key: cord-0275744-6jpoycu4
authors: Goehlich, Henry; Sartoris, Linda; Wagner, Kim Sara; Wendling, Carolin C.; Roth, Olivia
title: Pipefish locally adapted to low salinity in the Baltic Sea retain phenotypic plasticity to cope with ancestral salinity levels
date: 2020-11-12
journal: bioRxiv
DOI: 10.1101/2020.11.12.379305
sha: c5067da311d1168a4da15a292136a8635623cb36
doc_id: 275744
cord_uid: 6jpoycu4

Genetic adaptation and phenotypic plasticity facilitate the invasion of new habitats and enable organisms to cope with a rapidly changing environment. In contrast to genetic adaptation that spans multiple generations as an evolutionary process, phenotypic plasticity allows acclimation within the life-time of an organism. Genetic adaptation and phenotypic plasticity are usually studied in isolation, however, only by including their interactive impact, we can understand acclimation and adaptation in nature. We aimed to explore the contribution of adaptation and plasticity in coping with an abiotic (salinity) and a biotic (Vibrio bacteria) stressor using six different populations of the broad-nosed pipefish Syngnathus typhle that originated from either high or low saline environments. We hypothesized that wild S. typhle populations are locally adapted to the salinity and prevailing pathogens of their native environment, and that short-term acclimation of parents to a novel salinity may aid in buffering offspring phenotypes in a matching environment. To test these hypotheses, we exposed all wild caught animals, to either high or low salinity, representing native and novel salinity conditions and allowed animals to mate. After male pregnancy, offspring was split and each half was exposed to one of the two salinities and infected with Vibrio alginolyticus bacteria that were evolved at either of the two salinities in a fully reciprocal design. We investigated life history traits of fathers (offspring survival, offspring size) and expression of 47 target genes in mothers and offspring. Pregnant males originating from high salinity exposed to low salinity were highly susceptible to opportunistic fungi infections resulting in decreased offspring size and number. In contrast, no signs of fungal infection were identified in fathers originating from low saline conditions suggesting that genetic adaptation has the potential to overcome the challenging conditions of low salinity. Genetic adaptation increased survival rates of juveniles from parents in lower salinity (in contrast to those from high salinity). Juvenile gene expression indicated patterns of local adaptation, trans-generational plasticity and developmental plasticity. The results of our study suggest that pipefish locally adapted to low salinity retain phenotypic plasticity, which allows them to also cope with ancestral salinity levels and prevailing pathogens.

precipitation in the northern part may cause a decrease by up to 30% in surface salinity by 106 pathogens may compensate for the previously observed drop of immunological activity in 156 case of exposure to decreasing salinities (Birrer, Reusch et al. 2012 , Poirier, Listmann et al. 157 2017 and, hence, has the potential to reduce the negative impact of pathogens like Vibrio 158 bacteria (Roth, Keller et al. 2012) . 159

To explore how pipefishes have genetically adapted to long-term salinity changes 160 and how this adaptation influences their phenotypic plasticity to cope with short-term shifts 161 in salinity, we compared the potential of pipefish originating from either high or low salinity 162 environments to react towards salinity shifts with developmental and trans-generational 163 plasticity. Furthermore, we investigated how adaptation and acclimation of the pipefish host 164 and the bacterial Vibrio pathogen to high and low salinity changes the host-pathogen 165 interaction. We tested the following hypotheses: 1) S. typhle populations are genetically 166 adapted to the salinity in their local habitat, 2) adaptive trans-generational plasticity in 167 matching parental and offspring salinity results in enhanced juvenile survival and matching 168 gene expression pattern in the parental and offspring generation, 3) S. typhle populations 169 locally adapted to low salinity have reduced phenotypic plasticity and are not able to cope 170 with ancestral salinity levels, and 4) bacterial virulence is higher at low salinity. 171

