Specificity and host predictability: a comparative analysis among monogenean parasites of fish
Abstract
1. This article compares generalist (parasite species found on two or more host species) and specialist (found on only one host species) monogenean parasite species of fish. The reduction of the host range – that is an increase in host specificity – may correspond with a better adaptation of the parasite to a more predictable host environment. A more predictable environment may allow the parasite species to develop specific adaptations.
2. We assume that the more predictable host environment can be evaluated by host body size, since numerous life-traits, such as longevity, are positively correlated with size.
3. We found that specialist parasites parasitize larger hosts species than generalist parasites. We also found a good relationship between host body size and parasite body size for specialist parasite species.
4. An adaptation to the mechanical problems encountered in the host's gill chamber may lead to an increase in parasite body size. The infection of a larger part of the host population in order to decrease the chances of local extinction due to fluctuations of host abundance may be another adaptive mechanism.
5. We found a negative correlation between parasite body size and prevalence for generalist parasite species. This relationship disappeared when using the comparative method controlling for phylogeny, which proved that it was a phylogenetic effect.
Introduction
Parasites are viewed as highly host specific, but considerable variation in specificity occurs between and within parasite groups (Poulin 1992; Combes 1995). For example, among parasites of fish, acanthocephaleans and nematodes show little host specificity, whereas monogeneans seem to be highly specific (Euzet & Combes 1980; Poulin 1992). Even if there are no clearly identified determinants of specificity (Poulin 1992; Morand 1996a), one can postulate that specialization is an adaptive feature of a parasite (Ward 1992). The generalist vs. specialist strategy for a parasite consists of increasing or reducing host range, where host range is defined as the number of host species infected by a population of a single parasite species (Lymbery 1989). Host range reflects a reciprocal relationship between host and parasite (Ward 1992). Host specificity, which takes into account epidemiological indices, such as the percentage of infected hosts species or the intensity of infection (Lymbery 1989), may reveal more strong host–parasite interactions. Percentage of infected hosts may be used as an estimate of the proportion of susceptible hosts, this proportion being the result of both genetic and environmental factors (Minchella 1985).
Ward (1992) listed three factors which may explain the evolution of host specificity:
1. features of the host (behaviour, anatomy, physiology, etc.) requiring specific adaptation in the parasite;
2. restriction of the parasite's niche by competition or predation: this conforms to Jeffries & Lawton's hypothesis (1984) of specialization on enemy free-space;
3. specificity as a mean of finding mates on ‘rendez-vous’ hosts, i.e. Rodhe's hypothesis on seeking a host to find mates (Rohde 1979).
In this study, we investigate the determinants and consequences of host specificity among monogeneans of fish. As previously mentioned, monogenean species are very host specific, but some of them can infect several hosts (Euzet & Combes 1980; Poulin 1992). These parasites are good candidates for investigating the evolution of parasite specialization.
We hypothesize that specialization occurs on a predictable resource with a more predictable life history, that is for a parasite on a large-bodied or a long-lived host. Larger hosts are generally at the top of the food web and are the ones that live longest. Moreover, a recent comparative analysis of fish life histories (Winemiller & Rose 1992) has shown that fish body size correlates positively with several life-traits, such as longevity, growth or trophic status. The life span of the host may be particularly important for the evolution of direct cycle parasite as several parasite generations could sometimes occur on the same host (Llewellyn 1972).
The second hypothesis is that specialization leads to special adaptation to the host. We try to explain part of the host–parasite association by comparing the maximum sizes of the host and the parasite. A relationship between host body size and parasite body size may reveal co-evolution of life-traits (Morand et al. 1996; Sorci, Morand & Hugot 1997). Furthermore, we assume that the percentage of infected hosts can be considered as an indicator of the success of the parasite on one host, and we explore the relationship that may exist between host range and prevalence.
Host–parasite associations may reflect a long history of co-evolution, during which the ecology of the parasite is strongly linked with that of the host (Cameron 1964; Brooks & McLennan 1993; Poulin 1995). For these associations the effects of history and the effects of ecology are difficult to separate (Manter 1966; Holmes & Price 1980; Brooks & McLennan 1993; Winston 1995). It is thus necessary to take into account phylogenetic information when doing a comparative analysis (Brooks & McLennan 1991; Harvey & Pagel 1991; Harvey 1996).
