The broad-nosed pipefish Syngnathus typhle is sex-role reversed and lives in shallow eelgrass (Zostera marina) meadows along the coasts of Europe and has a lifespan of two or more years ^[30], ^[35]. S. typhle is regularly infected by a variety of parasite species ^[36], including our study species, the digenean trematode Cryptocotyle lingua. This generalist parasite shows a broad diversity of possible second intermediate hosts, but most of them are small coastal fishes ^[37]. Throughout European coastal waters C. lingua shows a high diversity where the majority of its European haplotypes can only be observed once ^[38]. Its larval cercarial stages penetrate the skin and lead to visible black spots (metacercariae). In the field we observed varying intensities (between 0 and 30 metacercariae) of infection with C. lingua. The complex life cycle of C. lingua includes three hosts: snails and fishes as first and second intermediate hosts respectively, and gulls as final hosts. The sexual reproduction takes place in the gull and parasite eggs are then shed in the water, with the gull's faeces. These eggs are then eaten by the first intermediate host, generally by the snail Littorina littorea. In the snail the eggs hatch into miracidia and penetrate the gut wall where they metamorphose into sporocysts ^[39]. In the sporocyst the rediae are formed and from its germ cells cercariae will be produced ^[40]. The cercariae are shed into the water, where the free-swimming forms search and penetrate the second intermediate host, generally fish, in which they transform into encysted metacercariae ^[40]. This complex life cycle excludes direct transmission from one fish host to another ^[40] and makes it a suitable model for laboratory infection.

We collected snails (Littorina littorea) and uninfected (no visible black spots) pipefish (Syngnathus typhle) in July and mid August 2010 at three different locations in North Europe in shallow water with eelgrass meadows (5^th and 6^th of July at Geltinger Bucht, Germany, N 54°75,57′; E 9°87,66′; on the 13^th–15^h of August in Fiskebäckskil, Sweden, N 58°24.80′; E 11°44.62′; at the 16^th of August at Doverodde, Denmark, N 56°71.90′ E 8°46.38′) by snorkelling using hand nets. Water temperature at the locations was between 14° and 18°C. We collected around 500 snails at each location at the shore on rocks, preferably at places with gull faeces to raise the possibility of an infection with C. lingua. From each location we collected between 90 and 100 juvenile pipefish, sex was mostly not identifiable yet. Before the experiment we let the fish acclimatise in 200 l barrels filled with 18°C water and artificial eelgrass for at least 15 days, and fed them ad libitum with live Mysis and frozen Artemia. Snails from the same location were kept in 20 l aquaria with constant aeration and fed ad libitium with Ulva sp. algae. Both, fish and snails were kept at 18°C and summer light conditions (day: 15 h, night: 9h).

We aimed to gain information about how a simulated heat wave, as environmental effect (E), affects a host-parasite interaction using a full factorial infection matrix of sympatric and allopatric host (H) and parasite (P) combinations (i.e. H×P×E). The pipefish were either kept at 18°C normal condition, or experienced a heat wave by increasing the water temperature to 25°C (incremental temperature increase of 1°C per day for 7 days), as it occurred in the Baltic Sea in 2003 ^[23], ^[41]. After reaching 25°C, this temperature was kept constant the following three weeks until the end of the experiment at either 25°C (high temperature) or 18°C (low temperature) to follow the 2003 summer heat wave situation in the Baltic Sea ^[21].

Pipefish from one population and the same temperature group were kept in a 200 l tank filled with aerated natural seawater (25 PSU) and artificial eelgrass before the start of the experiment. Original salinities were 16 PSU for German fish, 24 PSU for Danish fish and 27 PSU for Swedish fish, during the experiment salinity was kept rather high to prevent infections with Vibrio^[42]. The fish were fed ad libitum with Mysis and Artemia. All snails were constantly kept at 18°C. In order to trigger cercariae release, the snails were removed from the aquaria and placed individually in 50 ml Falcon tubes under a lamp. The released cercariae were checked under the stereomicroscope to confirm the species identity and the concentration was assessed. Cercariae from the same snail population were pooled and the concentration adjusted for the three different populations. This cercariae solution was then used for infection.

