Machine Learning for Conservation Planning in a Changing Climate

Abstract

Wildlife species’ habitats throughout North America are subject to direct and indirect consequences of climate change. Vulnerability assessments for the Intermountain West regard wildlife and vegetation and their disturbance as two key resource areas in terms of ecosystems when considering climate change issues. Despite the adaptability potential of certain wildlife, increased temperature estimates of 1.67–2 °C by 2050 increase the likelihood and severity of droughts, floods, heatwaves and wildfires in Utah. As a consequence, resilient flora and fauna could be displaced. The aim of this study was to locate areas of habitat for an exemplary species, i.e., sage-grouse, based on current climate conditions and pinpoint areas of future habitat based on climate projections. The locations of wildlife were collected from Volunteered Geographic Information (VGI) observations in addition to normal temperature and precipitation, vegetation cover and other ecosystem-related data. Four machine learning algorithms were then used to locate the current sites of wildlife habitats and predict suitable future sites where wildlife would likely relocate to, dependent on the effects of climate change and based on a timeframe of scientifically backed temperature-increase estimates. Our findings show that Random Forest outperforms other competing models, with an accuracy of 0.897, and a sensitivity and specificity of 0.917 and 0.885, respectively, and has great potential in Species Distribution Modeling (SDM), which can provide useful insights into habitat predictions. Based on this model, our predictions show that sage-grouse habitats in Utah will continue to decrease over the coming years due to climate change, producing a highly fragmented habitat and causing a loss of close to 70% of their current habitat. Priority Areas of Conservation (PACs) and protected areas might be deemed insufficient to halt this habitat loss, and more effort should be put into maintaining connectivity between patches to ensure the movement and genetic diversity within the sage-grouse population. The underlying data-driven methodical approach of this study could be useful for environmentalists, researchers, decision-makers, and policymakers, among others.

From Coastal to Montane Forest Ecosystems, Using Drones for Multi-Species Research in the Tropics

“…Previous studies have shown that machine learning provides good species detection performance [73][74][75], and our study concurred that using machine learning approaches, notably thermal and RGB imageries, detected both Javan deer and water buffalo populations as experimental species. We found that water buffalo detectability performed better than for Javan deer.…”

Artificial Intelligence in Ecology: A Commentary on a Chatbot's Perspective

“…For example, the models designed based on AI can serve as a tool for predicting the impacts of climate change on species distribution, abundance and interactions (Peterson 2001, Barlow and O'Neill 2020, Stupariu et al 2022). Additionally, AI can be employed for modeling the impact of human activities on ecosystem functioning, such as land‐use change (Carrero et al 2014, Wagner and de Vries 2019), pollution (Ye et al 2020, Masood and Ahmad 2021, Subramaniam et al 2022), or conservation planning (Mosebo Fernandes et al 2020). For example, AI algorithms could be used to predict the effects of land‐use change on biodiversity or to identify areas that are most at risk of pollution (Mosebo Fernandes et al 2020, Ye et al 2020).…”

Cost of Ecosystem Service Value Due to Rohingya Refugee Influx in Bangladesh

“…ANNs are analogous to the organic nervous system in that they use numerous hidden layers to anticipate LULC. 33,34 The input, hidden, and output layers make up a neural network, a computational model made up of significant nodes. 35 The output layer of a previous node could become the input layer of the following node in this method, and the network's output changes depending on linking styles, weight values, and incentive functions.…”