Skip to main content
Dryad logo

Data from: Seasonal dietary shifts enhance parasite transmission to lake salmonids during ice cover


Prati, Sebastian; Henriksen, Eirik H.; Knudsen, Rune; Amundsen, Per‐Arne (2021), Data from: Seasonal dietary shifts enhance parasite transmission to lake salmonids during ice cover, Dryad, Dataset,


Changes in abiotic and biotic factors between seasons in subarctic lake systems are often profound, potentially affecting the community structure and population dynamics of parasites over the annual cycle. However, few winter studies exist and interactions between fish hosts and their parasites are typically confined to snapshot studies restricted to the summer season whereas host-parasite dynamics during the ice-covered period rarely have been explored. The present study addresses seasonal patterns in the infections of intestinal parasites and their association with the diet of sympatric living Arctic charr (Salvelinus alpinus) and brown trout (Salmo trutta) in Lake Takvatn, a subarctic lake in northern Norway. In total, 354 Arctic charr and 203 brown trout were sampled from the littoral habitat between June 2017 and May 2018. Six trophically transmitted intestinal parasite taxa were identified and quantified, and their seasonal variations were contrasted with dietary information from both stomachs and intestines of the fish. The winter period proved to be an important transmission window for parasites, with increased prevalence and intensity of amphipod-transmitted parasites in Arctic charr and parasites transmitted through piscivory in brown trout. In Arctic charr, seasonal patterns in parasite infections resulted mainly from temporal changes in diet towards amphipods, whereas host body size and the utilization of fish prey were the main drivers in brown trout. The overall dynamics in the community structure of parasites chiefly mirrored the seasonal dietary shifts of their fish hosts.


Fish sampling and processing

In total 354 Arctic charr and 203 brown trout were sampled from the littoral habitat (<15 m depth) between June 2017 and May 2018 using multi-meshed gillnets with panels of eight different mesh sizes from 10 to 45 mm, knot to knot (Table 1). The sampling was carried out monthly during the ice-free season (June to November) and every second month during the ice-covered period (December to May). During the ice-covered period, gill nets were pulled out and retrieved through holes in the ice by means of submerged ropes. The ropes were positioned in the lake in December when the ice thickness was still modest. The nets were left in the lake overnight for approximately 12 hours during the ice-free period and approximately 16 hours during the ice-covered period. In the field, fork length in mm, weight, sex and gonad maturation of all fish were recorded. Stomachs were opened and the fullness degree was determined on a scale from 0 to 100%. Prey types were identified and their contribution to the total stomach contents was calculated according to the method described by Amundsen (1995). The stomach contents were preserved in 96% alcohol, and the intestines were frozen to preserve the content, allowing subsequent parasitological and dietary analyses in the laboratory.

Parasite sampling

The intestinal parasites were sampled by cutting the intestines open and sieving the contents including that of the pyloric caeca under running water with a 120-micron mesh size nylon net. The collected material was then placed into a Petri dish with a physiological saltwater solution (9% NaCl). We found 5 taxa: Crepidostomum spp., Cyathocephalus truncatus, Eubothrium salvelini, E. crassum, and Proteocephalus sp., which use Arctic charr or brown trout as their final host. (Table 2, Figure 1). At least four potentially different species belonging to the genus Crepidostomum are found in Lake Takvatn (Soldánová et al., 2017), here grouped as Crepidostomum spp. as they are only distinguishable via genetic analysis. The only representative of the genus Proteocephalus is here described as Proteocephalus sp. since the exact species is not known. Additionally, the larval stage (plerocercoids) of two different species of Dibothriocephalus (formerly Diphyllobothrium (Waeschenbach et al., 2017)) were also recorded in the intestines of both Arctic charr and brown trout and are here grouped together as Dibothriocephalus spp. (Figure 1). The Dibothriocephalus spp. plerocercoids analyzed in this study include only the unencysted larvae found in the intestine, not those encysted in the viscera. The presence of unencysted plerocercoids in the intestine has previously been considered accidental. However, a high correlation between the number of unencysted plerocercoids and the degree of piscivory (see later), particularly in brown trout, strongly indicates that their presence was the result of recent ingestion of infected fish prey, and the Dibothriocephalus spp. plerocercoids were therefore taken into consideration for the analyses.

