Museum records confirm least weasels are expanding southward, against the poleward trend

Deep in the archives of North American mammalogy lies a puzzle that has quietly divided biologists for decades. The least weasel, Mustela nivalis, is the world’s smallest carnivore, a diminutive predator with a Holarctic distribution stretching from Eurasia across the northern reaches of this continent. Under contemporary climate change, the overwhelming expectation is that species should shift their ranges poleward or upward in elevation, tracking the warming conditions to which they are ancestrally suited. Hundreds of endothermic and ectothermic taxa have been documented doing exactly that, with one landmark analysis reporting an average equatorward-to-poleward shift of 16.9 kilometres per decade. Yet in the early twentieth century, Nebraska marked the southern limit of the least weasel’s known North American range. By the 1950s, Eugene Raymond Hall, the renowned mustelid biologist and then director of the Natural History Museum at the University of Kansas, suspected the species almost certainly occurred in north-central Kansas. His graduate student, the eminent mammalogist J. Knox Jones, however, declared the weasel absent from the western and southernmost counties of Nebraska. The species was not documented in Kansas at all until 1965. After that, it became relatively abundant in the northern part of the state and appeared to disperse southward, directly contradicting the poleward paradigm.

The problem, until now, has been that such historical observations, however expert, rest on a logical foundation that modern ecology has learned to distrust. Finding a species in a new location has historically been treated as conclusive evidence of range expansion. But that framework conflates non-detection with absence. Consider a species that lives at the base of a mountain but was not found when the summit was sampled a century ago; if modern sampling of the summit later yields a record, has the species expanded, or was it simply overlooked the first time? For rare or elusive taxa, locations of perceived absence are far more likely to represent areas of non-detection. Occupancy modelling and species-accumulation analyses can resolve such questions, but they demand repeated, high-resolution sampling with a focused design that historical data rarely provide. A new study published in Frontiers of Biogeography has now tackled this problem head-on, adapting a statistical approach originally designed to measure true absence across space so that it measures absence across time, and applying it to one of the most counterintuitive suspected range shifts in North American mammalogy.

The research team, led by Peter D. Campbell of the University of Kansas, assembled an extraordinary volume of primary biodiversity data to test whether least weasels were genuinely absent from the central United States before the 1960s. From the Global Biodiversity Information Facility, they downloaded 877 North American least weasel specimen records with coordinates. Crucially, they also compiled specimen-backed occurrence records for all species in the weasel family Mustelidae, yielding 45,472 records, and for the entire order Rodentia, yielding 1,314,990 records. These rodent and mustelid specimens served as a target-group background dataset, an index of regional sampling effort. The logic is elegant: because least weasels, as the smallest species in the order Carnivora, are most frequently caught in Sherman live-traps or large rat-sized snap traps, the same traps used to survey rodents and other small mustelids, any intensive rodent trapping campaign represents sampling that could have yielded a least weasel. Where rodents were collected in abundance and least weasels were not, the weasel’s absence carries statistical weight rather than merely reflecting a gap in the historical record.

To perform the test, the researchers projected a grid of 100-kilometre hexagons across the least weasel’s North American range, creating 1,802 discrete geographic units. They split occurrence records into pre-1960s and post-1960s groups, deliberately ignoring records from the 1960s themselves to allow time for both the species’ range and environmental conditions to accumulate noticeable change. Each hexagon received four tallies: least weasel counts and background counts for each time period. In hexagons with zero pre-1960s least weasel records, they performed individual binomial tests, asking whether finding zero weasels was unexpected given the background sampling effort and the proportion of weasels detected post-1960s. A significant p value, below 0.05, indicated less than a 5 percent chance that weasels occupied that hexagon before 1960 and escaped detection by chance alone. The researchers also implemented a conservative correction, capping the probability of success at 1/77, the range-wide proportion of least weasels to background records, to prevent artificially high local proportions in hexagons with sparse post-1960s sampling. Hexagons with fewer than 77 pre-1960s background records were deemed to have insufficient historical sampling to warrant analysis at all.

The results were striking. Of the 1,802 hexagons, 66 percent had insufficient historical sampling, a sobering reminder of how unevenly the continent has been surveyed. But in areas with a deep history of specimen collection, like Kansas, statistically sound tests were possible. Thirty-five hexagons showed a pattern consistent with range expansion, zero weasel records before 1960 and at least one after. Sixteen of those showed significant evidence for a true pre-1960s absence given the sampling effort. Of the 36 hexagons overlapping Kansas, 19 were adequately sampled; 13 showed a pattern of expansion and 7 of those were statistically significant. The latitudinal and pairwise spatial clustering of the significant hexagons held up under permutation testing, confirming a coherent geographic pattern rather than a scattering of statistical flukes. In plain terms, given the tens of thousands of rodents and mustelids collected across the central Great Plains before 1960, the total absence of least weasels among them is extremely unlikely to reflect bad luck. The weasels were almost certainly not there.

