Biology subjects

Tilker, A.

Publications and source records attributed to Tilker, A..

3 recordsLinked to original sources

bioRxiv2023

Can snare removal safeguard protected areas caught in the Southeast Asian snaring crisis? Promises and perspectives

Unsustainable snaring is causing biodiversity declines across tropical protected areas, resulting in species extinctions and jeopardizing the health of forest ecosystems. Here, we used 11 years of ranger collected data to assess the impact of intensive snare removal on snaring levels in two protected areas in Viet Nam. Snare removal resulted in significant declines in snare occupancy (36.9, 95% BCI [4.6, 59.0] reduction in percent area occupied), but snaring levels nonetheless remained high (31.4, [23.6, 40.8] percent area occupied), and came with a substantial financial cost. Our results indicate that snare removal remains an important component of efforts to protect tropical protected areas, but by itself, is likely insufficient to address this threat. To stop snaring in protected areas, a multifaceted approach will be necessary that combines short-term reactive snare removal with long-term proactive programs that address the underlying drivers behind snaring.

ecology↗

bioRxiv2021

imageseg: an R package for deep learning-based image segmentation

O_LIConvolutional neural networks (CNNs) and deep learning are powerful and robust tools for ecological applications. CNNs can perform very well in various tasks, especially for visual tasks and image data. Image segmentation (the classification of all pixels in images) is one such task and can for example be used to assess forest vertical and horizontal structure. While such methods have been suggested, widespread adoption in ecological research has been slow, likely due to technical difficulties in implementation of CNNs and lack of toolboxes for ecologists. C_LIO_LIHere, we present R package imageseg which implements a workflow for general-purpose image segmentation using CNNs and the U-Net architecture in R. The workflow covers data (pre)processing, model training, and predictions. We illustrate the utility of the package with two models for forest structural metrics: tree canopy density and understory vegetation density. We trained the models using large and diverse training data sets from a variety of forest types and biomes, consisting of 3288 canopy images (both canopy cover and hemispherical canopy closure photographs) and 1468 understory vegetation images. C_LIO_LIOverall classification accuracy of the models was high with a Dice score of 0.91 for the canopy model and 0.89 for the understory vegetation model (assessed with 821 and 367 images, respectively), indicating robustness to variation in input images and good generalization strength across forest types and biomes. C_LIO_LIThe package and its workflow allow simple yet powerful assessments of forest structural metrics using pre-trained models. Furthermore, the package facilitates custom image segmentation with multiple classes and based on color or grayscale images, e.g. in cell biology or for medical images. Our package is free, open source, and available from CRAN. It will enable easier and faster implementation of deep learning-based image segmentation within R for ecological applications and beyond. C_LI

ecology↗

bioRxiv2019

Identifying conservation priorities in a defaunated tropical biodiversity hotspot

AimUnsustainable hunting is leading to widespread defaunation across the tropics. To mitigate against this threat with limited conservation resources, stakeholders must make decisions on where to focus anti-poaching activities. Identifying priority areas in a robust way allows decision-makers to target areas of conservation importance, therefore maximizing the impact of conservation interventions.\n\nLocationAnnamite mountains, Vietnam and Laos.\n\nMethodsWe conducted systematic landscape-scale surveys across five study sites (four protected areas, one unprotected area) using camera-trapping and leech-derived environmental DNA. We analyzed detections within a Bayesian multi-species occupancy framework to evaluate species responses to environmental and anthropogenic influences. Species responses were then used to predict occurrence to unsampled regions. We used predicted species richness maps and occurrence of endemic species to identify areas of conservation importance for targeted conservation interventions.\n\nResultsAnalyses showed that habitat-based covariates were uninformative. Our final model therefore incorporated three anthropogenic covariates as well as elevation, which reflects both ecological and anthropogenic factors. Conservation-priority species tended to found in areas that are more remote now or have been less accessible in the past, and at higher elevations. Predicted species richness was low and broadly similar across the sites, but slightly higher in the more remote site. Occupancy of the three endemic species showed a similar trend.\n\nMain conclusionIdentifying spatial patterns of biodiversity in heavily-defaunated landscapes may require novel methodological and analytical approaches. Our results indicate to build robust prediction maps it is beneficial to sample over large spatial scales, use multiple detection methods to increase detections for rare species, include anthropogenic covariates that capture different aspects of hunting pressure, and analyze data within a Bayesian multi-species framework. Our models further suggest that more remote areas should be prioritized for anti-poaching efforts to prevent the loss of rare and endemic species.

ecology↗