Cockle as Second Intermediate Host of Trematode Parasites: Consequences for Sediment Bioturbation and Nutrient Fluxes across the Benthic Interface

: Trematode parasites are distributed worldwide and can severely impact host populations. However, their inﬂuence on ecosystem functioning through the alteration of host engineering be-haviours remains largely unexplored. This study focuses on a common host parasite system in marine coastal environments, i.e., the trematode Himasthla elongata, infecting the edible cockle Cerastoderma edule as second intermediate host. A laboratory experiment was conducted to investigate the indirect effects of metacercarial infection on sediment bioturbation and biogeochemical ﬂuxes at the sediment water interface. Our results revealed that, despite high parasite intensity, the sediment reworking and bioirrigation rates, as well as nutrient ﬂuxes, were not impacted. This ﬁnding was unexpected since previous studies showed that metacercarial infection impairs the physiological condition of cockles and induces a mechanical obstruction of their feet, thus altering their burrowing capacity. There are several explanations for such contrasting results. Firstly, the alteration of cockle behavior could arise over a longer time period following parasite infection. Secondly, the modulation of cockle bioturbation by parasites could be more pronounced in older specimens burying deeper. Thirdly, the intensity of the deleterious impacts of metacercariae could strongly vary across parasite species. Lastly, metacercarial infection alters cockle ﬁtness through an interaction with other biotic and abiotic environmental stressors.


Introduction
Parasites represent 40% of the total known animal species, infecting autotrophic and heterotrophic hosts in all terrestrial and aquatic environments [1,2]. These ubiquitous organisms can profoundly impact ecosystem functioning, both directly by affecting ecosystem properties and indirectly by changing the functional role of their hosts [3]. More precisely, indirect effects may result from the limitation of dominant or keystone host species abundance (i.e., density-mediated effects) or from the modification of host-specific activities, which disproportionally influence ecosystem processes and functions (i.e., traitmediated effects) [4][5][6][7]. To date, the indirect ecological consequences, at the ecosystem level, of parasite-induced engineering behavior alteration (sensu Jones et al. [8]) remain poorly known and quantified, particularly in marine environments with contradictory observations depending on the studied processes and host-parasite system. Yet, it seems particularly important to assess how and to what extent parasites can indirectly influence ecosystem functioning, as climate change, particularly global warming, is expected to strongly enhance parasite expansion and transmission, decrease host resistance, and overall exacerbate detrimental effects [9,10].
A common marine parasite-host system, Himasthla elongata-Cerastoderma edule, was used in controlled laboratory experiments to investigate the potential indirect effects across the benthic boundary were expected in the presence of parasitized cockles, compared to nonparasitized conspecifics.

Field Sampling
Specimens of Cerastoderma edule and sediment (~80 L) were collected by hand in August 2020 on an intertidal flat at "Ile aux Oiseaux", Arcachon Bay, France (44 • 42'15.4" N, 1 • 11'05.1" W). In the laboratory, cockle shell length was measured to the lesser mm using a digital caliper, and cockles were selected according to their size. A total of 280 individuals ranging in size from 9 to 14 mm, corresponding to young cockles characterized by a low natural trematode infection [34], were stored in tanks with running seawater (~20 • C) for two days. The fresh sediment was sieved through a 1 mm mesh to remove macrofauna, larger debris, and shells, homogenized, and kept in separate tanks with seawater from the sampling site at ambient temperature for one day. Periwinkles (Littorina littorea) were collected in July 2020 in Norsminde Fjord (Denmark), where the prevalence of H. elongata is usually high. In the laboratory, periwinkles were individually kept in small plastic containers filled with seawater and exposed to constant artificial white light for 12 h to induce cercaria emergence [53]. Parasitized periwinkles were identified under a stereomicroscope by the presence of cercariae swimming in seawater. They were then maintained in aquaria at 15 • C to avoid cercaria emission and regularly fed with green macroalgae (Ulva spp.) until the experimental infection of cockles.

