---
title: "When being a triploid oyster flips the script on ocean acidification"
subtitle: "How extra chromosomes rewire an oyster's chemical response to a changing ocean"
author: "Roberts Lab"
date: today
format:
  html:
    toc: true
    toc-depth: 2
    theme: cosmo
    fig-cap-location: bottom
    embed-resources: true
categories: [epigenetics, oysters, ocean acidification, methylation]
---

## The short version

Oysters are living through an ocean that is slowly getting more acidic as it
soaks up carbon dioxide from the atmosphere. That change is hard on shell-
building animals. One trick the shellfish industry uses is to grow **triploid**
oysters — animals with three copies of every chromosome instead of the usual
two. Triploids often grow faster and stay marketable year-round.

We wanted to know something more basic: when the water turns acidic, do
triploid oysters *respond differently* than ordinary diploids — not just in how
fast they grow, but in how their cells dial genes up and down?

The answer turned out to be surprisingly clean. **Being a triploid doesn't just
turn the volume of the response up or down. It can flip the response backwards.**
A chemical mark that ordinary oysters *add* to their DNA under acidification,
triploid oysters *remove* — and vice versa, at thousands of spots across the
genome.

## What we measured (and what "methylation" means)

We didn't read the oysters' genes themselves. We read a layer of chemical
annotation sitting *on top* of the DNA, called **DNA methylation**.

Think of the genome as a very long instruction manual. Methylation marks are
like sticky notes and highlighter that the cell adds to certain pages to help
decide which instructions to use and when. The marks don't change the letters of
the manual — but they change how it's read. Crucially, the cell can move these
marks around in response to the environment, which makes methylation a good
place to look for how an animal is *coping* with stress.

We studied Pacific oysters (*Crassostrea gigas*) in a simple grid: **diploid or
triploid**, crossed with **normal seawater or acidified seawater** — four
groups, six oysters each, 24 animals in total. For every oyster we had a
genome-wide readout of methylation at roughly ten million individual positions.

## What the data looked like

Before asking any biological question, we checked data quality and looked at
the overall shape of the methylation. Oyster genomes (like most invertebrates)
have a distinctive "mosaic" pattern: most positions carry almost no methylation,
a smaller set is heavily methylated, and there's not much in between. Every one
of our 24 samples showed this same clean pattern, and coverage was even across
the board — so we could compare them fairly.

![**Quality and the methylation landscape.** Each of the 24 oysters was
sequenced deeply and evenly (panels A, C, D). Panel B shows the characteristic
invertebrate "mosaic" pattern — a big pile of near-zero positions and a small
highly-methylated peak.](/Users/sr320/.claude-science/orgs/8f4cdb45-0d59-4652-8ae1-4fd0eb8da9ba/artifacts/proj_652162724d5b/d2c6a1a1-1363-4c36-9dac-0e54180dc693/v1fe6bb93_qc_coverage_methylation.png)

Interestingly, if you just step back and ask "do the four groups look
different overall?", the answer is *no*. The oysters don't sort themselves into
ploidy or pH groups when you look at the whole genome at once. That's an
important clue, not a disappointment: it means the effects of acidification and
ploidy aren't sweeping changes across the entire genome — they're concentrated
at specific, targeted positions. So that's where we looked.

## The headline: ploidy flips the response

First, the big-picture number. When we averaged methylation across the whole
genome for each animal and asked how it responded to acidification, we found the
two ploidies going in **opposite directions**. Acidification nudged methylation
*up* in diploids and *down* in triploids. Statistically, the "it depends on
ploidy" effect (the interaction) was significant, while neither factor on its
own was — which is exactly the fingerprint of a response that flips.

![**The crossover.** Under acidification (low pH), average methylation rises in
diploids (2N) but falls in triploids (3N) — the two lines cross. This is the
whole story in one picture.](/Users/sr320/.claude-science/orgs/8f4cdb45-0d59-4652-8ae1-4fd0eb8da9ba/artifacts/proj_652162724d5b/4cf4fccb-0286-4413-a500-22b3d2eb1b1d/vdcc8c2cd_global_methylation.png)

Then we zoomed in to individual positions. Diploids responded to acidification
mostly by *adding* marks; triploids responded mostly by *removing* them — and
triploids reacted at more positions overall.