Prey types in the gastrointestinal tract

Only amphipods, insect larvae, zooplankton, and fish were considered for the stomach-parasite analysis, as they are the potential intermediate hosts of the identified intestinal parasites. The importance of these prey in the fish was expressed as frequency of occurrence (Amundsen & Sànchez-Hernàndez, 2019). The dietary information from the individual stomach samples was incomplete due to a high number of empty stomachs (Arctic charr N=115, brown trout N=29), especially during winter-time. To overcome this issue, the intestinal contents of each fish were carefully examined for the presence of identifiable prey remains. The frequency of occurrence of prey types in the present study is therefore a combination of stomach and intestinal observations (i.e., the whole gastrointestinal tract) of each individual fish. The implementation of the intestinal prey data covered the missing diet information for 40% of the empty stomachs.

Statistical analysis

Descriptive and statistical analyses were performed with the open source software Rstudio (version 1.1.423, Rstudio Inc.) and QPweb (version 1.0.14, Reiczigel et al., 2019), both based in R (version 3.5.1, R Core Team). To investigate differences in parasite load between Arctic charr and brown trout, five quantitative parameters (mean number of taxa, abundance, prevalence, intensity and mean intensity) were analyzed according to Bush et al. (1997) and Poulin (1998). Mean number of taxa is defined as the mean number of parasite taxa per host individual. Mean number of parasite taxa was used instead of observed parasite species richness, as no seasonal differences between Arctic charr and brown trout were detected using the Jackknife method (Zelmer and Esch, 1999) as an estimator of parasite richness. The mean number of parasite taxa was compared between Arctic charr and brown trout using the Mann-Whitney U-test. Abundance is the number of parasite individuals of a particular species in a single host species (infected and uninfected). Prevalence is defined as the proportion of host individuals infected by a particular parasite among the examined sample of a specific host species, usually expressed in percentage. Prevalence was compared between host species using a χ² test with Yates correction for each parasite taxa separately. Intensity is the number of parasite individuals of a particular species in a single infected host species. Mean intensity represents the average number of parasite individuals belonging to a particular species found in all hosts infected by that parasite (uninfected hosts excluded). To test for differences in mean intensity of each parasite taxa separately a non-parametric maximum test that combines Brunner-Munzel and Welch U-tests was used as suggested by Welz, Ruxton and Neuhäuser (2018).

To analyze seasonal variations in the infections of intestinal parasites, monthly data were merged into four seasonal periods to cope with the low winter sample size (Table 1). As the length of both Arctic charr and brown trout significantly differed among sampling seasons (One-way ANOVA, F(3)=3.844, P=0.01 and F(3)=21.78, P<0.001 respectively), any size effect on the seasonal variation in parasite infections was also tested using a negative binomial generalized linear model (GLM) with length as a covariate. Negative binomial GLM is best suited to model the overdispersion of parasites distributions among hosts which is typically aggregated (Wilson & Grenfell, 1997; Rózsa et al., 2000; Paterson & Lello, 2003; Lindén & Mäntyniemi, 2011). The model included parasite counts of infected hosts (i.e., intensity) as the response variable with seasons and fish length as predictors. The use of sex as a covariate did not produce significant results, and age was excluded from the analysis because part of the age data material was missing. The function glm.nb from the MASS package in R was used to run the model, and Anova (Type II) function from the Car package in R was adopted to assess the main effects. Similarly, to account for fish body size, seasonality in prevalence was tested using a binomial GLM.

To assess differences in parasite communities between seasons and host species, we used PERMANOVA (function Adonis in vegan package) on Bray-Curtis abundances matrices, thereafter illustrating the results using nonmetric multidimensional scaling (NMDS). Canonical correspondence analysis (CCA) was used to assess the relationship between the abundance of parasite taxa (response variable), presence-absence of prey types, and fish body length (explanatory variables). ANOVA-like permutations (999 cycles, function anova.cca in Vegan package) were used to test which variables explained a significant part of the variation in parasite abundance. Fish with no intestinal parasite infection were by default omitted from the CCA analyses. Species diversity across seasons was calculated using Shannon index (H’). Shannon index values of Arctic charr and brown trout were then compared with Hutcheson t-test. A correlation matrix with the Winsorized correlation coefficient (Wilcox, 2001) was further used to analyze potential correlations between parasite prevalence and frequency of occurrence of prey types. This method was preferred over the widely used Spearman-Rank and Kendall-Tau correlation coefficients as it is more robust to distribution shape, sample size, and outliers (Wilcox, 2001; Tuğran et al., 2015).