The experiment started at the 1^st of September when all fish were tagged with Visible Implant Elastomer (from NMT) with an individual colour code. For infection, pipefish from both temperature treatments (18°C and 25°C) were placed in 5 l tanks with the according acclimatisation temperature, in pairs of two as keeping fish singly imposes too much stress on pipefish. The fully reciprocal design included 3 hosts (DK, GER, SWE)×4 parasites (DK, GER, SWE+control)×2 temperatures (18°C and 25°C) = 24 treatments in total. Per treatment we used 10 replicates, which translated to a total of 240 experimental fish (24 treatments×10 replicates). During the next five days all fish were exposed daily to seawater containing cercariae or to seawater without cercariae as control treatment. Summed over five days the total dose were 80 C. lingua cercariae, coming from between 24 and 30 snails/population (with the following daily concentrations for all three locations [cercariae/aquaria]: 14, 26, 14, 13, 13).

During the infection week, 22 of 40 fish (infected and control) from the Swedish population in the normal water group (18°C) died, suffering from a fungal disease. 11 of the remaining fish died the following week. We therefore had to exclude this treatment group (Swedish fish, 18°C) from the data analysis.

After infection, 5–6 pipefish individuals (variation due to the Swedish fish) from the same origin and from the same temperature group were randomly placed together into larger, 80 l tanks (N = 36) in a circulating water system at either 18°C (control) or 25°C (heat wave). This minimized a possible stress response of the fish, confounding the latter immune measurements. Aquaria were arranged randomly and the two temperature groups (18°C and 25°C) were supplied with the same water (again at 25 PSU). The fish were fed twice a day with live Mysis and frozen Artemia and had artificial eelgrass and aeration in their tanks. After seven days the immune response of the fish and the transmission success of the parasite were measured.

To measure immune response the fish were killed and dissected in a random order during three subsequent days. We focused on the head kidney which is a major fish immune organ ^[43], ^[44], in particular for Syngnathids, as they lack a spleen ^[45]. We measured the proportion of lymphocytes and monocytes by means of flow cytometry, using a FACSCalibur and CellQuest Pro Software (both BD) to gain information about the adaptive and innate immune system. As a parameter for the activity of the adaptive immune system, the relative number of lymphocytes in the G_2-M phase and S phase of the cell proliferation cycle were determined in contrast to the resting phase G₀₁^[43].

Before the experiment we dissected over 50 naturally infected fish from the field to find the ratio of metacercariae in skin and organs or muscle tissue. In all cases we only could find metacercariae (also visible as “black spot”) in the skin. The appearance of a “black spot” therefore is to our understanding a suitable approximation for the success of the parasite entering the host. To assess the parasite's infectivity we counted metacercarial spots ^[46].

All experiments were performed according to current national legislation on Animal Welfare. Prof. Dr. Gerhard Schultheiss from the Ministerium für Landwirtschaft, Umwelt und ländliche Räume des Landes Schleswig-Holstein approved to perform the experiments with the permit number 187, named “Effects of global change on immunological interaction of pipefish and their natural parasites” described in this manuscript.

For all three host-parasite populations tested we found positive values for the specificity index, indicating local parasite adaptation. These effects were independent of temperature. Not all specificity indices were significantly different from zero (Table 1), but for each host-parasite combination, at least one specificity index differentiated significantly from zero. This indicates a clear pattern of local adaptation of the parasite to its local host. This pattern persists even if we allocate the host-parasite interaction into a new, warmer environment by imposing a heat wave.