Experimental Infection
Two days after field sampling, 240 cockles were randomly selected and evenly distributed among 20 containers (14 cm × 9 cm × 3 cm) filled with 350 mL of seawater (salinity = 31.5, 20 • C), corresponding to 12 cockles per container. In 10 containers, three periwinkles infected by the trematode Himasthla elongata were immediately introduced in each container [51]. They were continuously illuminated by artificial white light for 24 h (the seawater temperature then rapidly increased to 24 • C) to stimulate the emission of cercariae, which encysted in cockles, their second intermediate host. Cockles were left undisturbed for 6 h in a 15 • C tempered room following periwinkle removal in order to allow complete encystment of metacercarial larvae. The same protocol was conducted in 10 containers without periwinkles. The whole experimental process was reiterated three days later to ensure a higher and more homogeneous cockle infection and to avoid cercaria virulence variability across H. elongata strains [54,55].

Experimental Procedure
Twenty-eight experimental units (PVC tube, height 30 cm, internal diameter 9.3 cm) filled with 20 cm of homogenized sieved sediment were placed in a tidal mesocosm, which consisted of a large tank containing 600 L of natural seawater. An artificial tidal regime (6 h low tide/6 h high tide) was generated by pumping 200 L of water in a second tank, corresponding to a variation in the water level of about 20 cm. The transition between high and low tide lasted about 3 min and was carefully adjusted in order to avoid sediment resuspension. The seawater temperature (19 ± 1 • C) and salinity (33 ± 1) remained constant throughout the entire duration of the experiment. After two days of sediment stabilization, seven cockles that were not experimentally infested (nonparasitized "NP" treatment) were placed on the sediment surface of 12 experimental units (i.e., a total of 84 nonparasitized cockles) while seven cockles which were experimentally infested (parasitized "P" treatment), were also placed on the sediment surface of 12 other experimental units (i.e., a total of 84 parasitized cockles) (n = 12 for both "NP" and "P" treatments). Cockles displayed the same shell length distribution among the different experimental units. The corresponding density (1030 ind·m −2 ) was in the range commonly reported in the field for individuals of similar shell length [56][57][58]. Additionally, four control units were kept without cockles. At the beginning of the experiment, three days after the introduction of cockles, 10 g of luminophores (fluorescent particle tracers, d 50 = 250 µm, 2.6 g·cm −3 density) was homogeneously spread at the sediment-water interface of each experimental unit (sediment reworking analysis). Simultaneously, 630 g of sodium bromide (NaBr) was dissolved in the water of the tidal mesocosm (bioirrigation analysis). All experimental units were incubated in the tidal mesocosm during nine days following the addition of particulate and dissolved tracers.

Sediment Characteristics
The sediment grain size was determined using a laser diffraction microgranulometer (Malverns Panalytical ® Mastersizer hydro 2000G, Malvern, UK). The proportion of the mud fraction (<63 µm) and the median grain size (d 50 ) were used as synthetic parameters. Sediment organic carbon (OC) and total nitrogen (TN) contents were assessed on freezedried homogenized samples. For OC measurement, the sediment was decarbonated with 1.2N HCl [59]. OC and TN were analyzed using a CN auto-analyzer (Thermoscientific FlashSmart, Bremen, Germany). Sediment porosity was assessed in an additional experimental unit. At the end of the experiment, the sediment column was vertically sliced into 0.5 cm thick layers from the surface to 5 cm depth and into 1 cm thick layers between 5 and 11 cm depth. Slices were weighted and then dried at 60 • C for four days. The loss of water (assessed by the difference between wet and dry weight) was corrected for sea-salt content assuming a sediment bulk density of 2.65 g·cm −3 [60].

