Hempel, 2000). It is assumed that these measures reflect the ability of the host to control or eliminate an infection and so their links to fitness are highly plausible (see Schmid-Hempel and Ebert, 2003 for a recent review). For technical reasons we used the first measure as it appears better suited for screening large samples of natural populations.

The main finding of our study is that G. insensibilis exhibited the ability to melanize the metacercariae of all four microphallid species infecting this host at Pomorie lagoon. Both the overall frequency of melanization and the mean abundance of melanized metacercariae in the sample from Pomorie lagoon showed substantially higher values than those observed in ‘Sample 1’ at Thau (99·3% vs 35·6% and 3·22 vs 0·11, respectively). We believe that these differences are directly related to the significantly higher infection rates of microphallids in Pomorie lagoon. Our observations therefore indicate that amphipods invest more in defence and this can be attributed to the specific characteristics of this environment favouring microphallid infection. However, although the overall frequency of melanization approached the maximum, the total rate of melanization was rather low and this pattern was replicated in the sample stratified by size. Thus, although more than half of each size-class sample harboured melanized metacercariae, their abundance was rather low and they represented a very small proportion of the total. Furthermore, the overall rate of melanization showed a significant decrease with amphipod size and the associated higher parasite load in the absence of substantial differences in abundance of melanized metacercariae across size groups. These results illustrate an overall successful parasitism of microphallid larvae in their intermediate host at Pomorie lagoon.

Although studies on invertebrate host immune response against metazoan parasites are in their infancy, evidence is accumulating that these parasites may circumvent their hosts' immune defence (Rigaud and Moret, 2003 and references therein). Thus, infection of Gammarus pulex with acanthocephalans was found to be associated with lower levels of PO-enzyme activity (Plaistow et al. 2003; Rigaud and Moret, 2003). Inferring from these studies it appears plausible that microphallids also evade their host response through a similar mechanism of immunosuppression. However, the different mode of infection in amphipod-microphallid systems rather suggests a different effect on host defensive response. Microphallid cercariae infect amphipods by direct penetration of the cuticle thus inducing an immediate clotting and melanization response at the penetration holes, sometimes trapping penetrating cercariae (Jensen et al. 1998). In invertebrates, wounding induces an immune response, similar to that used to encapsulate parasites and pathogens and a recent study on G. pulex revealed a negative correlation between wounding abundance and PO activity (see Plaistow et al. 2003 and references therein).

The massive invasion by cercariae during a short period, possibly a common infection pattern in microphallids (Ginetsinskaya, 1988; see Fredensborg et al. 2004 for a discussion), therefore may result in significant loss of haemolymph (thus lowering PO activity levels) and/or monopolization of the defence resources for wound healing (Siva-Jothy et al. 2001). Both outcomes would lower melanization efficiency and favour successful parasitism of the metacercariae already established in the host. The overall low levels coupled with the contrasted rates of melanization in the youngest vs oldest individuals observed in our sample, tend to support this suggestion.

Thomas et al. (2000) found that only metacercariae of M. papillorobustus were melanized in samples of G. aequicauda collected ‘during spring 1999’ and ‘one month later’ in ‘Palavas les Flots’, Southern France (coordinates point to Thau lagoon) and speculated that the cellular defence reaction of the host is specifically targeted against cerebral metacercariae of M. papillorobustus since this species is ‘the most detrimental for the host’. We found no supportive evidence that host immune reaction towards the larvae of any of the two species encysted in the brain of G. insensibilis, i.e. M. papillorobustus and L. propinqua, is stronger. Specifically, few cephalic larvae were melanized and no significant differences with respect to location were detected in the frequencies and rates of melanization of both species. The brain, however, is a ‘privileged’ site with respect to evading host immune response in vertebrates and data on host-parasitoid systems suggest that this may well be true for invertebrates (see Eslin and Prévost, 2000). Our comparisons of the frequency of melanization and melanization rates of the combined cephalic vs corporal infections support this assumption since both parameters showed substantially higher levels for corporal infections. We conclude therefore, that the larvae of the two species encysting in the host's brain have better chances of escaping from immune recognition.

