Lec 13 Teaching Prep: Part 1 — Pavo vocab + Miyazawa labyrinths

2026-05-11 18:00:35 • 20:52

-

Lecture 13 Teaching Prep.

0:02

Part 1.

0:03

The Vocabricap.

0:04

The Conceptual Map.

0:06

And the Miyazawa Labyrinths.

0:07

Hey Michael.

0:09

Welcome to your prep for Lecture 13.

0:11

This is Tuesday, May 12th, Week 7, the second half of the Color Patterns Unit.

0:17

There are three things to hold in your head as you walk in.

0:19

First, this is a bridge lecture, not a fresh start.

0:23

Thursday you ended with the Walpert Versus Touring Dicotomy and the 4-cell Pallet.

0:28

Today you pick up exactly where you left off and push two ways at once.

0:32

You push deeper into the math with the Miyazawa Labyrinths and Murray Scaling Rule.

0:38

And you push wider into the genetics with the Master Regulator Toolkit and the ALX3 Stripe

0:44

Story.

0:45

The synthesis at the end is what makes the lecture work, three knobs of pattern evolution,

0:50

cell palette and patterning rule and master regulators.

0:54

The lecture earns that three knob pay off if you defend the pace through the middle and

0:58

don't lose ten minutes on touring math.

1:00

Second, Wednesday is demo seven.

1:03

The students show up to lab the next day and run pavo on real fish images.

1:08

The opening twelve minutes of today's lecture are explicit demo seven prep.

1:13

You are reactivating the metrics they need, K and JC and JT and M and A and M underscore

1:19

DS and M underscore DL.

1:21

Don't skip this.

1:23

If a student walks into the demo Wednesday without those abbreviations in their head, they

1:27

will spend forty minutes confused.

1:30

The pavo vocab block is a service you are providing to your own Wednesday lab.

1:34

Third, the deck is long.

1:36

Thirty-six slides.

1:38

You will not deliver every slide at full pace.

1:41

The two natural cut targets if you are running long are the Murray geometry slide, the one

1:46

about curved surfaces and effective diffusion, and the second mammalus slide on the guideless

1:51

gene.

1:52

Both are good but neither is the spine.

1:55

Cut them silently.

1:56

Don't apologize for skipping.

1:58

Part one of your prep covers the opening through the pufferfish pattern blending story.

2:03

About thirty-five minutes of class time.

2:06

Part two covers reaction diffusion math through the exit ticket.

2:09

Let's start.

2:10

Your title slide is up.

2:12

Lecture 13.

2:14

Color patterns two.

2:15

The subtitle, math evolution and the rules that scale is the spine of the lecture.

2:21

Hold the slide for three seconds.

2:23

Then advance.

2:24

Slide one is announcements.

2:26

Demo seven prep URL.

2:28

Office hours, Tuesday, two to four.

2:31

This is thirty seconds.

2:33

Point at the URL.

2:34

Say the line.

2:35

If you show up Wednesday without a working you will spend the first twenty minutes installing

2:40

instead of doing science.

2:42

Do it tonight.

2:43

Then advance.

2:44

Slide two is the color vocabulary slide.

2:47

Hue saturation luminance.

2:50

The big hue will SVG is the visual anchor on the right.

2:53

The three definitions are on the left.

2:56

This is ninety seconds of clean vocabulary.

2:59

Walk it briskly.

3:00

Hue is what color the wavelength identity.

3:04

Saturation is how pure.

3:05

Vivid magenta high.

3:07

Washed out pink low.

3:08

Gray zero.

3:09

Luminance is how bright.

3:11

Regardless of hue.

3:12

Then the key pavo move at the bottom.

3:15

Chromatic boundaries change hue or saturation.

3:18

Acromatic boundaries change brightness only.

3:21

A black and white striped fish has high luminance contrast, but zero chromatic contrast.

3:27

A rainbow fish can have huge chromatic contrast at moderate brightness contrast.

3:32

Plant this distinction firmly.

3:34

It is the conceptual basis for M underscore DS versus M underscore DL three slides from

3:39

now.

3:40

Slide three is the first TPS.

3:43

Pavo pattern metrics review.

3:45

Terminal phase.

3:46

Rass on the left.

3:47

Three questions on the right.

3:49

Ninety seconds discussion.

3:50

Sixty seconds cold call.