To investigate how S. typhle have adapted towards their local salinity and local pathogens 172 in the past (genetic adaptation) and to assess their consecutive acclimation potential 173 (phenotypic plasticity) towards salinity shifts and their immune response towards a bacterial 174 infection, we collected six S. typhle populations in the Baltic Sea. Fish were collected at 175 three sampling sites with high saline conditions and at three sampling sites with low saline 176 conditions. In a laboratory aquaria experiment, animals were exposed to either their native 177 salinity (high or low respectively) or the salinity of the other three populations (novel with a V. alginolyticus strain that evolved for 90 days either at low or high salinity in the 181 laboratory. In addition to life history traits and mortality, we investigated the expression of 182 47 target genes involved in (i) general metabolism, (ii) immune response, (iii) gene 183 regulation (DNA and histone modification) and (iv) osmoregulation. Table  189 1). Three sampling sites are characterized by relatively high salinity conditions (14) (15) (16) (17) 190 high salinity origin; H) and three sampling sites by relatively low salinity conditions (7 -11 191 PSU; low salinity origin, L; Table 1 ). Salzhaff was assigned the category low because 192 salinity drops are common after rainfall accompanied with freshwater discharge due to 193 enclosed morphology of the inlet. Therefore, pipefish in Salzhaff are likely to be exposed to 194 salinity levels below 10 PSU. A minimum of 30 non-pregnant males and 30 females were 195 caught snorkeling with hand nets at each sampling site at depths ranging between 0.5 and 196 2.5 m. At each sampling site, water temperature and salinity were measured from water 197 collected about 1 m below the surface using a salinometer (WTW Cond 330i). 198 200 Pipefish were transported in large aerated coolers to the aquaria facilities of the 204 GEOMAR (Westshore) in Kiel (Germany). Females and males were separated and kept in 205 groups of 5-7 per 80-liter tank resulting in a total number of 36 tanks. These tanks were 206 connected to two independent circulating water systems containing either high saline (15 207 disconnected from the circulation system during the time of salinity acclimation. 219

Subsequently, four to six randomly chosen males and four to six females originating from 220 the same sampling site and acclimated to the same salinity, were placed together in one of 221 the 36 tanks connected to circulating water systems of either high or low acclimation salinity 222 ( Figure 1 ). During mating and male pregnancy, fish maintenance and aquaria set-up 223 remained as previously described. 224

One week after mating, pipefish males started to show signs of infection with a 225 fungus growing inside and on the brood pouch. Three weeks after mating, we visually 226 assessed and photographically documented the prevalence of fungus. 

We sampled pipefish along the Baltic Sea coast, at three sampling sites from a relatively high saline 232 environment (high origin salinity: 14 -17 PSU; dark blue circles; subsequently labeled as italic H): 233 1) Flensburg Fjord, 2) Falckenstein Strand and 3) Fehmarn and three sampling sites with a relatively 234 low salinity level (low origin salinity: 7 -11 PSU; light blue circles; subsequently labeled as italic 235 L): 4) the Salzhaff and 5) Ruegen North and 6) Ruegen South. In the laboratory males and females 236 were kept separately and acclimated to the opposing salinity (acclimation salinity: 15 PSU (H, dark 237 blue), 7 PSU (L, light blue)) or their respective native salinity. Subsequently, males and females 238 were allowed to mate and pregnant males were kept at constant conditions. Half of the F1 generation 239 was either exposed to high (h) or low (l) salinity within 24h after birth (developmental salinity). Ten 240 days post hatch juveniles were injected with Vibrio alginolyticus evolved at 15 PSU (v15) or at 7 241 PSU (v7), sham injected with sterile seawater (W) or left naive (C) (treatment). Label in italic on the 242 right side correspond to the factors that were considered in the statistical models.

Four days after mating, females were removed from the tanks and immediately 246 euthanized using anesthetic tricaine methane sulfonate (MS-222, 500 mg/L). We measured 247 standard body length and total weight and removed the gills to store them in RNAlater at 4° 248 C overnight and subsequently at -80° C. Fin clips were taken and placed in 96% Ethanol for 249 Micro-Checker identifies genotyping errors caused by non-amplified null-alleles that either 286 appear due to mutations in the primer binding regions or generally occur in fragment analysis 287 because PCR shows greater efficiency in longer sequences. GENETIX (Belkhir et al. 2004 ) 288 was used to describe the level to which the genotype frequency differed from the expected 289

Hardy-Weinberg equilibrium (HWE) frequency by calculating a global FST value as a 290 correlation of inbreeding in the substructure vs. in the entire population. For completeness, 291

pairwise FST values were calculated to display distances between pairs of haplotypes and a 292 FIS value was calculated as a correlation of inbreeding vs. random mating within the 293 population. Although GENETIX has a greater statistical power, the population structure 294 within the multi-locus genotype data was further investigated by the STRUCTURE Software for Population Genetics Inference (Pritchard et al. 2000) . Based on the Bayesian clustering 296 method, STRUCTURE creates an admixture model, which provides likelihood scores for 297 each individual of belonging to a certain population. The model was tested with varying 298 numbers of expected populations ranging from a minimum of two (high salinity vs. low 299 salinity) to a maximum of six (number of sampling stations). Visualization of the population 300 clustering was performed using the PHYLogeny Inference Package PHYLIP (Felsenstein 301 1989) . PHYLIP provides a pipeline of programs to randomize comparisons, create 302 randomized trees, which are then assembled to a final phylogeographic tree that is based on 303 the most frequent combinations found within the randomized trees. As the retrieved 304 fragment data did not provide any lineage data that allows to draw conclusions with regard 305 to a common ancestor, we created an unrooted phylogeographic tree. 306 307

