Predicting future wolf density in Scandinavia under land-use and climate scenarios with an individual-based hybrid model
Abstract
Managing wolves in Scandinavia requires projections of the density and distribution of territories under future land-use and climatic change. We coupled machine-learning estimators of deer availability, habitat suitability and the spatial pattern of human-caused mortality with a deterministic pack-level model of recruitment, dispersal, territory formation and territory size on a 50 × 50 km grid, and simulated Scandinavian wolf (Canis lupus) territories from 1999 to 2060 under five Shared Socioeconomic Pathways (SSPs). We tested whether territorial wolf density increases (H1a), decreases (H1b) or remains unchanged (H1c), and whether its distribution becomes more homogeneous (H2a), more heterogeneous (H2b) or remains unchanged (H2c). Without parameters fitted to the coupled trajectory, the model reproduced the colonisation of 1999–2015 (territories, R² = 0.90) and, out of sample, the plateau of 2016–2024 (84 simulated against 76 observed territories in 2024). Territories peaked at about 90 around 2040 and declined to 81–84 by 2060 under all pathways, while mean territory size fell from 807 to 596–666 km² and the density-weighted centroid moved from 59.9 to 57.6–58.0 °N. Territorial density increased under every pathway (H1a; SSP5-8.5: +16% in the last epoch relative to 1999–2015, Holm-adjusted p < 0.001). Its spatial heterogeneity was 16–19% higher than in the baseline but showed no significant trend under SSP5-8.5 (H2c) and increased significantly only under SSP1-1.9 and SSP1-2.6 (H2b). Regime-shift analysis identified three to four shifts, the last between 2027 and 2035, followed by a regime that persisted to 2060. Territory counts differed by 3% among pathways and density by 12%, and both the density increase and the southward shift depended on the spatial pattern of human-caused mortality. Wolf dynamics to 2060 are therefore governed by human-caused mortality and inbreeding rather than by climate.
Related
- Predicting Habitat Suitability of Nine Major Grassland Insect Pests in Yunnan, China, Under Climate Change Scenarios
- Spain’s seafood futures to 2050 under climate change and contrasting transition scenarios
- A multi-scale dataset combining 3D plant architecture, leaf gas exchange, and whole-plant fluxes in young oil palm under controlled climate scenarios
- Land-use change causes rapid carbon losses in Congo Basin peatlands across climate scenarios, whereas the effects of climate change alone are uncertain
- Predicting the Potential Climatic Suitability of Typical Poisonous Weeds on the Qinghai–Tibet Plateau Under Climate Change Using Biomod2 and a Single Global Climate Model
BibTeX
@misc{ghosh2026predicting,
title = {{Predicting future wolf density in Scandinavia under land-use and climate scenarios with an individual-based hybrid model}},
author = {Sinchan Ghosh and Oskar Franklin and Barbara Zimmermann},
howpublished = {EcoEvoRxiv},
year = {2026},
month = nov,
doi = {10.32942/x2rx14},
url = {https://ecoevorxiv.org/repository/view/16106/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Predicting future wolf density in Scandinavia under land-use and climate scenarios with an individual-based hybrid model
AU - Sinchan Ghosh
AU - Oskar Franklin
AU - Barbara Zimmermann
PY - 2026
DA - 2026/11/14
PB - EcoEvoRxiv
DO - 10.32942/x2rx14
UR - https://ecoevorxiv.org/repository/view/16106/
AB - Managing wolves in Scandinavia requires projections of the density and distribution of territories under future land-use and climatic change. We coupled machine-learning estimators of deer availability, habitat suitability and the spatial pattern of human-caused mortality with a deterministic pack-level model of recruitment, dispersal, territory formation and territory size on a 50 × 50 km grid, and simulated Scandinavian wolf (Canis lupus) territories from 1999 to 2060 under five Shared Socioeconomic Pathways (SSPs). We tested whether territorial wolf density increases (H1a), decreases (H1b) or remains unchanged (H1c), and whether its distribution becomes more homogeneous (H2a), more heterogeneous (H2b) or remains unchanged (H2c). Without parameters fitted to the coupled trajectory, the model reproduced the colonisation of 1999–2015 (territories, R² = 0.90) and, out of sample, the plateau of 2016–2024 (84 simulated against 76 observed territories in 2024). Territories peaked at about 90 around 2040 and declined to 81–84 by 2060 under all pathways, while mean territory size fell from 807 to 596–666 km² and the density-weighted centroid moved from 59.9 to 57.6–58.0 °N. Territorial density increased under every pathway (H1a; SSP5-8.5: +16% in the last epoch relative to 1999–2015, Holm-adjusted p < 0.001). Its spatial heterogeneity was 16–19% higher than in the baseline but showed no significant trend under SSP5-8.5 (H2c) and increased significantly only under SSP1-1.9 and SSP1-2.6 (H2b). Regime-shift analysis identified three to four shifts, the last between 2027 and 2035, followed by a regime that persisted to 2060. Territory counts differed by 3% among pathways and density by 12%, and both the density increase and the southward shift depended on the spatial pattern of human-caused mortality. Wolf dynamics to 2060 are therefore governed by human-caused mortality and inbreeding rather than by climate.
ER -