3:52

The three questions are how many color classes define JC and JT in your own words.

3:58

And would you expect JC and JT to be high or low for this fish?

4:02

Walk among them.

4:04

Listen.

4:05

The most common error is over splitting.

4:08

Students see six or seven colors because they treat shading variations as distinct color

4:12

classes.

4:14

Push them to group similar hues.

4:16

The right answer is three to four colors.

4:19

Magenta body, cyan chin and tail, yellow pectoral patch, maybe a dark eye.

4:24

JC is moderate because three bold colors cover large areas.

4:29

Not low because there are multiple colors.

4:31

Not high because three is not a lot.

4:34

JT is low to moderate because with three colors in big patches there are only a few boundary

4:39

types.

4:40

Cold call two pairs.

4:42

Validate the moderate answers visibly.

4:44

Then advance.

4:46

!

4:47

Slide four is the answer reveal for K, JC and JT.

4:50

Don't dwell.

4:51

Sixty seconds.

4:52

The key intuition to land is that JT depends on both how many colors and how they are arranged.

4:58

Few colors in big patches means few transition types.

5:02

A mosaic with the same colors would have higher JT.

5:06

Then advance.

5:07

Slide five is the second TPS.

5:09

The harder one.

5:11

Compare two fish.

5:12

The colorful terminal phase ross on the left.

5:15

The juvenile thalasoma with horizontal stripes on the right.

5:19

Ninety seconds discussion.

5:21

Two questions.

5:22

Define m and a.

5:24

For fish b, predict whether a is greater than, less than or roughly equal to one.

5:30

Then define m underscore ds and m underscore dl.

5:34

Which fish has higher m underscore ds.

5:36

Which has higher m underscore dl.

5:39

This is where students need the most help.

5:41

M is transition density, total boundary complexity.

5:45

The striped thalasoma has more boundaries per unit area because the stripes are fine.

5:50

So fish b is higher m.

5:52

A is aspect ratio, vertical over horizontal.

5:56

The thalasoma has strong horizontal stripes, so vertical transitions outnumber horizontal

6:01

ones, which means A is less than one.

6:04

The ross has large patches with no strong directional bias, so A is roughly equal to one.

6:10

The harder pair is m underscore ds versus m underscore dl.

6:14

M underscore ds is mean chromatic boundary strength.

6:18

How much hue changes at a boundary?

6:20

Fish A, the rainbow ross, has magenta to yellow to cyan transitions, dramatic hue shifts.

6:27

High m underscore ds.

6:29

Fish b is stripes, but the stripes are mostly dark to light transitions with less hue change.

6:34

Lower m underscore ds.

6:36

M underscore dl is mean luminance boundary strength.

6:40

How much brightness changes?

6:42

Fish b is sharp dark stripe light interest stripe boundaries.

6:45

High m underscore dl.

6:47

Fish A also has brightness contrasts, but they're more gradual.

6:52

Moderate m underscore dl.

6:53

Cool cool.

6:54

Then advance.

6:56

Slide 6 is the answer reveal for m.

6:58

A m underscore ds.

7:00

M underscore dl.

7:02

About 2 minutes.

7:03

The right column has the abbreviations nomenic table.

7:06

M is mean.

7:07

A is aspect.

7:09

D is delta.

7:10

S is saturation, the chromatic term.

7:13

L is luminance, the brightness term.

7:16

So m underscore ds is mean delta saturation at boundaries, chromatic strength.

7:22

M underscore dl is mean delta luminance at boundaries, brightness strength.

7:27

Walk that table once.

7:29

Once a student sees that D means delta and S means saturation and L means luminance.

7:35

The rest of the abbreviations stop being mysterious.

7:38

Then deliver the bridge line.

7:40

These are the metrics you will compute on Wednesday with Pavo.

7:43

Now let's get back to the Turing answer from Thursday.

7:46

Advance.

7:47

You are about 12 minutes in.

7:49

The vocab recap is done.

7:51

The lecture pivots now.

7:53

Slide 7 is the four ways to make a pattern.

7:56

The conceptual map for everything that follows.

7:59

For boxed mechanism cards.

8:01

Turing self organization.

8:03

Global information.

8:05

Hybrid blending.

8:07

Regulatory switches.

8:08

Walk each box.

8:10

Turing is local activation plus long range inhibition.