Materials and methods
Parasite data
In order to avoid bias of the study due to geographical variation, data on parasite were compiled from a previously published study (Radujkovic & Euzet 1989) carried out at a regional scale (Adriatic Sea). Seventy-four monogenean species were used in this study, 51 specialist and 23 generalist species. When a parasite occurred on several host species (generalist species), we considered the mean percentage of infected hosts and the mean maximum fish sizes. For generalist species, when the percentage of infected hosts was not available for all host species, we took the mean values of the available data. The number of known hosts is the number of host species counted to be infected by a parasite species at the date of this study, even if no detailed data on infected hosts were available. Parasite species richness in specialist and generalist species of monogeneans, as well as percentage of infected hosts were controlled for host sampling size (Gregory & Blackburn 1991; Walther et al. 1995) and we used the residuals in the subsequent analysis. Phylogeny of parasites (Fig. 1) was obtained from the works of Boeger & Kritsky (1993, 1997).
Working phylogeny of the monogenean used in this study, adapted from Boeger & Kritsky (1993, 1997). Parasite body length is in micrometres and mean prevalence in percentage.
Host data
Parasite species were collected from 48 marine fish species (2547 individuals) from the above study (Radujkovic & Euzet 1989) from the Adriatic Sea (see details in Sasal, Morand & Guegan 1997). Data on maximum length of fish were taken from the check-list made by Whitehead et al. (1986) and ln transformed. Contrasts were performed using phylogenetic information according to Johnson (1993), Lecointre (1994) and Sasal et al. (1997).
Comparative analyses
The long-term historical relationship between a host and its parasites is very likely to have influenced the evolution of both hosts and parasites (Freeland 1983; Brooks 1985; Renaud & De Meeü 1991; Brooks & McLennan 1993). Practically everyone agrees that controlling for phylogeny is essential. Therefore, data are controlled for parasite or host phylogeny. However, we present comparisons between controlled and uncontrolled data, in order to highlight the importance of the phylogeny.
To solve the problem of non-independence of data in comparative studies, we used the phylogenetic independent contrasts method (Felsenstein 1985; Harvey & Pagel 1991; Garland, Harvey & Ives 1992; Pagel 1992; Harvey 1996). We used the caic program for Macintosh (Purvis & Rambaut 1995). Quantitative data were log transformed in order to stabilize variance (Harvey 1982). Because we wanted to verify that contrasts are properly standardized, we performed a regression of the absolute values of standardized contrasts vs. their standard deviations (Garland et al. 1992). All correlations between contrasts were forced through the origin (Garland et al. 1992). As it was not possible to determine branch length in the phylogenies, we assume that they were all of the same length.
Results
Host body size and parasite specialization
The frequency distribution of maximum body sizes of Mediterranean fish (613 species counted) was determined using the maximum sizes reported in the fish fauna (Fig. 2). Several host species (nine out of the 48 species studied) harbour both specialist and generalist parasite species. Finally, fish harbouring specialist parasites are, on average, larger than fish harbouring generalist parasite species (Fish harbouring specialists parasites = 1·85 ± 0·31 in log cm; Fish harbouring generalists parasites = 1·66 ± 0·26 in log cm; t-test: t-value = 2·4; P = 0·02; d.f. = 55; Fig. 2). Moreover, we found a significant positive relationship between host body size and host status (harbouring only specialists, only generalists or both type of parasite species; without controlling for phylogeny: slope = 0·50; R2 = 0. 14; n = 48; P < 0·01; controlled data: slope = 0. 26; R2 = 0. 15; n = 25; P = 0·052). We found no relationship between sample size and host range when controlling data for host size (P = 0·27).
Distribution of fish sizes in the Mediterranean Sea (in log): (a) a Mediterranean species; (b) host species harbouring generalist parasite species; (c) host species harbouring specialist parasite species. Note that 11 species harbour both specialist and generalist species.
A cross-species analysis showed that host body size was not correlated with parasite species richness in specialist or generalists species. However, when the data were controlled for host phylogeny, we found a significant positive correlation between host body size (in Ln) and parasite richness for specialist species (slope 0·53; R2 = 0·23; n = 22; P < 0·02).
Host body size and parasite body size
We found a significant correlation between host body size and specialist parasite species sizes. This result was significant for cross-species data (without correcting for phylogenetic relations; R2 = 0. 19; n = 50; P = 0·0014), as well as for data controlled for host phylogeny (R2 = 0·29; n = 22; P = 0·0075; Fig. 3). However, the result was no more significant when data were controlled for parasite phylogeny (n = 13, P = 0·07). Moreover, no significant correlation was found in the case of generalist parasite species (cross species, n = 23; P = 0·4; independent contrasts with host phylogeny, n = 11, P = 0. 14; independent contrasts with parasite phylogeny, n = 10, P = 0·54).
