Beaudreau, Nicholas; Page, Tessa M; Drolet, David; Mckindsey, Chris; Howland, Kimberly; Calosi, Piero (2024): Survival and metabolomics of Mya arenaria and Mya truncata after 12 days in experimental heatwave conditions, files and related R scripts [dataset]. PANGAEA, https://doi.org/10.1594/PANGAEA.964771

Two species of clams, Mya arenaria (1) and Mya truncata (2), were hand collected from Métis-sur-Mer, Québec, Canada (48° 40' 4.6092" N, 68° 1' 5.9484" W) and by SCUBA diving (~ 10 m) at Godbout, Québec, Canada (49° 19' 25.626" N, 67° 35' 17.034" W) respectively. The clams were brought to the wet labs of the Maurice-Lamontagne Institute in Mont-Joli, Québec for acclimatization to experimental conditions. A clams' ability (1) or inability (0) to bury itself into the substrate (sand) was observed visually to control for it's possible relationship to clam performance variability, as well as their common garden tank (pre.acclim) to control for possible tank variability. After > 1 month in this pre-acclimatization phase, during the months of november and december 2020, clams were measured for morophometrics (length, width, and height) with a vernier caliper to control for the relationship between size and clam physiology, and transferred to the experimental system and subjected to a combination of two stressors: one of seven levels of heatwave (2, 7, 12, 17, 22, 27, or 32 °C) crossed with one of two levels of harvesting (with, without). The true temperature was recorded (truetemp) with HOBO 8K pendant data loggers to measure the variation in real temperature experienced by the clams in each tank. Four tank replicates were used for each of the fourteen (7 x 2) experimental treatments to account for possible tank effects. In each of these tanks (1-4), eight individuals of each species were placed together to increase replication. At the end of the experimental period, mortality was assessed by prodding clams in each tank for each species (mya_spp_heatwave_mortality_data) as a measure of response to the stressors. In surviving individuals, three tissues -- mantle (m), gills (g) and posterior adductor muscle (a) were dissected from each individual to asses inter organ differences and the tissues were flash freezed for metabolomics analysis. A targeted metabolomics analysis was run over the months of March to May 2021 at the Iso-BioKem laboratories in Rimouski, Québec, Canada to quantify 48 metabolites with an Agilent 1260 Infinity II high performance liquid chromatographer (mya_spp_heatwave_metabolomics_data) as a measure of response to the stressors.

Mortality data:

This dataset includes information on individual clams and their survival assessed after 12 days in the experimental conditions (alive = 0, dead = 1), the targeted temperature (temp_treatment; 2 - 32 C) and the true temperature (truetemp), the harvesting treatment (with = 1, without = 0), the species (mya arenaria = 1, mya truncata = 2), and the replicate tank (replicate). Based on the presence of fixed (temperature, harvesting, species) and random effects (replicate) in the experiment, a generalized linear binomial mixed effects model was selected to analyze the effects of our treatments. For these analyses, the true temperature (truetemp) was used as a continuous independant variable.

*Some replicate tanks experienced technical problems which led to abnormal variation in temperature during the experiment, and must be removed for further analysis (see Code/Software section)

-----

Metabolomics data:

This dataset includes information on clam tissues (tissue), the individual they were dissected from (sample.y), the experimental tank replicate that individual was held in (replicate), the species of clam (mya arenaria = 1, mya truncata = 2), the pre-acclimitization common garden tank that individual was held in, its burying behaviour in that tank (buried; buried = 1, not-buried = 0), as well as its length (mm), width (mm) and height (mm). The experimental treatments for temperature (temp; 2 - 27 C), the true temperature experienced (truetemp) and harvesting (harv; with = 1, without = 2) levels are also reported. The dataset contains the levels of 48 metabolites (ug / mg) in these clam tissues after 12 days in the experimental conditions.

*Similarly to the mortality data section, some replicate tanks experienced technical problems which led to abnormal variation in temperature during the experiment. These same experimental replicates were not sampled for metabolomics and therefor do not appear in the dataset.

-----

## Code/Software

--

Mortality script (mya_spp_heatwave_mortality_script) :

Some explanation is given for

- line 39 (glmer): The binomial family of data distribution was selected due to the nature of the respones variable (mortality, 0 or 1) with the "logit" link function. A control option was added to optimize estimation of fixed effects.

- line 60 (multiple change points): The syntax for this analysis can be explained by consulting the following ressource: https://lindeloev.github.io/mcp/

- line 68-69 (changepoints) : These lines output the results of the multiple change point analysis, with the average changepoint of the dataset (combined species) which is the "22.33" (mean column) of "cp_1" (name column). From there, the function "ranef" yields species' changepoints, by calculating the deviation from this mean according to each species. This equates to a changepoint of 22.33 + 4.56 (26.89) for Mya arenaria and 22.33 - 4.56 (17.77) for Mya truncata.

----

Metabolomics script (mya_spp_heatwave_metabolomics_script):

All pertinent functions created for these analyses are labelled by their name/purpose in their respective sections.

Some explanation is given for:

- line 61 : some metabolites were not detected in a sufficient number of samples that they were removed from analysis

- lines 506-509 (workflow): this is an example of the preliminary workflow that was used in each subset of tissues.

- lines 512 - 516 (workflow): this is an example of the final workflow that was used in each subset of tissues. It illustrates how the linear mixed model fitting function normalizes the large amounts of variation brought upon by the experimental design.