Benthic Flux Measurement
After nine days, oxygen (O 2 ), ammonium (NH 4 + ), nitrates and nitrites (NO 2 − + NO 3 − , noted hereafter as NO x ), and dissolved silicate (dSi) fluxes across the sediment-water interface were quantified in total darkness. Twenty milliliters of overlying water was sampled in each microcosm for the measurement of initial nutrient concentrations, just before they were hermetically sealed with a PVC lid. The O 2 concentration was then continuously measured using a mini-optode (OXROB10, Pyroscience ® , Aachen, Allemagne) connected to a Firesting Oxygen Meter (Pyroscience ® , Aachen, Allemagne) placed through the lid 5 cm above the sediment surface. The optode was previously linearly calibrated between the oxygen concentration of the overlying water (100% saturation) measured by Winkler titration [61] and zero oxygen in a sodium ascorbate solution. A constant homogenization of the overlying water was ensured by a Teflon-coated magnetic stirring bar (~100 rpm) attached to the lid. As soon as the O 2 concentration reached 75% of its initial value, the incubation was stopped, and the overlying water was sampled for final nutrient measurements using a Quattro-AXFLOW auto-analyzer (Norderstedt, Germany).
The total oxygen uptake (TOU) and nutrient fluxes (mmol·m −2 ·day −1 ) were finally computed from concentration decreases within each experimental unit, according to the following equation: where a is the slope (mmol·m −3 ·day −1 ) of O 2 or nutrient concentrations as a function of time (day), V is the volume of overlying water (m 3 ), and S is the surface area of the sediment-water interface (m 2 ).
2.6. Quantification of Bioturbation Rates 2.6.1. Bioirrigation All experimental units were vertically drilled on one side at 1 cm intervals. For the duration of the experiment, the openings (2 mm diameter) were sealed with a strip of hydrophobic adhesive tape. Immediately following biogeochemical flux measurements, 500 µL of porewater was extracted at 0.5 to 11.5 cm depth with rhizon samplers (length 9 cm, mean pore size 0.15 µm) and stored at 4 • C until analyses. After oxidization with chloramine-T in the presence of phenol red, bromide concentrations were determined spectrophotometrically by fitting the absorbance measured at 595 nm to a standard curve [62,63].
An enhanced diffusion factor (ε) was then estimated by fitting vertical bromide profiles with the enhanced diffusion model [64]. The tracer concentration C (mM) in the sediment is described by Equation (2).
where C is the bromide concentration in the sediment (mM), t is the incubation time (day), x is the depth in the sediment (cm), D s is the molecular diffusion coefficient, φ is the porosity, and ε is a factor of enhanced diffusion (i.e., ε is equal to 1 when solute transport is exclusively driven by molecular sediment diffusion). ε was estimated by convergent iterations and weighted least-squared regressions of model prediction on observed vertical bromide profiles.

Sediment Reworking
Immediately after porewater sampling, the sediment column was vertically sliced into 0.5 cm thick layers from the surface to 5 cm depth. During slicing, cockles were gently removed and kept for parasite diagnosis (see below). The sediment of each slice was then freeze-dried and homogenized. A subsample of 3 g was spread in a Petri dish and photographed under UV light using a digital camera (Nikon ® D100, Tokyo, Japan) fitted with a yellow filter. Luminophore pixels were counted for each image after a binarization step (based on the RGB level) using AviExplore software [65].
Sediment reworking rates (D b , cm 2 ·year −1 ) were computed by fitting a one-dimensional biodiffusive model to luminophore vertical profiles (see [66] for further details) according to Equation (3).
where C is the proportion of luminophore (%), t is the time (year), and z is the vertical depth in the sediment column (cm). D b (cm 2 ·year −1 ) was estimated as a function of convergent iterations and weighted least-squared regressions of model prediction on observed vertical luminophore profiles. In addition, the proportion of the sediment surface reworked by the cockles (PRS) was assessed by analyzing pictures of experimental units taken at the beginning and at the end of the experiment (i.e., just before benthic flux measurements) using AviExplore software [65].

Parasite Diagnosis
The 168 cockles were measured once more and dissected for trematode inspection. The flesh was removed from the shell, squeezed between two glass slides, and observed under a stereomicroscope. Metacercariae, all belonging to the Himasthla genus, were counted in the whole body.