8:14

Examples are zebrafish stripes and pufferfish labyrinths.

8:17

Positional information is cells reading where they are on a tissue.

8:21

Examples are mimulus nectar guides and butterfly wing modules.

8:25

Hybrid blending is two parameter sets meeting in the same skin.

8:29

Examples are salmoned hybrids and a wroth-ron pufferfish.

8:33

Turing switches are small changes in when or where master regulators are expressed.

8:39

Examples are W and T A. Optics.

8:42

ALX3.

8:43

Connects in 41.8.

8:45

The teaching move here is not that these boxes are mutually exclusive.

8:49

The teaching move is that real pattern biology often combines them.

8:54

Butterfly wing modules are positional information at the developmental level, but their evolution

8:59

is dominated by regulatory switches.

9:02

Walk the room rhetorically.

9:03

Where would you put the rest we just scored?

9:06

Don't take answers.

9:07

Just plant the question.

9:09

Total time on this slide is about four minutes.

9:12

Slide 8 is the Thursday TPS callback.

9:15

Two cyclic photos.

9:17

The striped melanocromisoratus.

9:19

The Bard pseudotrophius demisoni.

9:21

The cross prediction.

9:23

Most of you predicted intermediate stripes.

9:25

Some stripes some bars of blend.

9:28

Thirty seconds.

9:29

Flash the images and ask who remembers what they predicted.

9:33

Don't answer yet.

9:34

Advance.

9:36

Slide 9 is the answer.

9:37

Labyrinths.

9:38

This is the dramatic reveal slide.

9:41

The big white photo card at the top has the Miyazawa Ellen model figure.

9:45

The spots to labyrinth to inverse spot sequence on a sphere.

9:49

The fragments below add the second figure with three different Turing models, all showing

9:53

the same path.

9:55

Start with only the first figure visible.

9:58

Say something close to this.

10:00

Imagine the patterning rule running on the surface of a sphere.

10:03

We are not drawing spots by hand.

10:05

We are changing one tuning parameter in the activator inhibitor system.

10:10

Walk left to right.

10:11

Initially local activator peaks stabilize the spots.

10:15

As the activation inhibition balance shifts, the spots connect into a labyrinth.

10:20

Farther along the sweep, the polarity flips and the background becomes the spot.

10:25

And advance the fragment to show all three rows.

10:29

The key teaching beat is that this is not a quirk of one model.

10:33

Three different Turing models, the linear Ellen model, the Giror-Mine heart model, the

10:37

Grayscott model, all show the same spots to labyrinth to inverse spot sequence.

10:43

The labyrinth zone is a generic property of reaction diffusion systems.

10:48

Then the biological logic.

10:50

Turing parameters are quantitative traits.

10:53

If they are determined by multiple loci, hybrids inherit intermediate values.

10:59

Intermediate values land in the labyrinth zone.

11:01

The hybrid pattern is predicted to be labyrinthin, not a blend of spots and stripes.

11:07

Something genuinely new.

11:09

Two minutes on this slide.

11:10

Then the bridge.

11:11

That's the prediction from math.

11:13

Can we test it with real fish?

11:16

Advance.

11:17

Slide 10 is Miyazawa 2010 the setup.

11:20

Salmoned hybrids.

11:22

Mark four panel figure on the left.

11:24

Story pros on the right.

11:26

Tell it as a story.

11:27

Two salmoned species live in the same Japanese rivers.

11:31

One has light spots on a dark background.

11:33

The white spotted char.

11:35

The other has dark spots on a light background.

11:37

The misuse salmon.

11:39

In Turing terms, these are on opposite sides of parameter space.

11:43

They occasionally hybridize naturally, and Miyazawa also made artificial crosses.

11:48

The hybrid, both natural and artificial, shows a labyrinthin pattern that neither parent

11:54

has.

11:55

Sinois worm-like markings.

11:57

Exactly what the simulation predicted.

11:59

Two minutes.

12:01

Slide 11 is the four by four hybrid grid.

12:04

Miyazawa 2010 figure four.

12:07

The full proof.

12:08

For salmoned species.

12:10

Within group crosses, light by light or dark by dark, produce parent-like patterns.

12:16

Across group crosses, light by dark, produce labyrinths every time.

12:21

Eight intergeneric combinations.

12:23

All labyrinthin.

