Using machine learning to predict habitat suitability of sloth bears at multiple spatial scales

Abstract

Background Habitat resources occur across the range of spatial scales in the environment. The environmental resources are characterized by upper and lower limits, which define organisms’ distribution in their communities. Animals respond to these resources at the optimal spatial scale. Therefore, multi-scale assessments are critical to identifying the correct spatial scale at which habitat resources are most influential in determining the species-habitat relationships. This study used a machine learning algorithm random forest (RF), to evaluate the scale-dependent habitat selection of sloth bears (Melursus ursinus) in and around Bandhavgarh Tiger Reserve, Madhya Pradesh, India. Results We used 155 spatially rarified occurrences out of 248 occurrence records of sloth bears obtained from camera trap captures (n = 36) and scats located (n = 212) in the field. We calculated focal statistics for 13 habitat variables across ten spatial scales surrounding each presence-absence record of sloth bears. Large (> 5000 m) and small (1000–2000 m) spatial scales were the most dominant scales at which sloth bears perceived the habitat features. Among the habitat covariates, farmlands and degraded forests were the essential patches associated with sloth bear occurrences, followed by sal and dry deciduous forests. The final habitat suitability model was highly accurate and had a very low out-of-bag (OOB) error rate. The high accuracy rate was also obtained using alternate validation matrices. Conclusions Human-dominated landscapes are characterized by expanding human populations, changing land-use patterns, and increasing habitat fragmentation. Farmland and degraded habitats constitute ~ 40% of the landform in the buffer zone of the reserve. One of the management implications may be identifying the highly suitable bear habitats in human-modified landscapes and integrating them with the existing conservation landscapes.

Habitat occupancy of sloth bear Melursus ursinus in Chitwan National Park, Nepal

“…Human–sloth bear conflict is common throughout the year in CNP, suggesting that sloth bears perceive humans as a threat (Acharya et al, 2016; Lamichhane et al, 2018; Silwal et al, 2017). Previous reports of sloth bears from degraded forests were likely because the study was conducted in an area of degraded forests and should not be taken as the norm in terms of sloth bear ecology (Rather et al, 2021) but rather as the manifestation of a high nexus between sloth bears and humans in the landscape. Sloth bears might use disturbed habitats in moderation for food, water, and shelter.…”

Explainable Boosting Machines for Slope Failure Spatial Predictive Modeling

“…Since a large portion of the landscape is not a slope failure initiation location, there is a greater chance of FPs than is represented in our validation sample. When class proportions are not known, such as for predicting habitat suitability or future landscape change, it is common to use a class-balanced validation set (for example [101][102][103]).…”

Structural complexity characterizes fine‐scale forest conditions used by Pacific martens

“…We used decision trees to evaluate fine‐scale use of marten denning and resting locations. Decision trees and other supervised machine learning approaches (e.g., random forest) attempt to model the relationship between a response and its predictors (Breiman 2001), offer powerful alternatives to traditional ecological modeling approaches (e.g., generalized linear models; De'ath and Fabricius 2000, Olden et al 2008), and have been increasingly applied in investigations of wildlife habitat use (Han et al 2017, Mi et al 2017, Cushman and Wasserman 2018, Carroll et al 2021, Rather et al 2021). We built decision trees by incorporating plot‐level data into a boosted C5.0 algorithm (Quinlan 1993), which we chose because of its nominal sensitivity to multicollinearity and unbalanced data, ability to manage missing values, and relative ease of interpretation (Guilherme et al 2018, Szilassi et al 2019, Moeinaddini et al 2020, Tanyu et al 2021, da Silveira et al 2022).…”