What a Doodle Coat Panel Can and Cannot Tell You

Poodle crosses are bred, priced, and sold largely on coat. A "fleece" puppy is advertised differently from a "wool" one, and a puppy that turns out to shed is a problem for everybody involved. So it is reasonable that coat panels have become close to standard, and reasonable that breeders make real mating decisions from them.

What follows is not an argument against testing. Testing at this locus is the single best thing a cross breeder can do. It is an account of which parts of that panel are load-bearing and which parts are not, because at the moment they are sold to you as though they were the same thing, and they are not.

Poodle crosses inherit their coat genetics from the poodle side as much as anything else, which is why this sits on a poodle genetics site. The alleles are the ones we already document. What changes in a cross is how much you can conclude from them.

---

The short version

What you are testing	How much weight it holds
Furnishings / improper coat	Solid. Predict it, breed on it.
Shedding	A shift in the odds. Real, but not a result.
Curl category	Not supported in crosses.
Wool / fleece / hair	No genetic definition exists.
"Weak furnishings"	No published evidence it exists.

The rest of this article is why.

---

First, check you are not reading your report backwards

Before any of the genetics, there is a practical problem that costs breeders real money: the labs do not agree on notation, and one lab uses opposite meanings for the same letters on two of its own pages.

From UC Davis's own furnishings page:

Dogs with N/N genotype (IC/IC in the Improper Coat test) are expected to lack furnishings.

Dogs with F/F genotype (N/N in the Improper Coat test) are expected to display furnishings.

Read that twice. N/N means unfurnished if you ordered the furnishings test and fully furnished if you ordered the improper coat test. It is the same locus and the same laboratory. They note it themselves, because the two products are the same assay reported from opposite directions.

It gets worse across labs. Here is what the same biology looks like on different reports:

Locus	Embark	UC Davis	Animal Genetics
Furnishings (RSPO2)	II / FI / FF	N/N N/F F/F — or IC/IC N/IC N/N on the improper coat test	n/n / n/F / F/F
Curl (KRT71)	CC CT TT — C means straight	N/N N/C1 C1/C1 — C means curl	n/n n/C C/C — C means curl
Shedding (MC5R)	TT = lighter shedding	SD = low shedding	SD/SD SD/N N/N
Coat length (FGF5)	ShSh / ShLh / LhLh	S/S S/L L/L — L means long	L/L = short, l/l = long

Three things to take from that table. C means straight at one lab and curly at another. L means long at one and short at another. And N/N at RSPO2 is genuinely ambiguous without knowing which test was ordered.

So the first rule of reading a coat panel is that a genotype string means nothing on its own. You need the lab and the test name before the letters mean anything. If you have ever compared two dogs' results from two different labs and concluded something about a mating, it is worth going back and checking which notation each report was using.

---

The one result that is solid: furnishings and improper coat

Furnishings are the moustache, beard, and eyebrows — the facial hair that makes a doodle look like a doodle rather than like a retriever. They are caused by an insertion in the RSPO2 gene, and furnishing is dominant: one copy is enough.

The evidence here is about as clean as canine genetics gets. In the original study, 297 of 298 furnished dogs carried the insertion, and none of the 406 unfurnished dogs did.

The flip side is improper coat: a dog that inherited no copy of the insertion in a breed where furnishings are expected. It is a flat, retriever-like face and legs instead of the furnished look, and it is recessive. Two furnished parents that each carry one copy can produce improper coat puppies, and roughly a quarter of that litter will be affected.

This is the result worth testing for and worth breeding on. It is also the one with money attached: an improper coat puppy is a fault in every registry that recognises these dogs, and it does not sell at the price the litter was planned around. Knowing both parents' furnishing genotype tells you in advance whether that outcome is possible.

If you test one thing, test this.

---

The result that is a shift in the odds: shedding

Shedding is associated with MC5R, and that association is genuine peer-reviewed work rather than lab marketing. It is also more limited than it sounds, in a way worth understanding before you promise anything to a buyer.

In the study that established it, shedding was never measured on individual dogs. It was gathered at breed level, from published breed descriptions, on a three-point scale. So the finding is that breeds reputed to shed less tend to carry this variant. That is a real signal. It is not a measurement of your dog.

There is also an unresolved disagreement about how shedding actually works. Furnishings themselves are strongly associated with low shedding, and the labs do not agree on how the two genes combine — one describes them as additive, another says a furnished dog tends to shed lightly more or less regardless of its MC5R result. No study has separated out how much each contributes.

And there is a whole component nobody tests at all. Whether a dog has a single or double coat is influenced by a region near the ADRB1 gene, identified in 2019. Poodles are single-coated; Labradors and Golden Retrievers are double-coated. That difference drives a great deal of real-world shedding and matting, and there is no commercial test for it. Every coat panel on the market has that hole in it.

So: report shedding as a direction, never as a promise. "Both parents carry the low-shedding variant and are furnished, so puppies are likely to be lighter shedders" is defensible. "This litter will not shed" is not.

---

The result that does not hold up in crosses: curl

This is the one that will be least welcome, so here is the evidence rather than an opinion.

Curl is associated with KRT71, and in purebred populations the association is strong. The problem is what happens in crosses specifically.

The only genome-wide study of Australian Labradoodles was published in 2020 by the Ostrander laboratory at the NIH. When they looked at KRT71 in that breed, they found the genotypes evenly distributed — 5 dogs with no copies, 8 with one, 8 with two — where the poodles in the same study were 0, 6, and 24. In the authors' own words:

This was surprising given our perception of how many ALBD have curly or wavy coats … Our data do not exclude the possibility of other genes with variable or weak alleles playing a role in ALBD coat curl.