To assess local adaptation to salinity and the potential of S. typhle to cope with novel 309 salinity conditions, we selected candidate genes from three different functional categories, 310

i.e. (i) immune response, (ii) metabolism and (iii) gene regulation (DNA and histone 311 modification) (Supplement 1 (Table S1)). Immune genes were further subdivided into 312 innate, adaptive and complement system genes and gene regulation genes into activating 313 and silencing genes. 314 315 2.4.1 RNA extraction and reverse transcription 316 RNA was extracted from gill tissue of adult pipefish that was stabilized in RNAlater 317 using the RNeasy® Universal Tissue kit (QIAGEN, Venlo, Netherlands). Tissue samples 318

were homogenized by adding a 5 mm stainless steel bead into each collection tube and 319 placing them into a homogenizer shaking for two times 30 seconds at 25 Hertz. Thereafter, we followed the manufacturer's protocol "Purification of Total RNA from Animal Tissues 321

Using Spin Technology". RNA concentration (extraction yield) and purity of the samples 322 were checked by spectrophotometry (NanoDrop ND-1000 Spectrometer; Peqlab, Erlangen, 323 Germany). Protein contamination was quantified using the adsorption ratio of 260/280 nm 324 (target > 2.0) and the ratio 260/230 nm (target > 1.8) was used to detect organic 325 contamination. A fixed amount of RNA (300 ng/sample = 50 ng/µl) was then reverse 326 and gene, two no-template controls (H2O), one control for gDNA contamination (-RT) and 344 one between plate control. 345

Within the first 24 hours after birth, half of the juveniles from each clutch was 349 exposed to native salinity conditions and half to novel salinity conditions in a fully reciprocal 350 design. Juveniles were fed twice a day with freshly hatched, nutrient enriched (Aqua Biotica 351 orange+TM) Artemia salina nauplii. Siblings were kept together in one non-aerated 1.5 l 352 tank, of which one third of the water was exchanged daily. Once a day, left-over food was 353 removed using single-use pipettes and mortality was documented. needle. Subsequently, all juvenile siblings with the same treatment were placed in one 500 361 ml Kautex bottle containing seawater with the respective salinity of the 1.5 l tanks. Survival 362 of juveniles was documented for six days and fish maintenance was according to the 363 procedure described for 1.5 l tanks. One day post infection, one juvenile from each treatment 364 (Kautex bottle) was euthanized and decapitated to assess expression of candidate genes. 365

Standard body length was measured and whole-body samples were stored in RNAlater 366 overnight at 4° C and subsequently at -80º C. 367 368

The Vibrio alginolyticus strain K01M1 used for injection of pipefish juveniles was 370 isolated from a healthy pipefish caught in the Kiel Fjord (Roth, Keller et al. 2012 ) and fully 371 sequenced (Chibani, Roth et al. 2020 ). The strain was evolved for 90 days either at 15 or 7 PSU (medium 101: 0.5% (w/v) peptone, 0.3% (w/v) meat extract, 1.5% (w/v) or 0.7% (w/v) 373 NaCl in Milli-Q deionized water) (Goehlich, Roth et al., unpublished data) . We used the 374 same strain and evolved it at two different salinities to ensure that salinity is the only driver 375 for differences in bacterial virulence, which can potentially also be influence by the presence 376 of filamentous phages (Waldor and Mekalanos 1996 , Ilyina 2015 , Chibani, Hertel et al. 377 2019 . 378

After 90 days the bacterial populations were diluted and plated onto Vibrio selective 379