Data Analysis
Results are generally reported as the mean ± standard error (SE) of n replicate measurements. A Wilcoxon-Mann-Whitney nonparametric test was used to compare the total abundance of metacercariae belonging to the Himasthla genus in "nonparasitized" (n = 84) and "parasitized" (n = 84) cockles. Differences among treatments ("control" (n = 4), "nonparasitized" (n = 12), and "parasitized" (n = 12)) concerning bioturbation rates (both bioirrigation and sediment reworking) and biogeochemical fluxes (oxygen, ammonium, nitrates and nitrites, and silicate) were assessed using one-way analysis of variance (ANOVA). Prior to analysis, the homogeneity of variance and the normality were tested using Levene and Shapiro tests, respectively. For bioirrigation, sediment reworking, and proportion of reworked surface (PRS), these conditions were not respected, and the data were transformed using the Boxcox function to compute an optimal power transformation. A pairwise t-test was performed when the ANOVA was significant (p < 0.05) using Bonferroni correction (p-value adjusted for multiple comparisons), to determine which treatment differed. A Kruskal-Wallis test was conducted for bioirrigation depth only when conditions of appli-cation were not respected despite transformation. Lastly, a principal component analysis (PCA) was applied to the entire standardized data (bioturbation rates and biogeochemical fluxes at the sediment water-interface) to explore a possible correlation between measured variables and treatments (NP vs. P). All statistical analyses were performed using the open-source program R (v3.6.1) in R studio (v1.3.1056) (www.R-project.org, accessed on 1 August 2020.

Parasite Diagnosis
Visual observations carried out throughout the experiment revealed that both nonparasitized ("NP") and parasitized ("P") cockles buried rapidly (<1 h) and remained just below the sediment surface for most of the 9-day period. Nonparasitized cockles were slightly naturally infected with 4.5 ± 0.4 metacercariae per individual. In contrast, parasitized cockles were significantly more infected with 74.2 ± 6.5 metacercariae per individual (Wilcoxon, p < 0.001). No additional parasite species were present in both parasitized and nonparasitized cockles. All cockle specimens were found alive during core slicing at the end of the experiment.

Sediment Characteristics
The sediment consisted of a muddy fine sand (d 50 = 235 µm), composed of 24% mud. Sediment organic carbon (OC) and total nitrogen (TN) contents were 0.37% OC and 0.04% TN. Sediment porosity slightly varied with depth, and the top 5 cm had an average porosity of 0.332 ± 0.005.

Bioirrigation
The natural background concentration of bromide in seawater of Arcachon bay was 0.8 mM. About 30 min following NaBr addition, the tracer concentration in the water of the tidal mesocosm was 11 mM. At the end of the experiment, the natural background concentration was reached at 6.0 ± 0.3 cm in the control treatment and at 6.7 ± 0.2 and 6.7 ± 0.3 cm in the experimental units inhabited by nonparasitized and parasitized cockles, respectively, with no significant difference among the three treatments (Kruskal-Wallis test, p = 0.25). The distribution of excess Br − in all experimental units exponentially decreased with depth ( Figure 1A). modeled factor of enhanced biodiffusion (ɛ) (data are represented by points; black crosses correspond to the mean value) following nine days of incubation for the three treatments ("control" (n = 4), "nonparasitized" (n = 12), and "parasitized" (n = 12)).

Sediment Reworking
At the end of the experiment, all the cockles were recovered in the first centimeter of the sediment column. The final distribution of luminophores in experimental units containing cockles consistently exhibited an exponential decrease with depth ( Figure 2A). factor of enhanced biodiffusion (ε) (data are represented by points; black crosses correspond to the mean value) following nine days of incubation for the three treatments ("control" (n = 4), "nonparasitized" (n = 12), and "parasitized" (n = 12)).
In control units, the average value of the enhanced diffusion coefficient (ε) was very close to 1 (1.1 ± 0.1), attesting that solute transport was mainly driven by molecular diffusion ( Figure 1B). In the presence of nonparasitized and parasitized cockles, ε values were slightly higher (1.7 ± 0.3 and 1.5 ± 0.2, respectively) but differences among the three treatments were not significant (ANOVA, p = 0.23), probably due to the high variability among replicates in both "NP" and "P" treatments.