The outcome of the infection (number of metacercariae) is higher, when sympatric hosts and parasites meet (ANOVA with subsequent Tukey HSD, p adj. = 0.003, Figure 1, Table 2). On the other hand we found that the heat wave altered important elements of the pipefish immune defence compared to the control condition of 18°C (Table 2 for details). The relative number of lymphocytes decreased (ANOVA with subsequent Tukey HSD, p adj. = 0.0001) in warm water indicating a downregulation of the adaptive immunity pathway. In contrast, the number of monocytes that are characteristic for innate immunity increased (ANOVA with subsequent Tukey HSD, p adj.<0.0001) in warm water compared to the control group. The water temperature also had a significant effect on the proliferation of lymphocytes. We found a significantly higher proportion of cells in S-G_2-M stage in warm water (ANOVA with subsequent Tukey HSD, p adj. = 0.007) than in the control group and consequently more cells in resting stage G₀₁ in the control treatment of 18°C than in the warm water (ANOVA with subsequent Tukey HSD p adj. = 0.02, Figure 2).

Pipefish from all three populations and parasite treatments (sympatric, allopatric and control), when exposed to a simulated heat wave had fewer cells of the adaptive immune system (lymphocytes) compared to the same groups that were kept at 18°C. The proportion of the innate immune cells (monocytes) was higher in all 3 treatment groups in the 18°C water compared to individuals kept in 25°C water (Figure 2). Moreover, cells of the adaptive immune system proliferated faster in the warm water. Compared to the constitutive and fast activation of the innate immune response the activation of the adaptive immune system is both slow and costly ^[47]–^[49]. Innate and adaptive immune pathways depend in several steps on each other which may result in resource allocation trade-offs in either the fast innate or the costly specific immune response ^[50]. Our findings of a higher activity of lymphocytes (Figure 2) under a heat wave compared to normal temperature combined with a lower proportion of lymphocytes suggest that the adaptive immune response is delayed under increasing environmental temperatures. Alternatively, a higher bacteria prevalence and virulence in warm waters could explain the decreased lymphocyte count at an increased proliferation, as cells may leave the head kidney to fight bacteria infections in the periphery ^[51].

Contrary to other studies, in which infectivity positively correlated with increasing temperature ^[52], ^[53], the temperature treatment had no effect on the infectivity of C. lingua. Infection patterns were consistent across treatments and not influenced by the heat wave, which indicates that local adaptation between genotypes was more important than first order temperature effects on the parasite. Due to the complex life cycle, the infectivity of C. lingua depends upon various biotic and abiotic factors acting on all stages. For example first intermediate hosts (invertebrates) of C. lingua were shown to have a decreased immune response under a scenario of global change ^[17]. The cercariae production is likely to be higher in warmer waters ^[32], However, a trade-off between cercarial activity and longevity was identified recently in 20°–25°C warm water in a different species ^[33].

C. lingua is locally adapted to its host as more parasites from the same location (sympatric) infected hosts than from a different (allopatric) location. This is of particular importance as C. lingua is a generalist parasite ^[36], with a broad range of different host species that has a mobile final bird host. Local adaptation is counterintuitive in generalists as they need to keep the potential to infect several hosts ^[54]. However, due to its final bird host and thus high geographic dispersal, it seems that C. lingua might be ahead in the evolutionary arms race, possibly due to a higher gene flow between adjacent parasite populations and therefore the parasite may profit from an inflow of new alleles that promote adaptation to new host resistance alleles ^[55].

Our heat wave experimentally displaced the different host – parasite combinations to a new thermal environment. If patterns of local adaptation were strongly influenced by the environment, as proposed by the geographic mosaic theory ^[12], ^[13], we expect that at least some patterns would alter in magnitude or sign. Yet, regardless of the temperature conditions all combinations of allopatric and sympatric hosts and parasites revealed a consistent sign of local adaptation, with sympatric parasites infecting their hosts better than allopatric ones.

This is, to the best of our knowledge, the first study that investigates immune parameters of the adaptive/innate response in a teleost under a realistic heat stress scenario to assess possible immune system suppression under global change. Our latter findings of a marked alteration under a realistic scenario of temperature stress highlight the importance of studies on ecological immunity ^[56]–^[58] to be included as response variables in global change research. We are also not aware of any study that measures the outcome of combinations of coevolved populations of hosts and parasites under an extreme event associated with global change. Because we only tested two environmental conditions, over a relatively short time, we cannot rule out that the patterns of local adaptation observed here are changed by the multitude of factors that are affected by global change.