Thiosulfate-Citrate-Bile-Saccharose (TCBS) agar plates (Fluka AnalyticalTM). The next 380 day, single colonies from each plate were picked and grown overnight in medium 101 with 381 the respective salinity. Subsequently, cultured bacteria were stored at -80°C as 33% glycerol 382 stocks. For the infection experiment, part of the glycerol stocks were plated onto TCBS agar 383 and one clone was grown in a 50 ml Falcon tube containing 30 ml medium 101 in the 384 respective salinity for 24 hours, at 25 °C with shaking at 230 rpm. Overnight cultures were 385 centrifuged for 20 min at 2000 rpm. The supernatant was discarded and the cell pellet was 386 resuspended in 3 ml sterile seawater (7 or 15 PSU respectively) to achieve similar bacterial 387 and histone modification) as described above for female pipefish S. typhle (Section 2.4). 399

Compared to female gene expression, eleven genes from the categories (i)-(iii) were 400 replaced by osmoregulation genes (iv). We selected osmoregulatory genes from teleost 401 studies (S 2) and designed specific primers with Primer3Web (Koressaar and Remm 2007, 402 Untergasser, Cutcutache et al. 2012) (S 3). RNA extraction and quantification of gene 403 expression were conducted as described before with the following modifications due to a 404 higher RNA yield: the fixed amount of RNA that was reverse transcribed into cDNA was 405 400 ng/sample (67 ng/µl) instead of 300 ng/sample (50 ng/µl) and pre-amplified cDNA was 406 diluted 1:10 and instead of 1:20. 407 408

All statistical analyses and visualizations were performed in the R 3.6.1 environment 410 (RCoreTeam 2020). 411 412

We used two-way ANOVAs to assess size and weight differences between adults as 414 well as differences in clutch size and in total length between juveniles at 10 days post-hatch. 415

Fixed factors included origin salinity (Salinity at sampling sites of origin two levels: High 416 or Low), acclimation salinity (High or Low), sex of the pipefish (male or female) and the 417 sampling site (Flens, Fehm, Falck, Salz, RuegN or RuegS) nested in origin salinity. ANOVA 418 of clutch size additionally included the average body length of males exposed to a given 419 treatment. Homogeneity of variances was tested by Fligner test and normal distribution of 420 data by using the Cramer-von Mises normality test. The clutch size was square root 421 transformed to achieve normal distribution of residuals. 422

We performed two spearman-rank correlations using the function ggscatter salinity measured at the sampling site on the day of capture as well as 2) between clutch size 425 and average male size of sampling site. The clutch size of males originating from high 426 salinity and acclimated to low salinity conditions was removed from the correlation due to 427 fungus infection. Post-hoc tests were carried out using Tukey's "honest significant Ten days post-hatch endpoint mortality of juveniles was analyzed as a ratio of "alive" 467 vs "dead" fish using a generalized linear model (package: lme4, function: glm) with 468 binomial error and the following fixed factors: Origin salinity (High or Low), acclimation 469 salinity (High or Low) and developmental salinity (high or low) and the sampling site nested 470 in origin salinity. Significance was tested using ANOVA type two partial sums of squares, 471 and models were simplified using Akaike information criterion (AIC) (Akaike, 1976). Post-472 hoc tests were carried out using Tukey's honest significant difference (TukeyHSD, package:

(control (c), sea water injection (sw), Vibrio 7 PSU (v7) and Vibrio 15 PSU (v15) injection) 476 as an additional factor. 477 

We found an interaction in the total length of adult pipefish between origin salinity 501 and acclimation salinity (ANOVA F1,320 = 7.4, p < 0.01) indicating that parental acclimation 502 salinity negatively affects growth of adult pipefish depending on the origin salinity. There 503 was a trend that adults from high origin salinity grew slower at low acclimation salinity 504 compared to high acclimation salinity (Tukey HSD, HH -HL: p = 0.085; S 6b), whereas 505 acclimation salinity did not affect size of pipefish from low origin salinity (Tukey HSD, LL 506 -LH: p = 0.535). Furthermore, all pairwise comparisons suggest that pipefish from high 507 origin salinity are in general larger than pipefish from low origin salinity (Tukey HSD, LL 508 -HH: p < 0.001, LL -HH: p < 0.001, LH -HL: p < 0.001). The significant factor sampling 509 site, which was nested in origin salinity (ANOVA F4,320 = 11.2, p < 0.01) indicates that 510 individuals from Salzhaff were larger compared to individuals from Ruegen North and 511

Ruegen South but did not differ from pipefish caught at the high origin salinity (Tukey HSD, 512 Salz -RuegN: p < 0.001; Salz -RuegS: p < 0.001; S 6c). 513