Relationship using independent contrasts, forced through the origin, between host size and parasite body size for specialist parasite species.
Host body size and percentage of hosts infected
For specialist parasite species (i.e. parasite living on one host species), there is a significant correlation between host body size and percentage of infected hosts (Table 1). The relationship is even stronger when the data are corrected for parasite phylogeny (Table 1). As host history could also be important for host body size, we repeat the test using hosts harbouring specialist parasite species. The result show that the relationship was significant when the data were corrected for host phylogeny (Table 1). No significant relationship was found for generalist parasite species when correcting for host or parasite phylogeny (Table 1).
| Specialist parasites | Host harbouring specialist parasites | Generalist parasites | Host harbouring generalist parasites | |||||
|---|---|---|---|---|---|---|---|---|
| Cross- species | Corrected for parasite phylogeny | Cross- species | Corrected for host phylogeny | Cross- species | Corrected for parasite phylogeny | Cross- species | Corrected for host phylogeny | |
| Host body size | 0.03 | <0.01 | 0.51 | 0.01 | 0.24 | 0.12 | 0.39 | 0.39 |
| ( n = 41) b = 0.18 |
( n = 19) b = 0.36 |
( n = 26) | ( n = 13) b = 0.30 |
( n = 17) | ( n = 8) | ( n = 18) | ( n = 8) | |
| Prevalence | ( R2 = 0.12) | ( R2 = 0.53) | ( R2= 0.43) | |||||
| Parasite body size | 0.01 ( n = 40) |
0.68 | 0.06 | 0.82 | 0.02 ( n = 17) |
0.30 | 0.98 | 0.98 |
| b = –0.11 | ( n = 18) | ( n = 26) | ( n = 13) | b = –0.23 | ( n = 8) | ( n = 18) | ( n= 8) | |
| Prevalence | ( R2 = 0·16) | ( R2 = 0·32) | ||||||
Parasite body size and percentage of hosts infected
We found a negative significant correlation between parasite body size and percentage of infected hosts for specialist (slope = – 0. 11; R2 = 0. 16; n = 40; P = 0·01) and generalist (slope = – 0·23; R2 = 0·32; n = 17; P = 0·02) parasite species (Table 1). However, these relationships were no longer significant when the data were corrected for the parasite phylogeny of specialist (P = 0·68) or generalist (P = 0·30) parasite species. No relationship was found when considering hosts harbouring specialist or generalist parasite species (Table 1).
Discussion
Monogeneans are usually considered to be very host specific parasites (Euzet & Combes 1980; Poulin 1992). However, some species are less specific than others for their host. Combes (1995) listed the costs of being a specialist as: (i) high risk of extinction when the host species becomes rare or disappears itself; (ii) smaller available niche space (Bush & Kennedy 1994); (iii) exposure to a single immune system that can evolve to eliminate the parasite (this is according to the Red Queen hypothesis, whereby specialist parasites should co-evolve to survive in their host); (iv) increased risk of mortality when entering an unsuitable host.
Our results reveal that parasite species richness of specific monogeneans correlates positively with host body size and that specific monogeneans are found on larger fish. We used body size as a mean to evaluate host predictability, because fish body size correlates positively with numerous life-history traits (Winemiller & Rose 1992) and especially longevity.
The fact that hosts of specialist parasites are significantly bigger than hosts of generalists, supports the hypothesis of specialization on hosts with a more predictable life history. Hosts with a long life expectancy may provide a more stable environment not subject to sudden changes (Basset 1992; Kitahara & Fujii 1994). Moreover, larger fish harbour higher monogenean species richness (Sasal et al. 1997), but only richness in specialists parasite species increases with host body size. These results agree with the general hypothesis of an increase of diversity with the stability and predictability of environment, but seem to disagree with a specialization in enemy free space that occurs where there is less competition (Jeffries & Lawton 1984). However, we may suspect that an increase in host size is linked with an increase in available niches for colonization (Kennedy, Bush & Aho 1986) and thus a greater parasite richness. Rohde (1979) has also suggested that interspecific competition has little influence on the organization of monogenean communities.
As the host constitutes the helminth's environment and, as the turnover time for a parasite generation is shorter than that of their hosts, we may expect evolutionary adaptations in parasites. The significant relationship between fish size and monogenean body size, found among highly specific species, seems to reveal a good adaptation to host size. This result concurs with the result obtained by Kirk (1991) on host-specific insect parasites or by Harvey & Keymer (1991) and Morand et al. (1996) on oxyurid nematodes of primates.