Sediment Reworking
At the end of the experiment, all the cockles were recovered in the first centimeter of the sediment column. The final distribution of luminophores in experimental units containing cockles consistently exhibited an exponential decrease with depth ( Figure 2A). Therefore, 17% ± 2% and 18% ± 1% of the sediment surface (PRS) was reworked in the presence of nonparasitized and parasitized cockles, respectively, and percentages were not significantly different (ANOVA, p = 0.18). In both treatments, tracers were mainly buried between 0.5 and 1.5 cm depth and never beyond 2 cm, thus indicating that particle mixing was restricted to the first 2 cm regardless of the parasitic status. Conversely, the sediment surface of control units was still entirely covered by luminophores after nine days, and 99.2% ± 0.3% of them were found in the uppermost sediment layer. modeled factor of enhanced biodiffusion (ɛ) (data are represented by points; black crosses correspond to the mean value) following nine days of incubation for the three treatments ("control" (n = 4), "nonparasitized" (n = 12), and "parasitized" (n = 12)).

Sediment Reworking
At the end of the experiment, all the cockles were recovered in the first centimeter of the sediment column. The final distribution of luminophores in experimental units containing cockles consistently exhibited an exponential decrease with depth ( Figure 2A). Therefore, 17% ± 2% and 18% ± 1% of the sediment surface (PRS) was reworked in the presence of nonparasitized and parasitized cockles, respectively, and percentages were not significantly different (ANOVA, p = 0.18). In both treatments, tracers were mainly buried between 0.5 and 1.5 cm depth and never beyond 2 cm, thus indicating that particle mixing was restricted to the first 2 cm regardless of the parasitic status. Conversely, the sediment surface of control units was still entirely covered by luminophores after nine days, and 99.2% ± 0.3% of them were found in the uppermost sediment layer.

Principal Component Analysis
The first dimension of the PCA explained 43.5% of the total inertia, and the second dimension explained 27.4% (Figure 4). The NH4 + (percentage contribution to dimension 1: 24%) and dSi fluxes (31%) and were both positively associated with the first dimension, whereas TOU (22%) was negatively correlated. This dimension discriminated the control treatment, characterized by low oxygen consumption and effluxes of NH4 + and dSi, from the experimental units with nonparasitized and parasitized cockles. However, "NP" and dissolved silicate across the sediment-water interface (data are represented by points; black crosses correspond to the mean value) for the three treatments ("control" (n = 4), "nonparasitized" (n = 12), and "parasitized" (n = 12)). Different letters indicate significant differences (p < 0.05) among treatments after performing a pairwise t-test with Bonferroni correction when one-way ANOVA was significant (p < 0.05).