![**Same stress, opposite direction.** Diploid oysters mostly gain methylation
under acidification; triploid oysters mostly lose it.](/Users/sr320/.claude-science/orgs/8f4cdb45-0d59-4652-8ae1-4fd0eb8da9ba/artifacts/proj_652162724d5b/e469ba0e-2294-4fd8-8c67-28fde611e655/v080d1455_dml_contrasts.png)

## Nailing it down: 17,000 positions that flip

A skeptic could ask: maybe diploids and triploids are just reacting at *different*
positions, not truly reversing at the *same* ones. So we ran a direct test,
position by position, asking specifically whether the acidification response
*differed between the ploidies* at each spot.

About **17,000 positions** passed this strict test — and when we sorted them by
*how* they differed, **96% were genuine reversals**: the mark goes up in one
ploidy and down in the other at the very same position. Only a small fraction
were simple "louder in one group" differences.

![**Reversals, not just louder or quieter.** Each point is a genome position;
its horizontal spot is the diploid response and its vertical spot is the
triploid response. The points pile into the corners where the two responses have
*opposite* signs — the signature of a flipped response.](/Users/sr320/.claude-science/orgs/8f4cdb45-0d59-4652-8ae1-4fd0eb8da9ba/artifacts/proj_652162724d5b/18c572da-b6df-4047-bd1f-b00007b6762d/v1d13d785_interaction_analysis.png)

We saw the same flip when we grouped the genome into larger regions instead of
single positions — so it isn't an artifact of looking too finely.

## Which genes are involved?

The flipped positions weren't scattered randomly. They landed
**inside genes** far more often than expected — the part of the genome that
actually codes for proteins and their instructions — and were largely absent
from the empty "desert" regions between genes. In invertebrates, methylation
inside gene bodies is the functional kind, so this is where we'd hope to see a
meaningful signal.

When we asked what those genes *do*, the flipped set was enriched for genes
involved in the cell's internal scaffolding and transport machinery (the
cytoskeleton, molecular motors), cell signalling, and the control of other genes
— the kind of general-purpose "how the cell runs itself" machinery you'd expect
an animal to retune when it's coping with a different internal chemistry.

![**Where the marks land and what they touch.** (A) Flipped positions sit inside
genes, not in the gaps between them. (B) The genes involved are enriched for
cytoskeleton, motor, signalling and gene-regulation functions.](/Users/sr320/.claude-science/orgs/8f4cdb45-0d59-4652-8ae1-4fd0eb8da9ba/artifacts/proj_652162724d5b/d3a9c58c-5640-4bab-89c3-acd902183445/v429db840_annotation_enrichment.png)

## Why it matters

For anyone growing or managing oysters, the practical takeaway is that **a
triploid oyster is not simply a diploid with a dialed-up or dialed-down stress
response.** At the molecular level it is running a *different program* when the
water turns acidic. That could help explain why triploids and diploids sometimes
differ in how they tolerate environmental stress — and it's a caution against
assuming results from one ploidy transfer directly to the other.

For the science of how animals cope with a changing ocean, it's a reminder that
the genome's "software layer" — the methylation marks — can be rewired by
something as fundamental as an extra set of chromosomes.

## A note on how we did it

This was a reanalysis of already-sequenced oysters, starting from the processed
methylation files. We used the R package **methylKit** for most of the work, but
the central question — does the response *flip* with ploidy? — needed a custom
statistical test, because standard tools compare two groups at a time and can't
directly ask an "it depends" question. All the analysis parameters, code, and
result tables are archived alongside this post so the work can be reproduced.

*This post describes an exploratory reanalysis. The sample information was
reconstructed from the analysis code that accompanied the data, and the findings
would be strengthened by validation in an independent set of animals.*