Robert M. Ewers’s research while affiliated with Imperial College London and other places

Modeling complex, nonlinear ecosystem processes across different timescales presents a significant challenge. We identify two key issues: selecting a representative timestep that captures interconnected processes across various timescales, and simulating these processes in an appropriate sequence. By synthesizing existing ecosystem frameworks, we find shared compromises between biological realism and computational performance. For the representative timestep, these include ‘selective elimination of timescales’, ‘biting the bullet’, ‘each in their own time’, and ‘capture the unseen’. For processing order, we identify hierarchical, logical, iterative, and random approaches. Similar challenges exist in other disciplines, and we show how transferring methods from multiple fields, along with smarter computing, can improve timescale integration. Overcoming these challenges requires innovative transdisciplinary solutions, and we outline directions for future research.

The impacts of degradation and deforestation on tropical forests are poorly understood, particularly at landscape scales. We present an extensive ecosystem analysis of the impacts of logging and conversion of tropical forest to oil palm from a large-scale study in Borneo, synthesizing responses from 82 variables categorized into four ecological levels spanning a broad suite of ecosystem properties: (i) structure and environment, (ii) species traits, (iii) biodiversity, and (iv) ecosystem functions. Responses were highly heterogeneous and often complex and nonlinear. Variables that were directly impacted by the physical process of timber extraction, such as soil structure, were sensitive to even moderate amounts of logging, whereas measures of biodiversity and ecosystem functioning were generally resilient to logging but more affected by conversion to oil palm plantation.

Small microphone arrays and sound‐source localisation algorithms are increasingly prevalent in the passive acoustic monitoring (PAM) of ecosystems. These technologies enable analysis of natural soundscapes' spatial features, yielding additional insights into biodiversity and ecosystem health. While many of these technologies have been evaluated in the field, there is a lack of controlled, repeatable methods to test them. We developed an ambisonic virtual sound environment (VSE) for simulating real natural soundscapes to evaluate spatial PAM technologies. We validated this novel approach using a PAM recorder with a six‐microphone array, from which we extracted a typical suite of ecoacoustic metrics, including acoustic indices and avian species predictions and localisations from the software BirdNET and HARKBird, respectively. We first verified whether the VSE could replicate natural soundscapes well enough to test PAM technologies by comparing these metrics between field and VSE‐based recordings. To pilot the VSE as an environment for testing PAM hardware, we assessed how orientation impacts the six‐microphone array's performance by using the same suite of metrics to compare VSE recordings made with the array at various pitch angles. Finally, we piloted the VSE as a test platform for PAM software by investigating how BirdNET and HARKBird perform on bird calls added to the VSE‐replicated soundscapes. While the VSE and field recordings had similarities in some metrics, including spectral composition and BirdNET predictions, ambisonics' perceptual bias and susceptibility to spatial aliasing limited the spatial analyses that could be undertaken. Our trials nonetheless revealed that device orientation impacts the performance of HARKBird and certain ecoacoustic indices, and that BirdNET and HARKBird perform best on louder, more directional bird calls. Our results demonstrate the potential for this approach, but highlight limitations to using an ambisonics‐based VSE. We thus provide guidelines for the use and refinement of such systems towards more standardised, controlled benchmarking of PAM technologies, empowering practitioners to make more informed decisions on using these vital tools.

The myriad interactions among individual plants, animals, microbes and their abiotic environment generate emergent phenomena that will determine the future of life on Earth. Here, we argue that holistic ecosystem models – incorporating key biological domains and feedbacks between biotic and abiotic processes and capable of predicting emergent phenomena – are required if we are to understand the functioning of complex, terrestrial ecosystems in a rapidly changing planet. We argue that holistic ecosystem models will provide a framework for integrating the many approaches used to study ecosystems, including biodiversity science, population and community ecology, soil science, biogeochemistry, hydrology and climate science. Holistic models will provide new insights into the nature and importance of feedbacks that cut across scales of space and time, and that connect ecosystem domains such as microbes with animals or above with below ground. They will allow us to critically examine the origins and maintenance of ecosystem stability, resilience and sustainability through the lens of systems theory, and provide a much-needed boost for conservation and the management of natural environments. We outline our approach to developing a holistic ecosystem model – the Virtual Ecosystem – and argue that while the construction of such complex models is obviously ambitious, it is both feasible and necessary.

