Parasite and host collection
Approximately 500 periwinkles (L. littorea) were haphazardly collected by hand at Årøsund, Denmark (55.25°N, 9.70°E) in August 2017, and transported to the GEOMAR laboratory in Kiel, Germany. Here, periwinkles were kept together in plastic containers of 25 L filled with seawater at 19 and 16 °C (i.e., the conditions at the site and time of sampling), aerated and fed ad libitum with Ulva lactuca and Fucus vesiculosus. On the day after collection (measurement 0), each periwinkle was individually placed in a petri dish containing 10 mL of seawater (19) and incubated under constant illumination underneath heating lamps at 25–27 °C for 4 h to promote cercarial emergence (e.g., Mouritsen 2002). Subsequently, petri dishes were checked under a stereomicroscope for the presence of cercariae. Periwinkles shown to be infected with H. elongata by this method were then kept in 10 L plastic containers in aerated seawater at 19and 16 °C (fed ad libitum with U. lactuca and F. vesiculosus).
Mussels (M. edulis) of 40–50 mm of shell length (i.e., 1–1.5 years of age) were collected from the ‘Kieler Meeresfarm’, a marine aquaculture facility in the inner Kiel Fjord (54.36°N, 10.16°E), as these are known to be free from trematode parasite infection. To confirm the uninfected status of mussels, 50 individuals were dissected immediately after collection and tissues (squeezed between two glass slides) were checked for parasites under a stereomicroscope. No infected mussels were found.
General experimental design and set-up
All experiments were conducted using temperature and salinity (fully crossed) as well as time (only for cercarial output) as fixed factors, and periwinkle/mussel identity nested within water bath as random factor (See Electronic Supplementary Material, ESM Fig.S1). Temperature levels applied were 19 and 23 °C. These two temperature levels were chosen based on representative summer temperatures in shallow Western Baltic Sea habitats (19 °C as shown for the past 15 years in Kiel Fjord; Pansch et al. 2018), and the expected increment by 2100 (+ 4 °C; IPCC Report 2019). Salinity levels applied were 13, 16, and 19. Following predictions for the Baltic Sea, salinity is expected to drop due to increasing precipitation rates and freshwater runoff (Vuorinen et al. 2015; Johnson et al. 2018). We, thus, chose two severity levels of desalination, i.e., reduction by 3 and 6, respectively, from 19 which is the average salinity of Årøsund, the site where L. littorea was collected. Furthermore, the activity of L. littorea was shown to considerably decrease at salinities below 13.5 (Rosenberg and Rosenberg 1972).
The experiments were conducted in six temperature-controlled water baths (50 × 50 × 50cm) located in a single climate chamber (16 °C). Cercarial motility and survival was assessed in a different room, using six smaller temperature-controlled water baths (8 L). Three of the six water baths were set at 23 and three at 19 °C, and each water bath included all salinity levels. There were (i) five independent containers (one periwinkle per container) per salinity level per water bath for the cercarial emergence phase, (ii) one 96-well plate (15 cercariae, one cercaria per well) per salinity level per water bath for the functional activity/survival phase, and (iii) four independent containers (one mussel per container) per salinity level per water bath for the infection success and susceptibility phases. For each salinity level, filtered Kiel Fjord seawater was used and either marine salt (Seequasal) or deionized water was added, until the desired salinity was reached. In each experiment, temperature and salinity were measured with a digital salinity meter (ProfiLine, Cond 3110, WTW), within the experimental jars every other day.
Exp. I cercarial emergence from periwinkles
Cercarial emergence was assessed by counting all cercariae that emerged from single periwinkles exposed to the different treatments. Infected periwinkles (20–25 mm length) were haphazardly assigned to the different treatment combinations. Periwinkles were placed individually into 250 mL glass jars filled with seawater at the different salinities. Water at the respective salinity conditions was prepared by adding salt (Seequasal) or DI water as necessary. After individually adding the snails, each jar was tightly closed with a lid preventing evaporation and provided with a small hole to allow oxygenation through a tube (from air pumps). Salinity and temperature were checked every other day before the water change, and only minor salinity increments were registered (+ 0.2–0.3). Five jars per salinity treatment were placed into each waterbath (three baths for each temperature level x five periwinkles/jars for each salinity level, i.e., 15 periwinkles in total per treatment combination), to acclimate periwinkles to the respective temperature and salinity levels. Periwinkles were fed daily ad libitum with U. lactuca and the water was exchanged every second day. Acclimation to the different temperatures and salinities lasted for 1 week, increasing/decreasing temperature and salinity by 1 °C and 1 psu per day, respectively.