12:25

Panel E is the scatter plot.

12:27

Hybrids sit at intermediate color tone, but higher pattern complexity than either parent.

12:33

Transgressive.

12:34

More complex than both parents.

12:36

This is the signature of nonlinear blending.

12:39

The punchline.

12:41

This is not one lucky cross.

12:43

It is a systematic repeatable result.

12:45

Two minutes.

12:46

Then the bridge.

12:48

So the 2010 paper proved it in salmoneds.

12:51

Ten years later Miyazawa asked, does pattern blending explain color diversity across all

12:56

fishes?

12:58

Advance.

12:59

Slide 12 is the bigger question slide.

13:01

Miyazawa 2020 figure one.

13:04

Bridge slide.

13:05

Ninety seconds.

13:07

Don't go deep here.

13:08

Just establish two facts.

13:10

One.

13:11

Color patterns are more similar across, distantly related species than among close ones.

13:15

Most relatives.

13:17

Massive homoplacee.

13:18

Two.

13:19

Miyazawa tested pattern blending across 18,114 fish species from 559 families.

13:27

The strongest association is between labyrinthin patterns and spot patterns.

13:33

Exactly what blending predicts.

13:35

Counterintuitive because labyrinths look like stripes, but the data say labyrinth co-occurs

13:40

with spots.

13:41

Then the teaser.

13:42

Can we catch it in the act?

13:44

Pufferfish coming up after the cell biology block.

13:47

Advance.

13:49

Slide 13 is the student question slide.

13:52

Why can small fish pattern but small mammals can't?

13:55

Two photos on the left, the house mouse uniform and the zebrafish bold striped at similar

14:00

body size.

14:01

This was an actual student question from Thursday.

14:04

Don't blow past it.

14:06

The setup.

14:07

Both use touring dynamics, but different implementations.

14:10

Two differences that matter.

14:12

When and how?

14:13

Again, mammals pattern the embryo when the skin domain is tiny.

14:18

Fish pattern post-embranically on growing skin.

14:22

So a mouse embryo can't fit even one wavelength and ends up uniform.

14:27

A zebrafish larva is already several millimeters and growing, so even a small adult fish has

14:32

many wavelengths of pattern.

14:34

How?

14:35

Mammals use molecular activator inhibitor pairs, WNT and DKK for hair follicle spacing,

14:42

and interpath way for skin appendages.

14:45

Molecular diffusion gives a long wavelength.

14:47

Fish use cell cell interactions, melanophoresanthophore repulsion plus gap junction support.

14:53

Cell contact and gap junction reach give a shorter wavelength.

14:57

Shorter wavelength means finer pattern.

14:59

The important caveat.

15:01

Don't say mammals can't do touring.

15:03

They can.

15:04

The Murray slide later will show that they do.

15:07

The question is why size constrains them more severely than fish.

15:11

Two minutes on this slide.

15:13

Then advance.

15:14

Slide 14 is the zebrafish three cell system slide.

15:18

Gertwein 2020 figure 1A on the left.

15:21

The Hebrew University a Doddy-Lad paper.

15:23

The big idea is that a stripe is not made by one pigment cell.

15:27

Three cell types.

15:29

Melanophores, black, light absorbing.

15:32

Santafores, yellow orange.

15:34

Iridophores, reflective with guanine crystal stacks.

15:38

The dark stripe and the light in terstripe are local neighborhoods.

15:42

Each neighborhood gives cells different signals.

15:45

And with the question.

15:46

Are stripe and terstripe Iridophores the same cells in different places or different cell

15:51

states?

15:52

That is the question Gertwein answered.

15:55

Two minutes.

15:56

Slide 15 is the old model slide.

15:59

Two panel figures showing model 1 respsification versus model 2 differentiation in Cetero.

16:05

The old model said Iridophores start in the interstripe, then some cells move into the

16:10

stripe and transform.

16:12

More phagenetic respsification.

16:14

The new model says precursors settle in each neighborhood and differentiate there.

16:19

Differentiation in Cetero.

16:21

The analogy I want you to use is not people moving and changing jobs.

16:25

The analogy is new arrivals in each neighborhood training for different jobs because the local

16:30

environment tells them what to become.

16:33

After it all did time lapse imaging and fate mapping.

16:36

They watched thousands of cells.