The adoption of machine learning in Passive Acoustic Monitoring (PAM) has improved prediction accuracy for tasks like species-specific call detection and habitat quality estimation. However, these models often lack interpretability, and PAM generates vast amounts of non-informative data, as soundscapes are typically information sparse. Here, we developed ecologically interpretable methods that accurately predict land use from audio while filtering unwanted data. Audio from habitats in Southern India (evergreen forests, deciduous forests, scrublands, grasslands) was collected and categorised by land use (reference, disturbed, and agriculture). We used Gaussian Mixture Models (GMMs) on top of a Convolutional Neural Network (CNN)-based feature extractor to predict land use. Thresholding based on likelihood values from GMMs improved model accuracy by excluding uninformative data, enabling our method to outperform models such as Random Forests and Support Vector Machines. By analysing areas of acoustic feature space driving predictions, we identified “keystone” soundscape elements for each land use, including both biotic and anthropogenic sources. Our approach provides a novel method for ecologically meaningful interpretation and exploration of large acoustic datasets independent of specific feature extractors. Our study paves the way for soundscape monitoring to deliver robust and trustworthy habitat assessments on scales that would not otherwise be possible.

Halting deforestation is essential to address climate change and biodiversity loss. However, in highly forested, low-income countries like Liberia, “zero deforestation” commitments (ZDCs) adopted by companies may restrict agricultural expansion that has been promoted in national strategies to alleviate poverty. In such situations, examining contrasting perspectives among stakeholders is important to inform ZDCs’ implementation. Here, we applied Critical Systems Heuristics in 94 interviews to explore stakeholders’ perspectives on, and thereby develop a systematic understanding of, ZDCs in Liberia’s concession-based palm oil sector. We found that regulatory, institutional, and political factors that were needed to support commitments’ implementation were missing. Concessions had initially been allocated without communities’ consent being adequately obtained, and oil palm expansion had subsequently been stalled by zero deforestation. This produced a situation where communities that lost farmland to oil palm were reluctant to allow further expansion, while communities in forest areas were frustrated by a lack of promised oil palm expansion. Consequently, although limited oil palm expansion suggests ZDCs were effective after they were adopted, this was perceived to have come at the expense of anticipated improvements in community welfare, with community members in highly forested areas feeling deprived of development. We argue that neither the complete development of Liberia’s oil palm concessions nor limited development with zero deforestation will necessarily improve communities’ welfare without reforming the concession system to promote community-led, deforestation-free agricultural development. This requires public governance reforms, novel mechanisms for agricultural investment, and the localisation of international standards to facilitate zero deforestation in smallholder agriculture.

Insects are posited to be declining globally. This is particularly pertinent in tropical forests, which exhibit both the highest levels of biodiversity and the highest rates of biodiversity loss. However, for the hyper-diverse tropical insects there are scant data available to evidence declines. Understanding tropical insect diversity and its response to environmental change has therefore become a challenge, but it is estimated that 80% of tropical insect species remain undescribed 1 . Insect biodiversity predictions are based mostly on well-studied taxa and extrapolated to other groups, but no one knows whether resilience to environmental change varies between undescribed and described species. Here, we collected staphylinid beetles from unlogged and logged tropical forests in Borneo and investigated their responses to environmental change. Out of 252 morphospecies collected, 76% were undescribed. Undescribed species showed higher community turnover, reduced abundance and decreased probability of occurrence in logged forests. Thus the unknown components of tropical insect biodiversity are likely more impacted by human-induced environmental change. If these patterns are widespread, how accurate will assessments of insect declines in the tropics be?

Stem respiration constitutes a substantial proportion of autotrophic respiration in forested ecosystems, but its drivers across different spatial scales and land‐use gradients remain poorly understood. This study quantifies and examines the impact of logging disturbance on stem CO2 efflux (EA) in Malaysian Borneo. EA was quantified at tree‐ and stand‐level in nine 1‐ha plots over a logging gradient from heavily logged to old‐growth using the static chamber method. Tree‐level results showed higher EA per unit stem area in logged vs old‐growth plots (37.0 ± 1.1 vs 26.92 ± 1.14 g C m⁻² month⁻¹). However, at stand‐level, there was no difference in EA between logged and old‐growth plots (6.7 ± 1.1 vs 6.0 ± 0.7 Mg C ha⁻¹ yr⁻¹) due to greater stem surface area in old‐growth plots. Allocation to growth respiration and carbon use efficiency was significantly higher in logged plots. Variation in EA at both tree‐ and stand‐level was driven by tree size, growth and differences in investment strategies between the forest types. These results reflect different resource allocation strategies and priorities, with a priority for growth in response to increased light availability in logged plots, while old‐growth plots prioritise maintenance and cell structure.