# UCP family evolution and mammalian UCP1 selection analysis

**Dormanomics Analysis Release:** 2026-10-25  
**Scope:** UCP1/UCP2/UCP3 family resolution and mammalian UCP1 coding-sequence evolution  
**Primary dataset:** 30 mammalian species; 12 deep hibernators and 18 non-hibernating controls

## Main conclusion

Mammalian UCP1 is predominantly under purifying selection at the whole-gene coding-sequence level. The primary analyses do not support a shared shift in UCP1 selection pressure among deep hibernators, selection relaxation or intensification on deep-hibernator branches, or episodic diversifying positive selection in those branches.

This result narrows the evolutionary hypothesis: a common coding-sequence adaptation across deep hibernators is not supported by the present data. It does not show that UCP1 is unrelated to dormancy. Expression regulation, brown-adipose-tissue-specific control, protein activity, individual sites, and lineage-specific mechanisms remain open questions.

## Data and quality control

- 202 representative UCP family proteins: 56 UCP1, 81 UCP2, and 65 UCP3.
- 55 of 56 UCP1 representatives passed CDS–protein consistency checks.
- 42 mammalian UCP1 sequences formed a supported clade in the family tree (UFBoot = 99).
- The primary analysis included 30 mammalian species from that clade.
- Protein alignment: MAFFT L-INS-i; codon alignment: CDS-guided back-translation.
- Full codon alignment: 30 sequences and 313 codons; 80% occupancy alignment: 30 sequences and 306 codons.
- Selection analyses used an independent species topology rather than treating the UCP1 gene tree as the species tree.

## Primary results

| Method | Estimate | Test statistic | P value | Interpretation |
|---|---|---:|---:|---|
| PAML M0 | whole-gene ω = 0.15558 | — | — | Predominant purifying selection |
| PAML two-ω branch model | deep hibernators ω = 0.16666; controls ω = 0.15345 | LRT = 0.278564 | 0.5976 | No supported between-group difference |
| HyPhy RELAX | K = 1.0262 | LRT = 0.005975 | 0.9384 | No supported relaxation or intensification |
| HyPhy BUSTED | — | LRT = 0 | 0.5000 | No supported episodic diversifying positive selection |

A ω below 1 indicates purifying selection on average; a difference between two ω estimates is not, by itself, evidence of positive selection.

## Sensitivity analyses

| Dataset | PAML hibernator ω | PAML control ω | PAML P | RELAX K | RELAX P |
|---|---:|---:|---:|---:|---:|
| Primary, 80% occupancy | 0.16666 | 0.15345 | 0.5976 | 1.0262 | 0.9384 |
| Full alignment | 0.16488 | 0.15472 | 0.6833 | 1.0769 | 0.8037 |
| Bats removed | 0.11276 | 0.13372 | 0.4606 | 1.1648 | 0.5943 |
| Sciurids removed | 0.17220 | 0.15922 | 0.6336 | 3.3071 | 0.1718 |
| Bears removed | 0.16323 | 0.15556 | 0.7598 | 1.0672 | 0.8534 |
| Hedgehog removed | 0.20292 | 0.15262 | 0.1125 | 0.8147 | 0.4169 |

None of the six PAML branch comparisons or six RELAX tests was statistically significant.

## Reproducibility

Software: MAFFT v7.505; IQ-TREE 3.1.2 (Q.BIRD+I+R4, 1000 UFBoot replicates); PAML/CODONML 4.10.10; HyPhy 2.5.103. The full archive contains inputs, alignments, trees, model outputs, scripts, logs, and result summaries. The small downloadable files on the release page are convenience copies of key outputs.