Kenneth F. Kellner’s research while affiliated with Michigan State University and other places

... Therefore, they may suffer from low interoperability, which can create friction when using multiple tools together. As an illustration, Kellner et al. (2025) highlight that about 20% of publications for abundance or distribution models are not reproducible because of issues in package dependencies. ...

Reference:

A Julia toolkit for species distribution data

... We thinned brown bear GPS points to one randomly-selected point every four hours to reduce spatial and temporal autocorrelation ("used points"; Swihart and Slade, 1985). Because brown bears on Afognak and Raspberry islands have large ranges and can use the entirety of both islands (Finnegan et al., 2021;Schooler et al., 2024), we compared individual habitat use to population level habitat availability to estimate landscape-scale brown bear relative probability of habitat selection ("probability of selection"; Thomas and Taylor, 2006). We generated random points equal to 10 times the number of brown bear GPS locations within the boundary of Afognak and Raspberry islands ("available points"; Fieberg et al., 2021;Fig. ...

Reference:

Sex, size, and dependent young influence how brown bears select habitat relative to salmon streams

... We identified research priorities necessary to further our understanding of wildlife harvests. First, the roles and ecosystem service benefits of wildlife harvests to human societies need to be quantified 12 . With these data, we can identify and forecast patterns of harvests, their distributions, and measure sustainability within established goals and frameworks. ...

Reference:

Wildlife harvests can advance food security and the food systems agenda

... A growing body of scientific literature has examined various effects of culling on wolf populations and pack dynamics, ranging from habitat selection shifts (Baillie-David et al. 2024) to increased hybridization with coyotes (Rutledge 2012) and turnover of individuals within packs (Bassing et al. 2019). Some studies suggest that culling can lead to pack dissolution, leaving behind less-experienced surviving members that may disperse (Brainerd et al. 2008;Wallach et al. 2009;Borg et al. 2015), while others have shown that low levels of harvest or agency removal do not significantly affect pack persistence (Zubiria Perez et al. 2024). Dispersing single wolves tend to navigate the landscape differently than those belonging to packs -avoiding threats posed by territorial packs (Kirilyuk et al. 2021) and wandering closer to towns and roads (Mancinelli et al. 2019;Barry et al. 2020). ...

Reference:

Landscape of risk: responses of grey wolves to lethal control in a mosaic landscape

... These regions are underrepresented even across the complete set of studies within BioTIME, suggesting that we lack long-term (≥ 10 years) monitoring studies across regions that face the strongest environmental filtering, with some of the most rapid rates of global change (Rantanen et al. 2022). There is strong evidence that biotic interactions can structure ecological communities (Godsoe et al. 2015), influencing how communities respond to change, such as in seasonal spatial patterns (Wehr et al. 2024). Accordingly, we expected that pairs of genera with documented GloBI interactions would exhibit stronger correlations of abundance change over time. ...

Reference:

Temporal changes in taxon abundances are positively correlated but poorly predicted at the global scale

... Wolf predation on adult-sized deer primarily occurs during October-April with predation generally peaking in February-April (Fuller 1990;Vucetich et al. 2012;Wehr et al. 2024). ...

Reference:

Wolf Predation on White‐tailed Deer Before, During, and After a Historically Mild Winter in Northern Minnesota

... Where additional data sources that are more directly informative about abundance than unmarked detection/non-detection data are available, for example, distance sampling data, or individual identity data, these data could be integrated into the model. This integration could significantly improve model performance and be achieved simply by using an additional observation model (e.g., that of a standard distance sampling model; Kéry et al., 2024 or a simple SCR model; Royle et al., 2014), with a shared abundance state. ...

Reference:

Abundance‐mediated species interactions

... Our results support the hypothesis that wolves avoid linear features with high levels of human activity (Dickie et al., 2017;McKenzie et al., 2012;van den Bosch et al., 2023). When in isolation, wolves exhibited a neutral response to primary road proximity. ...

Reference:

Movement‐integrated habitat selection reveals wolves balance ease of travel with human avoidance in a risk–reward trade‐off

... In Serengeti National Park, Tanzania, van den Bosch et al. (2023) examined aardwolves (Proteles cristata) and aardvarks (Orycteropus afer)-both nocturnal insectivores-to examine spatiotemporal co-occurrence of these two potential competitors. In fact, multispecies occupancy modelling showed a high degree of spatial and temporal overlap in the two species, with evidence that the two are commensals: aardvarks increase food accessibility for aardwolves (van den Bosch et al. 2023). In Mozambique's Gorongosa National Park, Grabowski, Phillips, and Gaynor (2024) used camera traps to explore patterns of spatial and temporal niche partitioning among mesocarnivores: large-spotted genet (Genetta maculata), African civet (Civettictis civetta), honey badger (Mellivora capensis), and marsh mongoose (Atilax paludinosus). ...

Reference:

Ecological Insights From Camera Trapping Span Biological Taxa, and the Globe

... Many popular R packages based on formula exist that can fit various combinations of models including N-mixture models, and/or generalized linear mixed models for assessing abundance (Doser et al.,2024). However, some packages can not accommodate multiple species within a multivariate framework, such as unmark (Kellner et al., 2023), dsm (Miller et al., 2013), hSDM (Vieilledent, 2019), and ubsm, and some packages can not explicitly account for imperfect detection, such as spBayes (Finley et al., 2013), Hmsc (Tikhonov et al., 2019), and sdmTMB (Anderson et al., 2024). spAbundance of R package fits a variety of analogous spatially explicit occupancy models, drastically reducing computational run times using Nearest Neighbor Gaussian Processes, while accounting for spatial autocorrelation and imperfect detection (Doser et al., 2024). ...

Reference:

Using PLS-SEM models to explore the interactions of meteorology and landscape pattern changes on waterbird diversity: A case of the Liaohe Estuary