With the historical absence validated, the team turned to ecological niche modelling to ask whether changing climate could explain the expansion. They calibrated Maxent models using pre-1960s occurrence data and projected those models onto post-1960s environmental conditions, drawing on the PRISM climatic datasets, which characterise climate for both periods using identical base models and methods at roughly 4-kilometre resolution. Five climatic variables were used: maximum temperature, minimum temperature, precipitation, maximum vapour pressure deficit and minimum vapour pressure deficit, transformed via principal component analysis to prevent multicollinearity. The geographic area for calibration was defined by an accessibility simulation based on the weasel’s dispersal capabilities, and a density raster of target-group background occurrences served as a bias layer to correct for geographic and environmental sampling bias. Ninety-eight candidate models were evaluated using partial ROC analysis, omission rates and AICc scores, with two final models selected. Jackknife analysis revealed that the fourth principal component, most strongly correlated with maximum vapour pressure deficit, was consistently the most influential variable, with higher suitability associated with low maximum VPD and high minimum temperature.

The niche models delivered a decisive answer. When binarised with a 5 percent omission threshold, the final models showed that 35.3 percent of Kansas was climatically suitable for least weasels before the 1960s, rising to 71.6 percent after, an increase of 102.7 percent in suitable area within the state. Across the Great Plains, suitable area grew by 14.6 percent, and across the entire accessible area by 18.1 percent. Most tellingly, post-1960s least weasel suitability within the perceived area of expansion increased by an average of 35.1 percent, a change that was statistically significant under a re-randomisation test comparing 100,000 mean differences. The area of expansion saw a significant decrease in minimum vapour pressure deficit of roughly 20 percent and a significant increase in precipitation of roughly 9 percent. In other words, the south-western leading edge of the least weasel’s range became substantially wetter and more suitable during precisely the time frame in which the weasels arrived, correlating with an increase in water availability. A sensitivity analysis, re-running the models without data thinning and at coarser 16-kilometre pixel resolution, produced qualitatively identical results.

These findings validate the expert, albeit anecdotal, accounts of mid-century mammalogists and confirm that least weasels genuinely defy the poleward migratory trend observed in other mammals. They also contradict a recent study by Cheeseman and colleagues, whose niche models incorporating land-use variables predicted a 54 percent decrease in suitable least weasel habitat in North America during the 1900s. The discrepancy likely turns on methodology. The history of land use in the Great Plains is extraordinarily complex: the Homestead Act of 1862 drove rapid conversion of native prairie to agriculture, while the Conservation Reserve Program of 1985 later reconverted more than 8 percent of farmland back into grass and shrublands, with vegetation fluctuating back and forth as economics dictated. The new study’s authors expect that including land use would contribute more noise than signal to their models, though they acknowledge that variables beyond climate alone may shape least weasel range dynamics. Whether the changing vapour pressure deficit and precipitation directly affect weasel fitness and physiology, perhaps through transpiration rates and water availability, or act indirectly through prey that prefers wetter conditions, remains an open question. The researchers suspect the weasels’ diminutive size allows them to exploit rodent burrows and similar microhabitats, behaviourally buffering them from warming ambient temperatures.

Beyond settling a decades-old debate about one small carnivore, the study offers a methodological blueprint with far-reaching implications. The weighted locality approach, adapted from Anderson’s 2003 framework and applied across time rather than space, provides a viable alternative to occupancy models when repeated survey data are unavailable, accommodating areas with limited or sporadic sampling effort and reducing bias from incomplete data. The authors suggest it could be applied to other rare or at-risk species distributed in otherwise well-sampled regions, including Franklin’s ground squirrel, Henslow’s Sparrow and the white-tailed jackrabbit. Least weasels have been in North America since the Wisconsin glaciation, yet only recently expanded south and west into the central United States. While we can never know a species’ true past distribution with certainty, statistically substantiating perceived absences reduces assumptions and better informs conservation. And the study’s deepest lesson may be about the irreplaceable value of natural history collections themselves: roughly 50,000 mammalian specimens collected in Kansas since 1900, 83 natural history collections contributing openly available data, and continued hands-on, expert-led surveys made this analysis possible. For a species not reliably identified from photographs, physical specimens and their associated data remain the only path to documenting change through time, and expanded resurveys of historically sampled sites will be critical to catching the next counterintuitive range shift empirically.

Subject of Research: Harnessing historical sampling to substantiate range shifts: southward movement of North American least weasels (Mustela nivalis)

Article Title: Harnessing historical sampling to substantiate range shifts: southward movement of North American least weasels ( Mustela nivalis )

Article References: Campbell, P. D., Wiens, B. J., Cobos, M. E., Townsend Peterson, A., & Colella, J. P. (2025). Harnessing historical sampling to substantiate range shifts: southward movement of North American least weasels (Mustela nivalis). Frontiers of Biogeography, 18, Article e131798. https://doi.org/10.21425/fob.18.131798

Image Credits: AI Generated

DOI: 10.21425/fob.18.131798

Keywords: Harnessing, historical, sampling, substantiate, range, shifts, southward, movement, North, American, least, weasels