Can a Quick Metabolism Test Predict Which Oyster Families Survive?

Using a simple dye assay to forecast heat-stress hardiness

SORMI-WRAC

2026-07-22

The problem

Oyster growers and breeders want to know which family lines survive — so they can farm the tough ones.

  • Finding out the usual way means breeding and waiting a season to measure performance.
  • Is there a faster, non-lethal shortcut?

Part 1 · The survival trials

How the survival trials work

We test oyster families side by side under a stress and record who lives.

  • 20 oysters per family, 9 USDA families, tested together.
  • Exposed to a stressor — heat (33–36 °C), low salinity, or freshwater.
  • Checked repeatedly over hours to days; we log when each animal dies.

Seven trials run so far, from mild to severe.

Seven trials, mild to severe

Trial Stress Oysters % that died
Apr 27 33 °C 128 49%
May 11 33 °C 84 99%
Jun 04 36 °C 39 95%
Jun 09 36 °C 77 86%
Jul 06 35 °C 378 97%
Jul 06 35 °C + low salt 204 84%
Jul 07 Freshwater → 36 °C 185 33%

Harshness varies a lot — so raw survival can’t be compared trial-to-trial.

What the “survival score” means

A single 0–100 hardiness number per family, built to be fair across trials:

  • Within each trial, rank the families on two things: (1) what fraction survived to the end, and (2) how long they lasted on average.
  • Turn each family’s standing into a percentile vs. the other families in that same trial — this cancels out how harsh the trial was.
  • Average those percentiles across all seven trials.

Higher score = consistently tougher across many kinds of stress.

Survival results: families differ

Family 5 is the toughest; family 7 the most fragile. The spread is real and repeatable across trials.

Part 2 · The metabolism test

How the resazurin assay works

Resazurin is a blue dye that turns pink as living cells burn energy — a non-lethal readout of metabolism.

  • Each oyster sits in a cup of dye; a small sample is read in a plate reader every ~30–60 min for a few hours.
  • We correct for background and adjust for body size.
  • The result is a metabolism curve for each animal over time.

Runs can be done at rest or during heat/freshwater stress.

What the metabolism curve tells us

From each curve we pull ~20 simple descriptions of its shape, for example:

  • How high metabolism climbs, and how fast it rises.
  • Whether it holds steady, spikes, or crashes under stress.
  • When the sharpest drop happens.

Families show distinct, repeatable curve shapes — the raw material for a prediction.

Part 3 · Linking the two

The idea

We have two independent results on the same 9 families:

  • A survival score (from the lethal trials).
  • A metabolism fingerprint (from the non-lethal dye test).

Question: can the metabolism fingerprint predict the survival score — provide data quicker?

How we tested it

  1. Summarize each family’s metabolism curve into simple numbers.
  2. Check which numbers line up with the survival score.
  3. The honest test: hide one family, predict its hardiness from the other eight, then reveal the truth — repeat for every family.

Reading the score: what the numbers mean

Both numbers ask: does the test rank the families in their true survival order? Each runs from −1 to +1.

  • Spearman rho — a ranking agreement score. +1 = same order (test’s toughest = truly toughest), 0 = unrelated, −1 = reversed. We use ranking (not raw values) so one odd family can’t distort it.
  • LOFO-CV rho — the same score, but measured honestly on families the model never saw: hide one family, predict it from the other eight, repeat for all nine, then compare.

Spearman rho = optimistic best case; LOFO-CV rho = the trustworthy one (how it’d do on a new family). A small gap between them means the prediction isn’t just memorizing our nine families.

The surprise: tough families “stay calm”

The clearest pattern was the opposite of what you might guess.

  • Families whose metabolism spikes the most when heated tend to survive the least.
  • The hardy families keep their metabolism restrained under heat — they don’t overreact.

Tough families also crash later

How to read it

  • Each dot is a family.
  • Left → right: how long before the metabolism curve makes its sharpest drop (“time to steepest decline”). Further right = crashes later.
  • Bottom → top: survival score (higher = tougher).
  • The upward trend = later-crashing families survive better.

The chart’s labels

  • time_to_min_slope = time to the steepest metabolic drop.
  • corrected_fc / fw_heat = the metabolism measure, taken during a freshwater-plus-heat run.
  • rho = how well two rankings agree (0 = not at all, 1 = perfect). 0.90 is very strong; the cross-validated 0.81 (the hide-one-family test) confirms it holds for families we didn’t train on.

What is the “resazurin index”?

One combined score that blends the best metabolism clues into a single “survivor-like metabolism” number per family — so their signals reinforce and individual noise averages out.

Built from the four strongest clues (measured under freshwater + heat):

Clue Plain meaning Survivor-like value
Time to steepest drop how long before metabolism crashes later
Steadiness how much metabolism bounces around steadier
Early rise how fast metabolism climbs at first faster
Depth of collapse how far it falls from its peak smaller

Each clue is put on a common scale, flipped so “higher = tougher,” and the four are averaged. Higher index = metabolism that looks like a hardy family’s.

Putting the clues together

How to read it

  • Each dot is a family.
  • Left → right: a combined resazurin index — several top metabolism clues merged into one “survivor-like metabolism” score. Further right = more survivor-like.
  • Bottom → top: survival score (higher = tougher).
  • The tight upward trend = the metabolism index tracks real survival.

The takeaway number

  • LOFO-CV rho = 0.79: even in the honest hide-one-family test, the index predicts the survival ranking strongly (1 = perfect).
  • The toughest family (5) and most fragile (7) land at opposite ends.

What a “survivor” looks like on the test

  • Restrained metabolism under heat — not a big spike.
  • A slower, later decline instead of an early crash.
  • The signal is strongest when the test is run during heat stress, not at rest.

What’s next

  • Add more assays to develop a more robust “index”.
  • Glycogen (heat v control 3 hours)
  • NaK ATPase
  • Citrate Syntase