The correlation between the salinity at the sampling site and the size of the adults, 514

i.e. length (Figure 3 B) and weight ( Figure S6 ) suggest that pipefish from low origin salinity 515 were smaller. The total length and weight in these plots are not corrected for age, which has 516 not been assessed. However, pipefish are usually all in the same age cohort when they are 517 caught in spring. Most of them were born the summer before and reached sexual maturity 518 around the time of catching. 519 

Males from a high origin salinity kept at high acclimation salinity had the largest 546 clutch size (HH, mean ± s.d., 41.8, ± 23.4, Tukey TSD; S 7b) followed by males from low 547 origin salinity kept at high salinity (LH, 27.8 ± 13.2) or low salinity (LL, 25.2 ± 16.9), which 548 corresponds to the lower body size at low salinity ( Figure 3 ). Pipefish from high origin 549 salinity exposed to low acclimation salinity were frequently infected by a brood pouch 550 fungus which reduced the clutch size (HL 19.4 ± 15.2; Tukey TSD, HL-HH, t = -4.8, p < 551 0.001;S 7b). In contrast, acclimation salinity did not affect clutch size of parents from low 552 origin salinity (Tukey TSD, LL-LH, t = -0.4, p = 0.976; S 7b) This divergent patterns caused 553 a origin salinity:acclimation salinity interaction (ANOVA F1,109 = 9.0, p = 0.003; S 7a). 554

Higher clutch size, was in general driven by a larger total body length of male pipefish 555 (Spearman rank correlation, R = 0.43, p < 0.001; Figure 5 

Clutch size R = 0.43, p = 2.1e-0.5

smaller 578

In the first ten days after hatching, patterns of juvenile survival suggest an origin 579 salinity:acclimation salinity interaction (GLM, χ 2 1 = 6.1, p = 0.031; S 9a). Whereas, survival 580 of juveniles from parents continuously exposed to the same salinity (Tukey HSD, LL -HH: 581 z = -3.4, p = 0.783; S 9b) or non-matching origin and acclimation salinity did not differ (LL 582 -LH: z = -4.7, p = 0.842; HH -HL: z = -2.1, p = 0.148), juveniles from high origin salinity 583 parents exposed to high acclimation salinity in the lab (LL) had higher survival rates 584 compared to juveniles from high origin salinity exposed to low acclimation salinity (Tukey 585 HSD, LL -HL: z = -3.4, p = 0.043; S 9b). The origin salinity:sampling site effect suggest 586 that patterns at single sampling sites differ. In particular, Flensburg offspring exposed to 587 high developmental salinity had reduced survival rates, when parents were acclimated to low 588 instead of high salinity (Tukey's HSD; Hh -Lh: z = -4.4, p = 0.046; S 9b). Following the 589 same pattern of non-matching acclimation salinity, Ruegen South offspring exposed to low 590 developmental salinity had reduced survival, when parents were kept at high acclimation 591 salinity (Tukey HSD; Hl -Ll: z = 5.8, p < 0.001). This suggests that exposure of parents to 592 novel salinities can negatively impact juvenile survival when juveniles experience salinity 593 conditions, which did not match parental acclimation salinity. 594 Ten days after hatching, juveniles from high origin salinity were larger (3.18 cm ± 612 0.37, n = 408) than juveniles from low origin salinity sampling sites (2.95 cm ± 0.37, n = 613 405) (origin salinity, ANOVA F1,782 = 86.2, p < 0.001; S 10). While acclimation salinity, i.e. 614 mating and male pregnancy, had no effect on size of juveniles ( 

Ten days post hatch, juvenile pipefish were challenged either with Vibrio 621 alginolyticus bacteria evolved at 15 PSU, 7 PSU, autoclaved seawater (sham injection) or 622 not treated at all (control) and survival was measured six days post infection, i.e. 623 approximately 16 days post hatch. Non-challenged control groups had the highest survival 624 rates (Mean ± s.d.; 83.0% ± 32.2; Figure 7) . The injection itself decreased survival of 625 juveniles by at least 10% in all salinity treatments combined, regardless whether seawater 626 (66.9% ± 38.2), Vibrio evolved at 15 PSU (73.0% ± 36.8) or 7 PSU (66.7% ± 38.6) was 627 administered. Vibrio strains evolved at 15 PSU caused a higher mortality in juveniles from 628 high origin salinity regardless of acclimation salinity compared to juveniles from low origin 629 salinity with low parental acclimation salinity (origin salinity x acclimation salinity x 630 treatment, GLM, χ 2 1 = 13.0, p = 0.005; S 11; Tukey HSD; LLv15 -HHv15: z = -3.4, p = 631 0.046; LLv15 -HLv15: z = -3.5, p = 0.038). When fathers from low origin salinity were 632 exposed to high acclimation salinity these positive effects on offspring survival were lost 633 (Tukey's HSD, LHv15 -HHv15 z = -1.5, p = 0.971). This suggests that mis-matching 634 salinity levels between the parental and juvenile generation can lead to reduced survival 635 rates and that juveniles of fathers from low salinity levels have higher survival rates 636 compared to juveniles of fathers from high salinity. 

