Tracking data from adult wild boar across southern and central Sweden reveals that individuals substantially contract their biweekly spatial ranges in areas characterized by higher population densities and abundant forest cover, according to a multi-scale study published in Movement Ecology. Researchers quantified the space use of 99 adult animals between 2004 and 2021 by pairing autocorrelated kernel density estimation with spatial Durbin econometric modeling and fine-scale integrated step selection analysis. The investigation evaluated space use across a 14-day temporal scale, capturing immediate behavioral adjustments within established ranges while aligning with the asymptomatic incubation window critical for pathogen transmission. Across the study cohort, biweekly space use averaged 20.31 square kilometers, with an overall median of 3.16 square kilometers. The findings demonstrate that spatial contraction is structured by both local habitat conditions and environmental characteristics in neighboring areas, establishing that broad landscape-scale space use emerges directly from fine-scale behavioral choices.
The investigation addresses accelerating ecological and agricultural concerns stemming from the rapid expansion of wild boar, scientifically known as Sus scrofa, across Europe over recent decades. This demographic surge has been facilitated by the species’ high adaptability to varied landscapes and the growing availability of human-derived food resources. Expanding populations have caused widespread crop damage and heightened the risk of spreading infectious diseases, particularly African swine fever. Wildlife managers require detailed knowledge of wild boar spatial dynamics because animals infected with African swine fever can remain infectious for several days before developing visible clinical symptoms. During this asymptomatic transmission window, roaming individuals can distribute the virus across the entire geographic extent of their active range before mortality occurs. Consequently, evaluating space use and movement paths over an intermediate biweekly timeframe provides a biologically meaningful foundation for designing effective surveillance barriers, population control interventions, and targeted containment strategies.
The research was conducted across southern and central Sweden between approximately 55 degrees and 60 degrees North latitude and 12 degrees to 18 degrees East longitude. The geographic distribution of wild boar in Sweden is situated predominantly south of latitude 62 degrees North, with national harvest records indicating that the species is rarely detected north of this boundary. The study region features a heterogeneous environmental mosaic, transitioning from dense boreal forests in the north to intensive agricultural landscapes in the south, interspersed with numerous lakes, wetlands, and mixed forest-farmland mosaics. Vegetation spans temperate to boreal zones, including mixed and coniferous woodlands, grasslands, and wetlands that offer fluctuating resource availability.
To reconstruct movement pathways, field biologists collared 99 adult wild boar, consisting of 16 males and 83 females, across southern and central Sweden between 2004 and 2021. Animals were captured using diverse capture protocols, including chemical immobilization from an elevated hide, capture from a vehicle, and corral live-trapping. The subjects were fitted with Vectronic Aerospace GPS and GSM collars, including Vertex Plus 2D and 3D, Vertex Lite 2D, and GPS Pro Light 3D units, which logged positional coordinates at intervals typically ranging from 30 to 60 minutes. To standardize data across individuals, researchers filtered records to retain fixes closest to regular one-hour intervals. Monitoring duration varied among individual animals, spanning from 13 to 364 days. All spatial processing, movement path derivation, econometric modeling, and cartographic visualizations were executed in R version 4.4.2 using specialized spatial packages including sf, spdep, spatialreg, ctmm, amt, terra, and data.table.
Space use was calculated across discrete 14-day intervals using autocorrelated kernel density estimation to produce 95 percent utilization distribution polygons. This biweekly window was chosen because it provides sufficient temporal breadth to maintain statistical stationarity while capturing immediate behavioral shifts rather than long-term home range relocations. To ensure robust estimates, intervals covering fewer than 10 tracking days were excluded. Variogram analyses using continuous-time movement models confirmed that 98.4 percent of intervals—representing 1,364 out of 1,386 evaluated segments—conformed to range-resident movement processes, primarily Ornstein-Uhlenbeck Foraging anisotropic models (966 intervals) followed by standard Ornstein-Uhlenbeck anisotropic forms. Only 22 intervals, or 1.6 percent, were best described by independent models lacking spatial autocorrelation. In total, 1,386 biweekly polygons were retained for statistical modeling from 1,604 initial polygons, with each tracked wild boar yielding between 1 and 26 polygons, reflecting an average of 14 polygons and a median of 13.
To analyze the fine-scale movement mechanisms generating these space-use extents, the authors implemented integrated step selection analysis within a conditional logistic regression framework. Movement paths were derived from tracking records by identifying movement bursts containing at least three consecutive locations, from which discrete step lengths and turning angles were calculated. Steps exceeding 10,000 meters were removed to filter out biologically unrealistic displacements caused by collar errors. For each verified empirical step, researchers generated 10 random available steps to represent potential alternative movement options within the landscape. This procedure produced 36,430 observed steps, averaging 383.5 steps per individual animal, which were paired with 364,300 available steps to create a combined dataset of 400,730 observations organized into 36,430 step sets. Models incorporated movement metrics alongside environmental covariates, individual sex, land cover classes, population density, and categorical indicators for hunting seasons where dogs were permitted.
