A project undertaken at the School of Environmental and Life Sciences, Murdoch University, and supervised by Stephen Beatty
Background
The South West Coast Drainage Division of Western Australia contains only ten species of native freshwater fishes, of which eight are endemic to the region. The south west also contains 13 species of alien fishes, of which goldfish (Carassius auratus) are among the most important Goldfish are a particular problem in the Vasse River system, where we have been been undertaking removal programs since 2005. In project APSF 04/2, we identified, for the first time, the parasites of native and alien fishes in the south west. One of these parasites was Lernaea cyprinacea, which was introduced with goldfish and has subsequently spread to native fishes. The parasite is more prevalent, and apparently more pathogenic on native species than on goldfish. A simple epidemiological model suggests that removal of goldfish may increase the prevalence of parasitic infection on native fishes (because goldfish are less competent hosts). On the basis of this model, we have stopped goldfish control programs in rivers where the parasite is present. The predictions of the model, however, depend critically on two parameters for which we currently have no reliable estimates; the relative susceptibility to infection of goldfish and native fishes when exposed to the parasite, and the fecundity of the parasite on goldfish and native fishes. In this project, we aim to obtain estimates of these parameters and test the predictions of our epidemiological model.
Project objectives
- Estimate the relative susceptibility of goldfish and the most common species of native fish (western pygmy perch, Nannoperca vittata) when exposed to Lernaea cyprinacea.
- Estimate the relative fecundity of Lernaea cyprinacea on pygmy perch and goldfish.
- Use the estimates of susceptibility and fecundity to parameterize the epidemiological model.
- Test the predictions of the model in mesocosm experiments to determine whether goldfish removal is likely to increase the infection pressure of Lernaea cyprinacea on native fish species.
- Test the predictions of the model in the field.
- Use these results to guide goldfish control programs in Western Australia.
Results
Lernaea cyprinacea was found on fishes in three river systems in the south-west of Western Australia; the Canning River and its tributary the Southern River, in which L. cyprinacea had been previously recorded, and the Serpentine and Murray Rivers, which represent new distribution records. The species identity of L. cyprinacea was confirmed by sequencing at the 18S and 28S rDNA loci.
Overall, L. cyprinacea infestation was more prevalent in native freshwater species than exotic freshwater fishes. The parasite was found on six native fish species, with a mean prevalence of 0.18 (± SE 0.04) and on three exotic fish species, with a mean prevalence of 0.02 (± SE 0.06) (Table 1). Goldfish and pygmy perch occurred together in two locations; the Canning River and the Serpentine River. Although in the Canning River there was no difference between these two species in either prevalence (Fisher exact test, P = 0.34) or intensity of infection (Kruskall-Wallis test, z = 0.72, P = 0.47), in the Serpentine River, prevalence was significantly greater in pygmy perch than in goldfish (Fisher exact test, P < 0.0001) (intensity could not be compared because no infected goldfish were found in the Serpentine River).
Table 1. Prevalences (proportion of infected fish) and mean intensities of infection of Lernaea cyprinacea of fish species in the Canning, Murray and Serpentine Rivers. 95% confidence intervals in parentheses. N = total number of fish sampled.
|
|||||
River | Fish Species | N | Prevalence | Intensity | |
|
|||||
Canning | Native | B. porosa | 54 | 0.11 (0.05-0.23) | 2.2 (1.0-3.0) |
G. occidentalis | 116 | 0.17 (0.11-0.25) | 1.3 (1.1-1.4) | ||
N. vittata | 269 | 0.07 (0.05-0.11) | 1.2 (1.0-1.4) | ||
P. olorum | 25 | 0.08 (0.01-0.26) | 1.5 (1.0-2.0) | ||
Alien | C. auratus | 33 | 0.12 (0.04-0.28) | 1 | |
G. holbrooki | 367 | 0.003 (0-0.02) | 1 | ||
P. caudimaculatus | 507 | 0.002 (0-0.01 | 1 | ||
|
|||||
Murray | Native | G. occidentalis | 127 | 0 | |
L. wallacei | 361 | 0.003 (0-0.02) | 1 | ||
N. vittata | 211 | 0 | |||
P. olorum | 297 | 0 | |||
T. bostocki | 255 | 0.32 (0.27-0.39) | 1.3 (1.2-1.4) | ||
Alien | G. holbrooki | 147 | 0 | ||
|
|||||
Serpentine | Native | B. porosa | 15 | 0.33 (0.14-0.60) | 1.6 (1.0-2.2) |
G. occidentalis | 101 | 0.11 (0.06-0.19) | 1.1 (1.0-1.3) | ||
N. vittata | 20 | 0.55 (0.32-0.75) | 2.2 (1.4-3.1) | ||
P. olorum | 28 | 0.14 (0.05-0.32) | 1.2 (1.0-1.5) | ||
T. bostocki | 59 | 0.51 (0.38-0.64) | 1.6 (1.3-2.0) | ||
Alien | C. auratus | 56 | 0 | 0 | |
G. holbrooki | 80 | 0 | 0 | ||
|
|||||
The greater prevalence on native freshwater fishes in the field may be due to a greater rate of exposure to the parasite and/or to a greater infectivity of the parasite on these species. To disentangle these possible causes, pygmy perch and goldfish were exposed to L. cyprinacea in experimental infections in the laboratory. Under these conditions, pygmy perch were significantly more likely than goldfish to be parasitised (Χ2 = 9.44, P = 0.002), with an overall prevalence for pygmy perch of 0.59 (95% CI 0.24-0.42), compared to 0.33 (95% CI 0.48-0.68) for goldfish. Mean intensity of infection was also significantly greater in pygmy perch than in goldfish (Χ2 = 4.44, P = 0.04, with a mean intensity of 3.0 (95% CI 2.4-3.8) for pygmy perch and 2.0 (95% CI 1.5-2.8) for goldfish. Mortality rate was significantly greater in pygmy perch than in goldfish (Χ2 = 29.63, P < 0.0001), with a mortality rate for pygmy perch of 0.41 (95% CI 0.27-0.55) and no mortalities in goldfish. Attached parasites appeared to be embedded much more deeply into the flesh of pygmy perch than goldfish. Mortality rate in pygmy perch, mortality was positively related to intensity (Χ2 = 18.51, P < 0.0001). The mean intensity of infection in fish that died was 3.9 (95% CI 2.6 – 5.1), compared to 2.2 (95% CI 1.6 – 2.9) in fish that survived.