The confidence ellipse explains 20 % of the variability. Exposing parents to low acclimation salinity led to an expression induction of two genes in 700 juveniles (PERMANOVA, complement; acclimation salinity F1,523 = 4.6, p = 0.014). Both 701 genes are associated with the complement system: Complement component 3 (c3, 702 complement system activation) and Complement component 9 (c9, membrane attack 703 complex). In addition to this parental effect a trans-generational effect was observed as a 704 parental acclimation salinity : developmental salinity interaction effect on adaptive immune 705 gene expression (PERMANOVA, adaptive F1,523 = 2.6, p = 0.034). Gene expression was 706 lower in four out of seven adaptive immune genes when acclimation salinity and 707 developmental salinity were matching compared to non-matching conditions (S 12): Human 708 immunodeficiency virus type l enhancer 2 (hivep2, transcription factor, MHC enhancer 709 binding) and 3 (hivep3; transcription factor, MHC enhancer binding), B-cell receptor-710 associated protein (becell.rap31, T-and B-cell regulation activity) and immunoglobulin 711 light chain (igM; antigen/pathogen recognition). The reduction in gene expression of 712 immune genes can hint at a reduced stress level in offspring fish when parents are acclimated 713 to the same salinity as their offspring. 714

Finally, we wanted to test whether genetic background, i.e. origin salinity, 725 acclimation salinity of the parents and the developmental salinity influenced the ability of 726 juveniles to cope with infections of the opportunistic pathogen V. alginolyticus, which 727 evolved in the lab at either 15 or 7 PSU. However, we found no interaction between any 728 salinity regime of parents or juveniles interacting with gene expression after juvenile 729

infection. An injection, regardless of the component, i.e. autoclaved seawater, V. 730 alginolyticus evolved at 15 PSU or 7 PSU caused similar changes in gene expression patterns 731 that could only be differentiated from the untreated control group. In 24 genes injections 732 caused a higher gene expression (PERMANOVA, all genes; F3,523 = 11.3, p = 0.001; S 13), 733

including genes from all groups. In five genes, injections caused a lower gene expression 734 compare to the control group. Using post-hoc tests on ANOVAs of single genes, we found 735 no differences in gene expression between the three injection treatments. 736 737

In the present study we investigated the role of genetic adaptation and phenotypic 739 plasticity as well as their interaction on the ability of the broad-nosed pipefish Syngnathus | 36 offspring size, while small parents may rather invest in survival (Nygard, Kvarnemo et al. 776 2019) via genetically determined gene expression patterns. 777

Such genetically determined gene expression patterns that are inherited from 778 generation to generation, can be indicative signs for local adaptation (Larsen, Schulte et al. 779 2011 , Fraser 2013 , Heckwolf, Meyer et al. 2020 . Females from high origin salinity had an 780 induced baseline innate immune gene expression compared to females originating from low 781 salinity environment. This induced innate immune gene expression pattern was inherited to 782 their offspring: juveniles from animals caught from high saline environments generally had 783 an induced expression of innate immune genes. We suggest that the observed induction of 784 innate immune genes in pipefish originating from high saline origins is indicative for the 785 existence of sufficient resources allowing to keep the innate immune response at a high 786 baseline level. This can result in a faster and stronger and eventually more effective immune 787 response. In contrast, pipefish from low origin salinity may rather suffer stress induced by 788 the above stated resource allocation trade-off, which decreases the resources available for 789 the innate immune system. 790

This suggests that osmoregulation is costly 756 (Rolfe and Brown 1997, Boeuf and Payan 2001) and that the negative impact of low salinity 757 can potentially not be fully compensated through local adaptation This implies that trade-758 offs for osmoregulation reduce growth rates, which ultimately correspond to a decreased 759 fitness. Studies of other marine teleosts that originated from fully marine environments

PSU) are possible, especially when close or slightly above isosmotic levels

the parental phenotype 764 correlated with the body weight or length of the offspring. The heritability of morphological 765 traits was suggested to be lower for ectotherms than for endotherms