16:39

Dense and terstripe Iridophores did not migrate out and become loose stripe Iridophores.

16:44

So model 2 wins.

16:46

Three minutes.

16:47

Slide 16 is the new model slide.

16:49

Two Iridophore crystallotypes.

16:52

Crystalotype is the word to teach.

16:54

A cell type distinguished by the architecture of its guanine crystals.

16:59

Ritophore is our loose stellate cells with more ordered crystal arrays and blue shifted

17:03

reflection.

17:05

Interstripe Iridophores are dense cuboidal cells with different crystal organization and

17:10

silvery yellowish appearance.

17:12

Both use guanine.

17:13

The crystal architecture differs.

17:15

Why it matters for patterning.

17:17

The reflective layer is itself patterned.

17:20

Fish are not just arranging black and yellow pigment cells.

17:23

They are also locally building different structural color cells.

17:27

This is the clean bridge back to Tuesday's lecture.

17:30

Fish can make fine adult patterns because their chromatophore palette is richer than mammalian

17:34

melanocytes.

17:35

Three minutes.

17:37

Slide 17 is the pattern blending hypothesis slide.

17:41

Miizawa 2020 figure one again.

17:44

This time with the framework on the right.

17:46

The teaching move.

17:47

Cross two simple patterns and you get a labyrinth.

17:50

The strongest association in the 18,000 species data set is labyrinth and plus spot patterns.

17:57

This is the pufferfish on the next slide.

17:59

Two minutes.

18:01

Slide 18 is the pufferfish slide.

18:04

Pattern blending caught in the act.

18:05

Miizawa 2020 figure three.

18:08

The arothron pufferfish vertical panel on the left.

18:11

This is the centerpiece of the Miizawa vignette.

18:15

Walk through carefully.

18:16

The spotted parents.

18:17

A. Firmamentum and A. Stelaedis on opposite sides of parameter space.

18:23

The labyrinthin species.

18:25

A. Multilineatus.

18:27

A. Cardiose.

18:28

A. Mappa.

18:30

Then the genomic proof.

18:32

Whole genome sequencing of 12 arothron species.

18:35

The labyrinthin species are hybrids of spotted parents.

18:39

PCA on 28 million SNPs put them between the parents.

18:43

Admixture analysis estimated 50-50 ancestry.

18:47

Interclass hetero-segosity was 97 to 99%.

18:52

They are early generation hybrids.

18:54

The punchline.

18:55

A. Multilineatus and A.

18:58

Cardiose were originally described as distinct species based on their peculiar skin patterns.

19:04

They are not distinct species.

19:06

They are hybrids.

19:07

Pattern blending created camouflage phenotypes that fooled taxonomists for over 100 years.

19:14

78% of possible trios among spotted and labyrinthin arothron species

19:19

showed significant patterns in D, indicating widespread hybridization.

19:24

3 minutes.

19:25

Slide 19 is the 18,000 species quantification.

19:29

Figure 4A.

19:31

Simulations match real pufferfish.

19:33

Blue dots are R.D. predictions.

19:36

Colored symbols are real arothron species.

19:39

Spotted at the extremes.

19:41

Labyrinthin in the middle.

19:42

Intermediate tone, but higher complexity.

19:45

Transgressive.

19:46

Across all fishes, Bayesian phylogenetic analysis found correlated evolution of labyrinthin

19:52

and spot motifs in 8 of 10 major fish orders.

19:57

Over 900 fish species with labyrinthin patterns across over 160 families.

20:03

Two minutes.

20:04

Slide 20 is the pattern blending framework synthesis.

20:08

Two coupled fields same skin.

20:10

Different wavelengths.

20:11

Different cell systems.

20:13

Additive in the final phenotype.

20:15

The big idea.

20:16

Pattern diversity is cheap to generate.

20:19

Reformation makes new patterns with no new genes needed.

20:23

This makes plain explosive cycloid radiations.

20:26

Same logic across fish butterflies and snakes.

20:30

Land the punchline.

20:31

The cell does the math.

20:33

Selection picks the winners.

20:35

End of part 1.

20:36

You are about 35 minutes into the lecture.

20:39

Have done.

20:40

Take a breath.

20:41

Part 2 starts with the reaction diffusion math and Murray's scaling rule.

20:45

In the master regulator toolkit, the ALX 3 stripe story, plants and the 3 knob synthesis.