The braincase of a long-clawed shrew gets lower in winter. That much matches a pattern first described in European common shrews in 1949. What the Hokkaido skulls add is the other end of the same head: the nasal region enlarges in those same winter months, then shrinks again in early spring.
Yugo Ikeda of Toyo University, Yukina Suwa and Satoshi D. Ohdachi of Hokkaido University report the pairing in Proceedings of the Royal Society B, in a paper dated 19 August 2026. They call the snout’s winter growth a reverse Dehnel’s phenomenon. Their preferred reading is that a larger nasal cavity may help keep the brain warm while the brain itself is getting smaller. They say that outright as a proposal, not as a measurement of heat flow.
The winter head that shrinks in one place
Dehnel’s phenomenon is a reversible seasonal change in body size, skull, brain and some internal organs. August Dehnel first described it in the common shrew Sorex araneus. Later work, mainly on that species, has reported winter reductions exceeding 20 percent in several cranial and visceral structures, followed by spring regrowth. That 20 percent figure is the prior literature’s, not a number this paper measured on the Hokkaido animals.
The authors present the new study as the first empirical account of the phenomenon in the long-clawed shrew Sorex unguiculatus, and the first detailed one in any Asian Sorex. Northeast Asia is a hotspot of shrew diversity. Until now, seasonal morphological change there had been limited to a few preliminary observations.
The authors argue from that result that the winter remodelling is not a European quirk. Their wording in the discussion is strong: they call it a widespread and universal adaptive strategy among small mammals at high latitudes. The data underneath that sentence are one species, one locality, and a collection of dead animals. The paper also notes convergent winter remodelling in moles, voles, stoats and weasels, which is why they reach for “widespread.” Universal is their word, not a count of species tested here. Later in the same discussion they cite recent comparative work suggesting the phenomenon is phylogenetically constrained rather than universally expressed, pointing to a white-toothed shrew from warmer regions that shows no seasonal reduction at all.
Linear measurements and landmark shape
The sample is 136 skulls in the Botanic Garden of Hokkaido University, all collected between 1948 and 1988, most of them in 1977 to 1979, all from Tomakomai. Only animals with known collection dates were kept. The series includes at least one specimen for each month from a first-year June through a second-year November. The youngest is a first-year animal taken in June. The oldest is an overwintered animal taken the following November.
Most soricine shrews die before a second winter. The authors split the sample, following earlier work by Ohdachi and Maekawa, into young of the year and overwintered adults, using tooth wear when the original age notes were missing. The final count is 67 males, 51 females and 18 of unknown sex. Because some cells were small, they pooled the sexes for the principal-component and canonical-variate analyses.
This is not a study of live shrews measured twice. No animal in the series was followed from autumn into spring. The seasonal curve is a comparison of different individuals collected in different months. That design can show a population pattern. It cannot show one skull changing.
Body weight was taken from the original collection notes to the nearest 0.1 gram. Skull dry weight was taken on the museum specimens to 0.1 milligram. The team photographed dorsal, ventral and right-lateral views of the cranium and a right-lateral view of the mandible, with a 0.1 millimetre scale in the frame. They then measured ten cranial and four mandibular traits in ImageJ to 0.01 millimetre. Braincase height is one of those ten. They also placed landmarks: 32 on the ventral cranium, 27 on the right lateral cranium, 15 on the mandible. Completeness cut the landmark samples to 131 ventral crania, 128 lateral crania and 125 mandibles.
Collection months were grouped into six seasonal categories. Wintering, in their scheme, is January to April.
A snout that does the opposite
Eleven of the 16 measurements, the two weights and nine of the fourteen linear traits, showed seasonal change by month. Braincase height had no significant sex effect on its own, but it did show a sex-by-month interaction. Braincase width and length did not differ by month, then did differ when the months were binned into those six seasonal categories.
The paper’s figure 1 colour-codes traits that decrease in January to April, traits that increase, and traits with no significant seasonal difference. The paper’s running text gives no millimetre means. The geometric result is the one the authors lean on for the reverse pattern.
A two-way Procrustes ANOVA found seasonal differences in all three landmark views: ventral cranium and lateral cranium at p equal to 0.001, mandible at p equal to 0.002. A sex effect appeared only on the ventral cranium. Canonical variate 1, the dominant seasonal axis, accounted for 43.17 percent of variation in the ventral cranium, 46.16 percent in the lateral cranium and 33.48 percent in the mandible.
Animals from the wintering months, their category iv, shifted farthest along that axis on both cranial views; on the mandible the divergence instead built gradually toward the second-year adults, category vi. On the lateral cranium, the winter deformation is shrinkage of the occipital elements of the braincase and enlargement of the nasal region. That is the reverse pattern in geometric form: the back of the head flattening while the snout goes the other way.
The mandible did not follow. It reinforced the body and ramus with age, and it did not show the corresponding reverse cycle. The authors treat that mismatch as evidence against a simple “stronger bite in winter” story, because a jaw that is doing the chewing does not enlarge in step with the snout.
Heat exchange is the authors’ preferred reading
They offer three non-exclusive hypotheses for the winter snout.
The first is mechanical: a sturdier rostrum might let the animal handle tougher prey when food is scarce. S. unguiculatus eats earthworms as its main prey. Three sympatric shrews eat more surface arthropods. In European S. araneus, prey composition stays largely consistent across seasons. A study of two European mole species, which the authors cite, concluded that Dehnel’s phenomenon is an adaptation to winter climate and not to a changing resource landscape alone. Combined with the jaw’s failure to reverse, they say current evidence gives little support to reinforcement as the primary driver.
The third hypothesis is smell: more olfactory epithelium might help find prey in winter. They leave it standing. They do not test odour thresholds.
The second is the one they call more strongly supported by comparative anatomical and physiological evidence. Reindeer muzzles carry dense vascular networks that exchange heat and keep the brain from cooling. If winter brain shrinkage is partly an energy trade-off, a larger nasal cavity might stabilise brain temperature even as the braincase flattens. Their geometric results, they write, show enlargement of the nasal region in the wintering animals. They did not put a thermocouple in a live shrew. Heat exchange is an inference from shape plus other species.
The pattern is a between-animal seasonal series, not a within-animal time series. Sexes were pooled for the canonical-variate analysis that produced the seasonal axis. Eighteen skulls had no sex recorded. The collection is one town on Hokkaido, mostly from three years in the late 1970s. Whether the reverse snout occurs in other Sorex on the island, or in live animals tracked through a winter, is not shown.
The 20 percent winter reductions that appear in popular summaries of Dehnel’s phenomenon belong to earlier work, mainly on S. araneus. This paper demonstrates direction (braincase down, rostrum up) and statistical seasonal separation. It does not, in the body text, give a percentage change for the Hokkaido snout.
The thermoregulation story is the authors’ best-supported guess among three. It is not a result. A reader who leaves with “shrews grow bigger noses to heat the brain” has gone one step past the paper.
What the series does show, cleanly, is that the winter shrew head is not a uniformly smaller version of the summer one. The back flattens. The front does the opposite. Then, in spring, the snout comes back down.
The 136 skulls from Tomakomai are the record of that split: a braincase that still shrinks for winter, and a snout that does not go along.