To measure cercarial emergence, periwinkles were individually transferred from the jars into 50 mL Plexiglas beakers (Omnilab, ESM Fig. S2a), 1 day after the 7-day acclimation (measurement 1), 1 week after the start of exposure to the treatments (measurement 2), and 2 weeks after the start of exposure to the treatments (measurement 3), representing three sequential measurements for each individual periwinkle. Each incubation lasted 8 h. Each beaker was filled with 40 mL of seawater of the respective salinity and small amounts of U. lactuca were added. The beakers were covered with a net and incubated for 8 h in the same water baths from which the periwinkles originated (ESM Fig. S2b) At the end of the incubation period, the periwinkles were individually placed back into their 250 mL glass jars at treatment conditions, while the water from each beaker was transferred into 50 mL Falcon tubes. In addition, the beakers were immediately washed with 5 mL ethanol (99%), and the solution was also added to the Falcon tube to ensure the collection of all cercariae and their preservation for later counting. Samples were centrifuged for five minutes at 800 g, excess water was discarded, and cercariae at the bottom were poured into a petri dish for counting under a stereomicroscope (Nikon, SMZ1000). Periwinkles that did not shed any cercariae over all the three incubations were excluded from the analysis.
Exp. II cercarial activity and survival
Infected periwinkles were acclimated for 3 weeks under the different treatments (n = 10 individuals per treatment) and incubated as described above for 4 h to allow for cercarial release. At the start of the experiment, approximately 45 cercariae per treatment, obtained from the pooled cercariae released within four hours from the ten snails per treatment, were added individually to wells of three 96-well plates (n = 15 wells into three replicate 96-well plates per salinity; 45 wells in total). The amount of water (of the respective salinity levels of 13, 16, and 19) in each well was around 0.25 mL. The 96-well plates were tightly closed with the respective manufacturer’s lids throughout the 45 h period preventing evaporation, except when cercariae were examined for activity and survival. The 96-well plates were then exposed to the different temperature treatments placing them in the six thermobaths (three thermobaths for each temperature level x one 96-well plate for each salinity level = three 96-well plates per treatment combination). Activity and survival were then assessed by visual assessment under a stereomicroscope after 4, 6, 9, 18, 27, and 45 h. Cercariae were considered fully active when they were constantly swirling around (category: fully active). When cercariae were laying at the bottom of a well and not reacting after physical stimuli (through pipette’s tip), these were considered dead (category: dead), when still reacting, even if slowly, these were still considered alive (category: alive). This last category included “fully active”. Wells that accidentally received more than just one cercaria during the inoculation were excluded from the analysis.
Exp. III cercarial infectivity in mussels
The infectivity of H. elongata cercariae was investigated by counting metacercariae in M. edulis, after a standardized exposure to cercariae acclimated at different temperatures and salinities. Infected periwinkles were acclimated to the treatments for 4 weeks prior to the infection success assays. Mussels were collected 2 days prior to the start of the assays, kept in ambient salinity and temperature conditions and fed with Rhodomonas salina. A total of 72 mussels (n = 12 for each treatment combination) of 40–50 mm shell length, kept at 19 psu and 16 °C, were individually placed in 50 mL beakers at the respective temperatures and salinities, and exposed to a standardized number of cercariae released from the all periwinkles from the different treatment combinations. In each waterbath, four beakers per salinity combination were placed, i.e., a total of 12 beakers per bath. To obtain cercarial stages for the infection assays, periwinkles were individually placed in petri dishes at the different salinities and incubated for four hours under constant light (see above). After release, all cercariae (4–5 h old) of a given treatment combination were collected mixing the water from the different petri dishes into one single container, counted under the stereomicroscope and immediately added to the mussels. This accounted for genotypic and phenotypic variation among cercariae. Each of the 12 mussels of a given treatment combination was inoculated with 22 (13–19 °C, 16–19 °C and 19–23 °C), or 23 (19–19 °C) cercariae, except for the 16–23 °C treatment, in which only five mussels where infected with 20 cercariae each due to the low number of cercariae obtained from such pre-treated periwinkles. The 13–23 °C treatment was excluded from the analyses due to extremely low numbers of cercariae released under this particular treatment combination. The beakers with mussels were then returned to the water baths at the treatment temperature and incubated for 24 h. This incubation period was chosen to ensure cercarial encystation (de Montaudouin et al. 2016), but to also limit possible metabolic stress for the host, that could have affected the experiment. After the incubation, each mussel was collected and dissected, and H. elongata metacercariae were identified and counted under a stereomicroscope.