Principal Component Analysis
The first dimension of the PCA explained 43.5% of the total inertia, and the second dimension explained 27.4% (Figure 4). The NH 4 + (percentage contribution to dimension 1: 24%) and dSi fluxes (31%) and were both positively associated with the first dimension, whereas TOU (22%) was negatively correlated. This dimension discriminated the control treatment, characterized by low oxygen consumption and effluxes of NH 4 + and dSi, from the experimental units with nonparasitized and parasitized cockles. However, "NP" and "P" treatments were not strongly differentiated, with the gravity center of "NP" corresponding to slightly higher values of the first dimension than the "P" treatment. Concerning the second dimension, NO x fluxes (26%) and ε (44%) were respectively positively and negatively correlated. There was no net difference among the three treatments regarding the second dimension. Only the gravity center of the "P" treatment corresponded to negative values of this dimension. Lastly, the contribution of Db was too low to characterize one of the two dimensions, even if a negative correlation with TOU could be noted.
corresponding to slightly higher values of the first dimension than the "P" treatment. Concerning the second dimension, NOx fluxes (26%) and ε (44%) were respectively positively and negatively correlated. There was no net difference among the three treatments regarding the second dimension. Only the gravity center of the "P" treatment corresponded to negative values of this dimension. Lastly, the contribution of Db was too low to characterize one of the two dimensions, even if a negative correlation with TOU could be noted. Black arrows represent the projection of measured variables (sediment reworking rates (Db), bioirrigation rates (ε), total oxygen uptake (TOU), and ammonium (NH4 + ), nitrates and nitrites (NOx), and silicate (dSi) fluxes) on the first plane of the PCA. Points correspond to experimental units including the three treatments ("control" treatment (C) in the blue circle, "nonparasitized" cockles (NP) in the yellow triangle, and "parasitized" cockles (P) in the red square). Larger points represent the gravity center of each group corresponding to a treatment, and lines show the confidence ellipse at the 0.95 level.

Discussion
The abundance of metacercarial larvae observed in experimentally infected cockles was 16-fold higher than that in noninfected specimens, indicating that (1) the experimental infection procedure was efficient, allowing high parasite intensity to be attained, and (2) the parasitic status between the two treatments (nonparasitized, NP vs. parasitized, P) was highly contrasted. Moreover, the abundance of H. elongata in parasitized cockles was in the upper part of the range commonly reported in the field [24,67] or recorded in previous experimental studies [39][40][41]. Despite this high infection rate, we found that metacercarial larvae impacted neither the mode of bioturbation generated by cockles nor its intensity. Indeed, the exponential shape of both luminophore and bromide depth profiles revealed that, irrespective of the parasitic status, particle and solute transports were mainly driven by diffusive-like processes [64,68]. Parasitized cockles, thus, behaved as nonparasitized conspecifics, inducing a constant and random biomixing of the first 2 cm of the sediment column characteristic of surficial biodiffusors [18]. Sediment reworking rates measured per gram dry weight were similar in the two treatments (8.09 and 8.21 cm 2 ·year −1 ·g DW −1 in NP and P treatments, respectively) and slightly higher than those previously reported for the same species by Mermillod-Blondin et al. [69] (2.14 cm 2 ·year −1 ·g DW −1 ). They were also on the same order of magnitude as other infaunal suspension-feeding bivalves such as the clam Polititapes aureus [70]. Similarly, the bioirrigation activity of nonparasitized and parasitized cockles during the nine days of the experiment was not different. ε values (1.7 ± 0.3 and 1.5 ± 0.2 in NP and P treatments, respectively) indicate that solute displacements were low, mainly driven by molecular diffusion in most experimental units, regardless of parasite intensity. The weak influence

