Species distribution models built from four broad climate predictors indicate that North American trees and shrubs occupy on average only about 48.6% of their potential climatic ranges (Seliger et al., 2021), yet the cause of this underfilling remains unattributed. The open problem is to quantitatively partition the observed underfilling across the North American tree flora into (a) postglacial dispersal and colonization lag (including intrinsic, trait-influenced limitations and extrinsic environmental barriers) and (b) habitat constraints omitted from climate-only models — soil properties, disturbance regimes, fine-scale microclimates, and biotic interactions — while accounting for the methodological artefacts the source documents as affecting these filling estimates (expert range-map inaccuracy for rare or fragmented species, spatial-grain mismatch, and model commission error); and to establish what this attribution implies for forecasting and conserving small-ranged, non-climatically limited tree species under climate change.

Background

A species' realized range is the area it actually occupies, while its potential range is the area where environmental conditions appear suitable. Species distribution models (SDMs) — statistical models relating occurrence records to environmental predictors — are routinely used to map potential ranges, most often from climate variables alone. Range filling is the occupied proportion of the modelled potential range, and climatic equilibrium is the assumption that realized ranges closely track climatically suitable areas; when they do not, the species or flora is in climatic disequilibrium. Because trees are long-lived and sessile, the Quaternary glaciations offer a natural experiment: ice sheets retreated roughly 11,000–18,000 years ago, and species that could not keep pace with shifting suitable climate should still occupy less of their potential range than they could (dispersal lag). Seliger, McGill, Svenning and Gill (2021, Journal of Biogeography 48(2): 359–371, doi:10.1111/jbi.14001) performed a range-filling analysis of 447 North American tree and shrub species using SDMs driven by exactly four climate predictors tied to plant physiology (minimum temperature of the coldest month, growing degree days above 5 °C, water balance as precipitation divided by potential evapotranspiration, and precipitation timing). They computed range filling without thresholding suitability maps, ran the analysis at 5- and 10-arc-minute resolutions with three algorithms (bioclimatic envelopes, MaxEnt — a maximum-entropy machine-learning method for presence-only occurrence data — and generalized additive models), and compared realized filling against a null model built from simulated ranges generated by a spreading-dye algorithm — a null model in which suitable area is filled by random contiguous spread, representing expectation without ecological dispersal or habitat constraints. Range shape ratios (the latitudinal divided by the longitudinal extent of a range) served as a second diagnostic: climatically constrained ranges tend to follow climate bands and differ in shape from ranges blocked by directional dispersal barriers. The headline result was widespread underfilling — a mean range filling of 48.6%, comparable to earlier estimates for European trees (about 38%; Svenning and Skov, 2004) — with filling strongly increasing with range size: large-ranged species exceeded the null model and showed climate-conforming shapes, while small-ranged species fell below the null model and showed shapes indicative of dispersal limitation. The paper leaves the central attribution unresolved. With only climate predictors, an unfilled area inside the climatic potential range is ambiguous between two interpretations: the area is suitable but was never reached (postglacial dispersal lag or other barriers), or the area is not actually suitable because determinants absent from the model — soils, disturbance regimes such as fire, fine-scale microclimates, and biotic interactions such as competition, facilitation, and mycorrhizal symbioses — exclude the species there. The paper also documents why the question is hard to settle with existing tools: attempts to link species' functional traits (seed mass, dispersal vectors, generation time, mutualists) to range size, filling, or postglacial migration rates have been surprisingly unsuccessful at biogeographic scales; expert-drawn range maps are least accurate for the rare, fragmented species whose underfilling is most extreme; coarse grid cells overstate potential ranges for elevational specialists; and SDMs biased toward commission errors (predicting presence where the species cannot persist) systematically underestimate filling. Adjacent mechanistic routes exist but have not been applied to this flora: demography-based potential ranges, defined by where the intrinsic population growth rate r0r_0 is positive rather than by climatic suitability, revealed that niche–distribution mismatches in South African plants are largest for poorly dispersed and highly persistent species (Pagel et al., 2020); and a global analysis of thermal tolerances found temperate terrestrial species systematically absent from warm, thermally tolerable equatorward areas, suggesting cold-adaptation trade-offs and biotic exclusion rather than dispersal lag (Moore et al., 2023).

Problem Statement

For the North American tree and shrub flora analyzed by Seliger et al. (2021) — 447 species whose climate-modelled potential ranges are broadly underfilled (mean filling 48.6%) — determine, quantitatively and with uncertainty, how much of the observed underfilling is attributable to (a) dispersal and colonization lag (including postglacial failure to reach suitable areas, intrinsic trait-influenced limitations, and extrinsic barriers) and (b) non-climatic habitat constraints absent from the four-predictor climate SDMs (soil properties, disturbance regimes, fine-scale microclimates, and biotic interactions such as competition, facilitation, and symbioses), while accounting for the methodological artefacts the source documents as contaminating the filling estimates being partitioned (expert range-map inaccuracy for rare or fragmented species, spatial-grain mismatch for elevational specialists, and SDM commission error). The partition must be a validated attribution of the drivers of climatic disequilibrium, robust enough to determine what it implies for forecasting and conserving small-ranged, non-climatically limited tree species under climate change (e.g., whether forecasting requires edaphic and biotic layers and explicit dispersal models, and where managed relocation or other conservation measures become necessary). The attribution must cover the full 447-species flora analyzed by Seliger et al. (2021) and must resolve both factor classes (a) and (b).