Landscape-scale spatial dependencies were evaluated using a spatial Durbin model, which simultaneously estimates spatial autoregression in space-use area and indirect spatial spillovers arising from explanatory variables in surrounding areas. Lagrange multiplier diagnostics on ordinary least squares residuals revealed significant spatial lag and error dependence, confirming the necessity of a spatial model. The spatial weight matrix explicitly excluded successive observations from the same individual, restricting autoregression strictly to spatial dependence among distinct animals. To identify the optimal geographic interaction threshold, candidate distances from 1 to 20 kilometers were tested. The 7-kilometer threshold demonstrated the best model performance, yielding the lowest Akaike information criterion value of 16,135.5, eliminating residual spatial autocorrelation (Moran’s I dropped from 0.218 to a non-significant 0.031), maintaining 79.87 percent network connectivity, and generating a maximum spatial autoregressive coefficient of 0.233. Effect decomposition was evaluated using 1,000 Monte Carlo replications.
Durbin model estimations revealed that land cover composition exerts complex, scale-dependent influences on biweekly space-use size. Proportions of forest and open land within the focal polygon exhibited significant negative direct effects, driving substantial range contraction. In contrast, the proportion of arable land within the focal area showed no statistically significant direct effect. However, arable land in neighboring areas within the 7-kilometer radius produced a statistically significant negative indirect effect, leading to a significant negative total effect. Neighboring forest cover also exerted a significant negative indirect effect on local space-use size, whereas neighboring open land produced no significant indirect spillover. These patterns indicate that while open land restricts movement locally due to perceived exposure and elevated risk, abundant agricultural fields and forests in the surrounding landscape provide rich nutritional resources that satisfy energetic demands locally, thereby eliminating the need for wide-ranging foraging movements across exposed areas.
Population density and hunting practices demonstrated distinct patterns across spatial scales. In the Durbin model, higher wild boar density within the focal area showed a significant negative direct effect and a significant negative total effect on biweekly range size, confirming that animals contract their space use in high-density environments. The indirect effect of neighboring density was positive but not statistically significant. Periods permitting hunting dogs produced no statistically significant direct, indirect, or total effects on biweekly range size, and the interaction between density and dog hunting was non-significant at the landscape scale. Conversely, integrated step selection analysis showed that fine-scale movement velocity responded to both factors. Higher population densities and the presence of hunting dogs were significantly associated with reduced step lengths. The negative effect of density on movement velocity was slightly amplified during dog-hunting periods, demonstrating that disturbance constrains fine-scale movement without altering broad 14-day range dimensions.
Pronounced behavioral differences emerged between sexes across both analytical scales, accompanied by important methodological considerations. Males occupied substantially larger biweekly areas, averaging 54.27 square kilometers (median 6.90 square kilometers across 240 observations), than females, which averaged 14.07 square kilometers (median 2.75 square kilometers across 1,307 observations). The Durbin model revealed a significant positive direct effect for males and a positive neighboring male effect approaching significance (p = 0.075), while step selection models indicated higher movement rates for males, reflecting expansive roaming mating strategies. The authors noted several limitations, including annual hunting bag data compared to biweekly movement tracking, poorly documented supplemental feeding across hunting grounds, an uneven sample comprising 16 males and 83 females, and an exclusive focus on resident adult animals rather than dispersing juveniles. Nonetheless, integrating spatial Durbin modeling with step selection successfully links movement decisions to landscape-scale space use, supporting spatially targeted wildlife management and pathogen control.
Subject of Research: Multi-scale analysis of environmental drivers, population density, and space use in Swedish wild boar
Article Title: Multi-scale analysis of spatial drivers in wild boar space use at a biweekly scale across landscape patterns and step selection
Article References: Kim, H., Widgren, S., Johansson, R., Kjellander, P., MÃ¥nsson, J., Malmsten, J., Thurfjell, H., & Rosendal, T. (2026). Multi-scale analysis of spatial drivers in wild boar space use at a biweekly scale across landscape patterns and step selection. Movement Ecology, 14(1), Article 61. https://doi.org/10.1186/s40462-026-00696-0
Image Credits: AI Generated
DOI: 10.1186/s40462-026-00696-0
Keywords: wild boar, Sus scrofa, space use, movement ecology, spatial Durbin model, integrated step selection analysis, autocorrelated kernel density estimation, GPS tracking, Sweden, African swine fever, population density, hunting bag statistics, forest cover, arable land, step length, sex differences
News Source: Drew Townsend. (November 20, 2026). Wild Boar Contract Biweekly Ranges in High Density and Dense Forests. Scienmag.