Discussion
The abundance of metacercarial larvae observed in experimentally infected cockles was 16-fold higher than that in noninfected specimens, indicating that (1) the experimental infection procedure was efficient, allowing high parasite intensity to be attained, and (2) the parasitic status between the two treatments (nonparasitized, NP vs. parasitized, P) was highly contrasted. Moreover, the abundance of H. elongata in parasitized cockles was in the upper part of the range commonly reported in the field [24,67] or recorded in previous experimental studies [39][40][41]. Despite this high infection rate, we found that metacercarial larvae impacted neither the mode of bioturbation generated by cockles nor its intensity. Indeed, the exponential shape of both luminophore and bromide depth profiles revealed that, irrespective of the parasitic status, particle and solute transports were mainly driven by diffusive-like processes [64,68]. Parasitized cockles, thus, behaved as nonparasitized conspecifics, inducing a constant and random biomixing of the first 2 cm of the sediment column characteristic of surficial biodiffusors [18]. Sediment reworking rates measured per gram dry weight were similar in the two treatments (8.09 and 8.21 cm 2 ·year −1 ·g DW −1 in NP and P treatments, respectively) and slightly higher than those previously reported for the same species by Mermillod-Blondin et al. [69] (2.14 cm 2 ·year −1 ·g DW −1 ). They were also on the same order of magnitude as other infaunal suspension-feeding bivalves such as the clam Polititapes aureus [70]. Similarly, the bioirrigation activity of nonparasitized and parasitized cockles during the nine days of the experiment was not different. ε values (1.7 ± 0.3 and 1.5 ± 0.2 in NP and P treatments, respectively) indicate that solute displacements were low, mainly driven by molecular diffusion in most experimental units, regardless of parasite intensity. The weak influence of C. edule on porewater movement can be explained by its shallow burying depth (especially in young specimens), as well as its feeding mode. Indeed, as a suspension-feeder, it directly pumps water with its inhalant siphon just above the sediment surface so that the overlying water transported through the paleal cavity is not in contact with the surrounding sediment [69]. Porewater movements are, thus, mainly generated by shell-valve abduction (i.e., opening of the valves to allow the siphons to extend) and closure of the exhalant siphon, which both induce limited pressurization of the sediment over short periods of time [71]. However, while the influence of cockles on sediment bioirrigation was not genuinely perceptible using bromides at the timescale of the experiment, the high degree of variance of ε values, as well as the enhancement of dSi efflux, observed during benthic flux measurement revealed that these burrowing bivalves, irrespective of their parasitic condition, can nonetheless periodically increase the exchange of solutes between interstitial and overlying waters. As observed for bioturbation processes, parasitic infection with H. elongata did not impact the influence of its second intermediate host on biogeochemical fluxes across the sediment-water interface. Therefore, the sediment total oxygen uptake (TOU), which mainly results from the (1) organic carbon remineralization via aerobic microbial respiration, (2) reoxidation of reduced products from anaerobic degradation, and (3) macrofaunal respiration [72], was very similar in the two treatments (−35.5 ± 1.7 and −35.1 ± 1.6 mmol·m −2 ·day −1 in NP and P treatments, respectively). Estimation of the bivalve oxygen consumption based on the length/weight allometric relationship [38] and the respiration/AFDW equation developed by Newell & Bayne [73] suggested that the TOU increase observed in the presence of young cockles was largely due to their own respiration (96% ± 6% and 99% ± 6% in NP and P treatments, respectively) rather than the stimulation of diagenetic processes through sediment bioturbation. This was confirmed by the very low intensity of NOx effluxes, suggesting that nitrification was not significantly stimulated despite the higher ammonium concentrations in both treatments. This also highlighted that the contribution of cockles to nutrient fluxes across the sediment-water interface in both NP and P treatments mainly resulted from the direct excretion of metabolic wastes (e.g., ammonium) [13,74]. Our findings are consistent with previous studies which showed that shallow-burrowing filter-feeders, including C. edule, have generally limited effects on porewater oxygen concentrations due to moderate bioturbation rates, and they consequently induce low modifications of sediment microbial-mediated processes [13,[75][76][77]. However, although the oxygen consumption by encysted larvae is negligible [78], Magalhães et al. [40] observed that, two days following H. elongata metacercarial infection, cockles can drastically reduce their respiration rate. In the present study, the absence of discernable effects of trematodes on TOU 12 days after infection (and, thus, on oxygen consumption by parasitized cockles since bioturbation rates were similar between the two treatments) reinforces the interpretation of the authors that these immediate effects following encystment reflect a quick defense mechanism rather than a direct long-term impact of parasites on the host oxygen requirement. The