References

  1. Seliger, B.J., McGill, B.J., Svenning, J.-C. & Gill, J.L. (2021) Widespread underfilling of the potential ranges of North American trees. Journal of Biogeography 48(2): 359–371. doi:10.1111/jbi.14001. https://doi.org/10.1111/jbi.14001 (source paper)
  2. Svenning, J.-C. & Skov, F. (2004) Limited filling of the potential range in European tree species. Ecology Letters 7(7): 565–573. doi:10.1111/j.1461-0248.2004.00614.x. https://doi.org/10.1111/j.1461-0248.2004.00614.x
  3. Pagel, J., Treurnicht, M., Bond, W.J., Kraaij, T., Nottebrock, H., Schutte-Vlok, A., Tonnabel, J., Esler, K.J. & Schurr, F.M. (2020) Mismatches between demographic niches and geographic distributions are strongest in poorly dispersed and highly persistent plant species. Proceedings of the National Academy of Sciences 117(7): 3663–3669. doi:10.1073/pnas.1908684117. https://doi.org/10.1073/pnas.1908684117
  4. Moore, N.A., Morales-Castilla, I., Hargreaves, A.L., Olalla-Tárraga, M.Á., Villalobos, F., Calosi, P., Clusella-Trullas, S., Rubalcaba, J.G., Algar, A.C., Martínez, B., Rodríguez, L., Gravel, S., Bennett, J.M., Vega, G.C., Rahbek, C., Araújo, M.B., Bernhardt, J.R. & Sunday, J.M. (2023) Temperate species underfill their tropical thermal potentials on land. Nature Ecology & Evolution 7(12): 1993–2003. doi:10.1038/s41559-023-02239-x. https://doi.org/10.1038/s41559-023-02239-x
  5. Arnell, M. & Eriksson, O. (2022) Landscape-scale range filling and dispersal limitation of woody plants. Journal of Biogeography 49(11): 2028–2036. doi:10.1111/jbi.14485. https://doi.org/10.1111/jbi.14485
  6. Puglielli, G., Tordoni, E., Humphreys, A.M., Kalwij, J.M., Hutchings, M.J. & Laanisto, L. (2021) The cold-drought tolerance trade-off in temperate woody plants constrains range size, but not range filling. bioRxiv preprint. doi:10.1101/2021.07.24.453622. https://doi.org/10.1101/2021.07.24.453622
  7. Puglielli, G., Tordoni, E., Laanisto, L., Kalwij, J.M., Hutchings, M.J. & Humphreys, A.M. (2023) Abiotic stress tolerance can explain range size and filling in temperate woody plants. Perspectives in Plant Ecology, Evolution and Systematics 59: 125734. doi:10.1016/j.ppees.2023.125734. https://doi.org/10.1016/j.ppees.2023.125734
  8. Prasad, A., Peters, M., Pedlar, J., Gougherty, A., McKenney, D., Mora, F., Matthews, S., McNulty, S. & López-Mata, L. (2025) Habitat and Migration Dynamics of North American Tree Species Under Climate Change. Journal of Biogeography 52(7): e15138. doi:10.1111/jbi.15138. https://doi.org/10.1111/jbi.15138
  9. Essl, F., García-Rodríguez, A., Lenzner, B., Alexander, J.M., Capinha, C., Gaüzère, P., Guisan, A., Kühn, I., Lenoir, J., Richardson, D.M., Rumpf, S.B., Svenning, J.-C., Thuiller, W., Zurell, D. & Dullinger, S. (2024) Potential sources of time lags in calibrating species distribution models. Journal of Biogeography 51(1): 89–102. doi:10.1111/jbi.14726. https://doi.org/10.1111/jbi.14726
  10. Hanberry, B.B. (2022) Climate Envelopes Do Not Reflect Tree Dynamics after Euro-American Settlement in Eastern North America. Land 11(9): 1536. doi:10.3390/land11091536. https://doi.org/10.3390/land11091536
  11. Baranger, A., Cordonnier, T., Charrier, G., Delzon, S., Larter, M., Martin-StPaul, N.K. & Kunstler, G. (2026) Living on the edge – physiological tolerance to frost and drought explains range limits of 35 European tree species. Ecography 2026(8): e07528. (First published online 3 December 2024.) doi:10.1111/ecog.07528. https://doi.org/10.1111/ecog.07528