An Outlook for Deep Learning in Ecosystem Science

Abstract

Rapid advances in hardware and software, accompanied by public- and private-sector investment, have led to a new generation of data-driven computational tools. Recently, there has been a particular focus on deep learning—a class of machine learning algorithms that uses deep neural networks to identify patterns in large and heterogeneous datasets. These developments have been accompanied by both hype and scepticism by ecologists and others. This review describes the context in which deep learning methods have emerged, the deep learning methods most relevant to ecosystem ecologists, and some of the problem domains they have been applied to. Deep learning methods have high predictive performance in a range of ecological contexts, leveraging the large data resources now available. Furthermore, deep learning tools offer ecosystem ecologists new ways to learn about ecosystem dynamics. In particular, recent advances in interpretable machine learning and in developing hybrid approaches combining deep learning and mechanistic models provide a bridge between pure prediction and causal explanation. We conclude by looking at the opportunities that deep learning tools offer ecosystem ecologists and assess the challenges in interpretability that deep learning applications pose.

Tree and mycorrhizal fungal diversity drive intraspecific and intraindividual trait variation in temperate forests: Evidence from a tree diversity experiment

“…Leaf traits prediction and consecutive statistical analyses were conducted in the R environment with R version 4.1.3 (R Core Team, 2021). As deep learning has recently emerged as a promising tool in trait‐based ecology (Perry et al., 2022; Vasseur et al., 2022), we used a convolutional neural network (CNN) approach for leaf trait prediction based on the spectral data. First, input spectra were augmented from 2501 to 12,906 features by using transformations based on a combination of standard normal variates and Savitzky–Golay derivatives (Figure S3; Passos & Mishra, 2021).…”

Twenty Years of Progress, Challenges, and Opportunities in Measuring and Understanding Soil Respiration

“…Another analysis found no significant differences or biases among Rs estimates derived from semi‐empirical, statistical or ML approaches (Hashimoto et al., 2023). We suggest that the field needs convincing demonstrations of what ML approaches add to more interpretable mechanistic or empirical approaches (Peters et al., 2014), careful consideration of the interpretability of ML models, and awareness of their potential biases (Perry et al., 2022). ML may be particularly useful for large‐scale assessments of the importance of parent material (Aka Sagliker et al., 2018; Dacal et al., 2022), geochemistry (Doetterl et al., 2015), and presence of soil carbonates (Gallagher & Breecker, 2020).…”

Reviewing the role of AI in environmental monitoring and conservation: A data-driven revolution for our planet

“…Assessing the vitality of habitats and identifying areas in need of restoration demands a level of insight that transcends human capacity alone. Artificial Intelligence (AI) offers an aerial lens that unravels the secrets of Earth's diverse ecosystems through its unparalleled image analysis capabilities, often propelled by sophisticated convolutional neural networks (CNNs) (Flück et al, 2022, Perry et al, 2022). AI's ability to process enormous datasets swiftly and accurately is a game-changer in habitat assessment, transcending the limitations of manual surveys and traditional monitoring methods.…”