absence of significant effects of parasites on sediment bioturbation and biogeochemical dynamics is in contrast with our a priori predictions. Indeed, although the biochemical and physiological effects of trematode parasites are usually much more detrimental on their first than second intermediate hosts [29,31,33,42], recent studies nevertheless highlighted that metacercarial infection can have profound metabolic consequences [39,40,48]. For example, Magalhães et al. [79] observed a strong increase in energetic reserve (e.g., glycogen content) of infected cockles, thus reflecting an enhanced metabolic activity to cope with parasiteinduced additional energetic requirement, as well as fuel defense mechanisms. Moreover, several experimental or field studies demonstrated that metacercarial invasion of different trematode species such as Curtuteria australis or Himasthla spp., both using infaunal bivalves as a second intermediate host, can induce a mechanical obstruction of their feet, thereby potentially altering their burrowing capacity [50,80,81]. This has been interpreted as a behavioral manipulation aimed at increasing the transmission rate to the final host (e.g., a shorebird) that preferentially feeds on surface-dwelling prey. Therefore, we anticipated that infected cockles would remain buried less deep and/or would strongly reduce their lateral displacement. Since the volume of sediment and porewater displaced by burrowing suspension-feeding bivalves fundamentally depends on burying depth and movement distance [15], we expected lower bioturbation rates in parasitized specimens. As an indirect consequence, through a cascading effect, we anticipated a reduction in oxygen penetration depth into the sediment, leading to reduced remineralization rates and inorganic nutrient fluxes toward the overlying water. Such negative indirect effects of parasites on benthic ecosystem functioning through an alteration of host bioturbation activity have been, for example, recently quantified in another marine parasite-host system Upogebia pusilla-Gyge branchialis [7]. However, it is interesting to note that parasite infection can also have the opposite effect. Indeed, Williams et al. [6] showed that infected individuals of Gammarus duebeni with an acanthocephalan parasite, Polymorphus minutus, surprisingly increased their digging activity, thereby reworking an extended sediment surface area compared to noninfected congeners.
Several assumptions can be formulated to explain the absence of significant indirect effects of metacercariae on sediment bioturbation and nutrient fluxes observed during the present study. Firstly, trait-mediated indirect effects would be more obvious and pronounced in older host specimens, which bury deeper into the sediment [82,83]. We chose young bivalves, in which parasite intensity is naturally low, as a model host to ensure contrasted experimental parasitic conditions between our NP and P treatments. However, the bioturbation potential of young cockles is limited due to their low burrowing depth, which does not exceed 1-2 cm compared to 5-6 cm in adults [84]. Therefore, the resolution of vertical tracer profiles every 0.5 cm may not be fine enough to allow for accurate detection of weak alteration both particle and solute transport just below the sediment-water interface. Secondly, deleterious behavioral effects caused by metacercariae would appear over a longer-term period. For example, Bakhmet et al. [85] only observed a significant impact of H. elongata metacercarial infection on the heart rate of the blue mussel Mytilus edulis after 12 months. Thirdly, the degree of pathogenicity and the intensity of host responses are species-specific. Hence, Magalhães et al. [40] showed that the effects of trematode infection on host metabolic activities tightly depend on the parasite species. The effects of H. elongata infection would, thus, be less deleterious or may require a longer time period to arise than for other larval stages or species, such as Bucephalus minimus, which quickly reduces the bioturbation potential of C. edule, as well as their influence on the sediment stability [16]. Lastly, parasites alter the fitness of their second intermediate host via interactions with other biotic and abiotic environmental stressors. Paul-pont et al. [43], for instance, reported that the impact of H. elongata metacercarial infection on C. edule defense-related performance profoundly depended on its combination with other biotic (e.g., Vibrio tapetis) or abiotic (e.g., cadmium) stressors.

Conclusions
The present study suggests that the parasitic alteration of the biochemical and physiological performance of an ecosystem engineer host species does not necessarily translate into marked behavioral changes with obvious consequences for ecosystem functioning. Nevertheless, to better understand the ecological role played by trematode parasites in marine benthic ecosystem processes (e.g., bioturbation) and functions (e.g., nutrient recycling), it would be interesting to further investigate the potential long-term effects of metacercarial infection on older and larger host specimens, which more intensively interact with their physical and chemical environments, by considering the possible interactions with other trematode species or other sources of stress.

Data Availability Statement:
The data presented in this study are available on request from the corresponding authors.