The greater infectivity of L. cyprinacea to pygmy perch than to goldfish may be due to differences between the fish species in defensive behaviours and/or immunological responses. Putative defensive behaviours (gulping, jerking, scraping, pecking) were observed much more frequently in infected goldfish than in pygmy perch (Fisher exact tests: for gulping, P = 0.0001; for jerking, P < 0.0001; for scraping, P = 0.06; for pecking, P = 0.01). When goldfish and pygmy perch were treated for infection and then re-exposed to the parasite, the re-infection rate for pygmy perch was 0.75 (95% CI 0.50 – 0.91, while that for goldfish was 0.39 (95% CI 0.18 – 0.62 (significantly different by Fisher exact test, P = 0.04), suggesting that the species also differ in acquired immunity to L. cyprinacea.
Over all fishes the mean fecundity of attached adult female L. cyprinacea was 226.7 eggs/parasite (95% CI 142.9 – 310.4). There was no difference in fecundity of L. cyprinacea between fish hosts (Wilcoxon signed rank test, z = 1.16, P = 0.24) and no relationship between fecundity and intensity of infection (Spearman’s r = 0.01, P = 0.98).
An epidemiological model was constructed, based on the “reservoir potential” model of Mather et al. (1989; American Journal of Epidemiology 130: 143-150). The competence of the ith host species to transmit infection (Ci) was defined as the number of infective stages (Ii) produced by hosts of that species per day, divided by the total number of infective stages produced by hosts of all species per day (IT). For the ith host, Ii = Di Pi (Bi/Ti) Fi, where Di is the host density, Pi is the proportion of hosts infected, (Bi/Ti) is the mean intensity (burden) of parasites per infected host per day, Ti is the mean lifespan of an attached adult female and Fi is the mean lifetime fecundity of an attached adult female. Over all hosts, IT = ΣIi.
The epidemiological model was initially parameterised using estimates of prevalence and intensity of infection for naïve and recovered hosts obtained from laboratory experiments, and estimates of parasite lifespan and fecundity obtained from the literature. Using these estimates, simulation results suggested that reservoir potential (i.e. the number of infective parasite stages in the environment) is inversely related to goldfish density. These initial simulations, however, did not account for the increased mortality of infected pygmy perch compared to goldfish and therefore assumed that parasite lifespan was the same for both host species. Using differential host mortality rates to adjust parasite lifespan removed this trend, predicting no relationship between reservoir potential and goldfish density.
To test these predictions, infection experiments were conducted in laboratory aquaria, stocked with either eight pygmy perch or four pygmy perch and four goldfish. Infection prevalence of pygmy perch was 0.52 (95% CI 0.42 – 0.61) in single species communities and 0.59 (95% CI 0.48 – 0.68) in mixed species communities (not significantly different by Fisher exact test, P = 0.33). Mean intensity of infection was lower in single species communities (2.0 parasites/fish, 95% CI 1.7 – 2.4) than in mixed species communities (3.0, 95% CI 2.4 – 3.9), but this difference was not quite significant (Wilcoxon signed-rank test, z = 1.89, P = 0.06). Mortality rate was also lower in single species communities (0.35, 95% CI 0.26 – 0.44) than in mixed species communities (0.41, 95% CI 0.27 – 0.56), but again this difference was not significant (Fisher exact test, P = 0.58).
In the field, parasite prevalence and intensity were comapred in populations of pygmy perch from paired locations in each of the Canning and the Serpentine Rivers, where goldfish were prevalent at one location, but not found at the other location. Aside from the presence or absence of goldfish, fish community structure was similar in each pair of locations. While a comparison of only two paired locations does not permit statistical analysis of the data, there was no evidence that pygmy perch were more heavily infected when goldfish were absent (Canning River, prevalence 0.14 (95% CI 0.07-0.25) when goldfish present compared to 0.06 (95% CI 0.03-0.09) when goldfish absent; Serpentine River prevalence 0.55 (95% CI 0.32-0.75) when goldfish present compared to 0.16 (95% CI 0.07-0.32) when goldfish absent.
From these results, there is no evidence that removal of goldfish will increase the prevalence or intensity of infections of L. cyprinacea in native fish species. Although native fish species are more readily infected and habour more attached parasites than goldfish, because they appear to lack both innate behavioural defenses and acquired immunity to the parasite, they also suffer greater mortality and this markedly reduces their competence in transmitting further infections.