Apparently, the amphipod-microphallid relationship is far more complex in the Pomorie lagoon and this can to a degree be related to the substantially higher infection rates in this locality (see Kostadinova and Mavrodieva, 2005). Thus the overall mean intensity of the microphallids is about 5 times higher than that recorded by Thomas et al. (2000) (39·7 vs 7·04) and so is the mean intensity of melanized metacercariae (3·24 vs 0·66). Furthermore, there are considerable differences in the structure of microphallid communities in the amphipod hosts under comparison (see Table 4). Thus the relative abundances of M. subdolum and M. hoffmanni are substantially higher in the sample from Pomorie lagoon whereas the opposite is observed for M. papillorobustus. Half of all metacercariae of this species recovered in gammarids from Thau lagoon were located in hosts brain (vs 2·2% in our sample) and these were all melanized (vs 1·2% in our sample). Finally, the fact that no melanized larvae of the 3 species co-occurring with M. papillorobustus have been recorded by Thomas et al. (2000) may simply reflect the fact that they were virtually absent in their sample (i.e. a total of 9 L. tridigitata, 41 M. subdolum and 54 M. hoffmanni metacercariae have been recovered in 500 gammarids, see prevalence range in Table 4). All above comparisons suggest that host melanization response in the amphipod sample from Thau lagoon has, in fact, been aimed at the most prevalent and abundant microphallid species. Our analyses clearly indicate that at high levels of multispecies infections the larvae located in the haemocoel are more prone to host immune response.

Another distinctive feature of the host-parasite system studied by us is the differential melanization of the microphallids. Not only the 4 microphallid species infecting G. insensibilis at Pomorie lagoon exhibited differing infection parameters (Kostadinova and Mavrodieva, 2005) but they were also melanized at different rates. The stronger association between the numbers of melanized metacercariae and conspecific parasite load, support this conclusion. Surprisingly, the specific melanization patterns confirmed neither the predictions of the previous study (Thomas et al. 2000) nor our initial expectations, based on the overall infection parameters and high levels of association between the microphallid species infecting G. insensibilis. In particular, host defensive response was largely directed against the larvae of M. hoffmanni and M. subdolum, and to a lesser extent against M. papillorobustus. The differing position of the 2 Microphallus spp. in this host-parasite relationship and specifically the leading position of M. hoffmanni with respect to levels of melanization, appears most puzzling. Both species show similar, and distinctly lower than those of M. subdolum, infection parameters and this precludes considering the difference as an effect of the latter. Two possible explanations can be suggested for the observed differential pattern of melanization: (i) M. hoffmanni is less adapted to avoid/suppress its host immune response; and (ii) M. hoffmanni is more harmful for its host. Although M. hoffmanni has been relatively recently ‘discovered’ and the data on its geographical distribution are scarce (see Kostadinova and Gibson, 1994) it may appear a good candidate for maladaptation. In contrast to the other 2 abundant parasites in G. insensibilis, M. subdolum and M. papillorobustus, whose populations are sustained by the local gull populations and therefore infect amphipods in several multiple waves during the year, the transmission of M. hoffmanni only occurs during late Autumn – early Spring and is probably related to overwintering bird populations from Northern Europe (most likely Anatidae but also Podiceps nigricollis, see Kostadinova and Gibson, 1994). A single infection wave with rare parasite genotypes might initiate mounting an immune response in the local amphipod population. Another possibly more parsimonious explanation is that although not numerically dominant, the metacercariae of M. hoffmanni occupy substantially more space in individual amphipods than the other microphallids. An approximate (using minima for the cyst size measurements) estimation of the biovolume of M. hoffmanni shows that it represents at least half of the total parasite biovolume in our sample (see Table 4 for a comparison with the remaining species). We assume that the physical effect of the metacercariae (e.g. haemolymph flow obstruction, organ displacement, etc.) of this species is stronger and the host perceives it as most harmful. A larger biovolume is also associated with a higher contact surface for PO adhesion.

In conclusion, our correlative study suggests that (i) amphipods invest more in defensive melanization in an environment where the likelihood to become infected is higher; (ii) host immune response is differential rather than generalized but not exclusively focused on the cerebral metacercariae of the manipulative species, M. papillorobustus; (iii) multiple waves of multiple parasites seem to exhaust hosts immune reserves and this favours successful parasitism.