The view that larger hosts harbour larger parasites has received comparative support (Harvey & Keymer 1991; Morand et al. 1996). It has been hypothesized that larger hosts provide more resources and space, which, in turn, promote parasite growth. Positive relationships between body size and fecundity have been found in other parasites (Morand 1996b). Because large body size is correlated with high fecundity, a fecundity advantage is thus associated with an increase in parasite body size as pointed out by Skorping, Read & Keymer (1991), Keymer et al. (1991) and Morand (1996b).
The increase of parasite body size in relation to host size may be the result of a costly adaptation to the host environment. In the case of the monogeneans, parasites may have to develop larger, and thus more efficient hooks and muscles to be able to stay attached to the host and to avoid being dislodged by the water current. Llewellyn (1972) reported the importance of ventilating currents in the gill chamber of the host for the monogeneans. There is also an obvious relationship between the size of fish and the volume of their gill chamber (Hughes 1966). As the water flow increases with the fish size and as more active fish (larger fish) have larger gill area per unit of body weight (Gray 1954; Hughes 1966), parasites living on large hosts have to adapt their hooking system to the gill filament diameter if they do not want to be eliminated by their host. The significant relationship between host and parasite size for specialist species seems to indicate an adaptation of the parasite to the gill chamber's mechanical constrains. It may be reasonable to assume that the co-evolution of sizes could be a purely mechanical result. This hypothesis is reinforced by the existence of intra-specific variability in the monogenean Ligophorus imitans (Euzet & Suriano) (Caltran, Silan & Roux 1995) showing a morphological plasticity, and a positive relationship between fish body size and parasite body size. As this intra-specific variability for monogenean species could have heritable changes over a short time-scale, this phenotypic plasticity may lead to ecological isolation and, in time, to a divergence of species (De Meeüet al. 1993).
Even if it seems more reasonable to see the co-evolution of body sizes as a consequence of parasite specialization, the lack of correlation of host–parasite body sizes for generalist species may be explained in terms of energy allocation (Fox & Morrow 1981). Generalist species should allocate a larger part of their resources to resistance, as they encounter a broader range of host defences, which could prevent allocation of energy for size adaptation. This may also lead to maladaptive states due to a less efficient use of the host (Adamson & Caira 1994). Large parasite body size will also be a disadvantage with regards to cleaning fishes, as cleaning rate increases with host size (Grutter 1995, 1996). Finally, it may be that the costs of being a specialist can be compensated in terms of reproductive success since specialization may be an indirect result of the increase in body size (Skorping et al. 1991; Morand 1996b; Sorci et al. 1997).
In conclusion, the significant correlation found between host body size and percentage of infected hosts, reinforces the idea of a trade-off between a few performing well and many performing poorly for specialist parasite species (Adamson & Caira 1994; Wilson & Yoshimura 1994). Consequently, if we imagine that there is a cost to finding a mate in infecting fewer host species for specialist parasites (as there are fewer potential hosts in their close environment), a larger proportion of the host species should be infected. Finally, a parasite species that infects a large part of a host population may exert a stronger selective pressure on their hosts, in the same way as host species frequently infected may be important in selection for traits of the parasite. As the parasites can develop responses to host defences, i.e. increase their impact on hosts, generalists may progressively parasitize fewer species (Lanyon 1992). This strategy would reduce the host spectrum and increase the specificity. This conclusion does not take into account parasite virulence as our data do not allow us to test for it. One consequence of the global specificity of monogeneans is that they are usually highly aggregated species (Rohde, Hayward & Heap 1995). This could favour mate location (Rohde 1979; Adamson & Caira 1994), but also decrease the impact of parasites on host population (Anderson & Gordon 1982). Testing aggregation for generalist and specialist parasite species may help to test the often reported (but not really admitted) notion of a decrease of pathogenesis with parasite adaptation (Humphrey-Smith 1989; Carton & Nappi 1991).
Acknowledgements
This work was financially supported by the Conseil Régional du Languedoc: Roussillon, the Programme National sur le Déterminisme du Recrutement (P.N.D.R.), the French Ministère de l’enseignement supérieur et de la recherche and the CNRS. Authors would like to thank Gabriele Sorci, Thierry DeMeeüs and Louis Euzet for helpful comments on the first draft of this paper. We are also very grateful to Joseph Jourdane, Ioannis Michalakis, François Renaud, André Théron, and two anonymous referees for constructive suggestions and comments which greatly improved the final form of the manuscript.
References
Received 22 December 1997