In other words: in the exact population doodle breeders care about, the curl gene did not sort the way the curl story requires.

There is a second problem. The widely repeated rule is that one copy gives a wavy coat and two copies give a curly coat. That rule is not traceable to the original research. The 2009 paper that identified the gene never describes it as incompletely dominant, and its own supplementary table groups one-copy dogs together with two-copy dogs as curly. The wavy-heterozygote model appears to enter the literature through a later review table carrying no citation, and from there into commercial test descriptions. The labs themselves now disagree: one states incomplete dominance with heterozygotes wavy, another describes curl as simply dominant.

There is also more than one curl allele. A second variant was reported independently by two groups in 2019, and a single-marker curl test cannot see it. One curly dog in that work carried neither known variant at all.

Practical consequence: a curl result is worth recording, and it is not worth promising a coat texture on. If you have been pricing puppies by predicted curl, that is the part of the panel with the least underneath it.

---

What has no genetic definition at all

Wool, fleece, and hair

These are the categories the entire market runs on, and no published study uses, tests, or validates them.

The registries do not agree with each other either. Two of the major Australian Labradoodle organisations recognise two coat types and call the third a fault; two others recognise three. They disagree on how long a fleece coat should be and on how dense a wool coat should be. In at least one published standard, the fleece description and the wool description overlap enough that a single dog could satisfy both.

The most telling detail: when the Ostrander laboratory wrote the one genome paper on this breed, they declined to use the breed's own vocabulary, describing coats as single or double and as curly, wavy, or straight, and noting that the standard permits curl anywhere on a continuum from straight to tightly curled. A continuum is precisely not three named categories.

You can use these words as descriptive shorthand for what a coat looks like today. They are not genetic categories, and no test predicts them.

"Weak furnishings"

Several laboratories sell a furnishings test with a third allele, variously called weak furnishings, Fw, or F2, described as producing a lightly furnished coat.

There is no published evidence that this allele exists. The reference catalogue of canine genetic variants lists exactly one RSPO2 coat variant. A 2022 survey of 118 breeds plus wolves found only the original insertion. No laboratory publishes a position, a coordinate, or a variant identifier for a second furnishings allele, and at least one lab selling the three-allele version lists only the 2009 paper as its reference — a paper that contains no such allele.

Lightly furnished dogs certainly exist. Furnishings vary in expression, and breeders describing weak furnishings are describing something real that they can see. What has not been shown is that a second allele explains it.

Predicting adult coat from a puppy

There is no published study on when or how coat texture changes as a dog matures. Breeders and groomers broadly agree that something changes between roughly six and fourteen months. That is worth knowing, and it is not evidence.

Notably, breeders also report that lightly furnished puppies can look unfurnished until nine to twelve months, which would confound any study that phenotyped puppies rather than adults.

F1b as a guarantee

No study measures coat variance by generation. The 2020 Labradoodle work found multigenerational dogs still genetically heterogeneous, to the point that the authors could not cleanly assign generation labels from the pedigrees they were given and had to group later dogs together.

A generation label tells you the expected dose of poodle alleles. It is not a measurement of how the coat came out.

---

The claim to stop making

"Hypoallergenic" is not merely unproven for these dogs. The evidence runs the other way.

A 2012 study measured Can f 1, the major dog allergen, in hair and coat samples and in homes. Breeds marketed as hypoallergenic — poodles and labradoodles among them — carried higher levels on their hair than control breeds, with poodles highest of all. The authors concluded there is no evidence for classifying any dog breed as hypoallergenic. A separate study of 173 homes found no difference in household allergen levels under four different classification schemes.

The mechanism explains why. Can f 1 comes from saliva, skin, and dander, not from hair. A coat that does not shed does not remove allergen from the dog; it retains it. What a furnished, low-shedding coat may plausibly do is reduce how much allergen is dispersed around a house. That is a reasonable thing to say. It is not the same claim, and buyers with allergies deserve the difference.

---

And one thing no laboratory will tell you

No commercial laboratory publishes an accuracy figure for coat prediction.

We checked ten of them. Not one states how often its predicted coat outcome matches the adult dog. Some argue from methodology; at least one quotes figures from the 2009 paper rather than from any validation of its own product. Where an accuracy percentage does appear, it is generally the accuracy of genotyping — how reliably the test reads the DNA — which is a different thing entirely from how reliably that genotype predicts a coat.

Both numbers can be true at once. A test can read your dog's KRT71 genotype with 99% reliability and still tell you very little about what the puppy's coat will be, because the uncertainty is in the biology, not the assay.

---

What to actually do

1. Test furnishings on every dog you breed. It is the one result that is solid, and it prevents an expensive, avoidable outcome.
2. Record the lab and the test name alongside the genotype. Not just N/N. Without that, your own records may not mean what you think in two years.
3. Treat shedding as a probability, and say so to buyers in those terms.
4. Stop pricing on predicted curl. Record it; do not sell on it.
5. Describe coats, do not promise them. Say what the parents look like and what the last litter did. That is honest, and it is more informative to a buyer than a category no one can define.
6. Never describe a puppy as hypoallergenic. Reduced dispersal in the home is the defensible version.

Testing is not the problem here. Testing is the best tool you have. The problem is that it has been sold to you as answering questions it was never able to answer, alongside one question it answers very well.