Exp. IV susceptibility of acclimated mussels to cercarial infection
The effects of temperature and salinity on the susceptibility of M. edulis to infection by H. elongata cercariae were assessed by acclimating mussels to the above-mentioned treatment combinations (salinity × temperature) for 1 week, before these were exposed to non-acclimated cercariae. In this experiment, cercariae did not receive any temperature/salinity treatment and only the mussels did (after acclimation), to specifically investigate a putative shift in susceptibility under the treatments by the second intermediate host. We assumed acclimation effects to be stronger in the (acclimated) hosts and less likely on the very short-term basis in the added (non-acclimated) cercariae. However, some potential confounding factors by cercariae very quickly acclimatised cannot be fully excluded. Cercariae were obtained as described above from periwinkles collected 2 days earlier from the sampling site at Årøsund, Denmark. The exposure to cercariae of each mussel (n = 12 mussels per treatment) was realized as described for Exp. II, but this time, the number of cercariae added to individual mussels was 54 (16–19 °C, 16–23 °C and 19–23 °C), 53 (13–19 °C), and 56 (19–19 °C). Beakers with mussels were incubated at the different treatments for 24 h. Mussels were then dissected and the number of metacercariae in each mussel (squeezed between two glass plates) was counted under a stereomicroscope.
Statistical analysis
Prior to analysis, the normality of data distribution was tested through Shapiro–Wilk tests (Shapiro and Wilk, 1965) and residual plots were checked visually. Non-parametric Generalised Linear Mixed Models (GLMMs), using the packages lm4 and MASS, were then applied to cercarial emergence, infectivity, and mussel susceptibility. Residual extraction, the marginal R2, and conditional R2 were used to reveal individual identity and thermobath effect, using the function r.squared GLMM (package MuMIn; Nakagawa and Schielzeth 2013). Individual identity nested within thermobath was then used as random factor in all the steps analysed with GLMM. The Akaike Information Criterion (AIC) was used to select the best GLMM model, from models including all fixed factors and reduced models (without temperature or salinity as fixed factors). Two-way ANOVA was instead used for activity and survival. All statistical analyses were performed using the software R 3.5.0 (R Development Core Team 2018).
For cercarial emergence, a negative binomial GLMM, using the lme4 package, was used to test for the effects of temperature, salinity, and sampling time (fixed factors) on the number of H. elongata cercarie (Exp. I). Negative binomial structure was chosen due to high over-dispersion and presence of zeros in the data.
For cercarial activity/survival (Exp. II), the analysis applied was selected to link Exp II with Exp. III. For this purpose, we accounted of the differential slope of decreasing activity/survival during these 24 h by considering the integral over time (integrated area), which should correspond to the (changing) infection pressure that the mussels underwent in the infection experiments. The average proportion of “fully active” and “alive” cercariae present in one 96-well plate was considered as one replicate. Proportions were calculated through the ‘survfit’ function (package survminer). The integral areas of both 0–24 h and 0–45 h intervals were calculated using the mean proportions of the three 96-well plates to fit the curves. Curves of activity were fitted using geometric model (Power Low family), and the curves of survival using logistic power model (Sigmoidal models) (software: CurveExpert Professional 2.6.5). To test the effects, two separated two-factorial ANOVA were applied for each interval. Integrated motility and survival at 24 h and 45 h were the dependent variables, and temperature and salinity were the independent variables. Post hoc tests [Tukey honestly significant difference (HSD)] were performed following ANOVA.
Cercarial infectivity in M. edulis (Exp. III) was tested by a GLMM fitted with a Poisson structure with the number of metacercariae found in mussel tissue as dependent variable, and temperature and salinity were set as fixed factors. The same methodology as for infectivity was applied for susceptibility, however, here applying a GLMM fitted with a negative binomial structure.
Using the raw data, the effects of i) strong future freshening (salinity 13) in relation to present conditions (salinity 19) and ii) strong warming (23 °C) in relation to present condition (19 °C) were expressed as logarithmic response ratios (Lajeunesse 2015). For cercarial activity, only the 0–24 h interval was re-considered, linking this period to the infectivity experiment, in which mussels were incubated for 24 h. Effects were considered significant when the confidence interval (CI) bar was not overlapping zero. Effect sizes (in percentages) of the different life cycle steps were calculated applying the inverse of the log mean ratio for (i), all temperature levels were pooled, and for (ii), all salinity levels were pooled.