Ecological Informatics

The automated classification of animals from photos is important in ecology and conservation biology for organizing and understanding the immense diversity of species, as well as facilitating effective conservation and management practices. It is equally important for disease surveillance systems, allowing prompt detection of anomalies in species distributions and boosting citizen-scientist platforms by making user-reported data more accurate and convenient. Image classification uses photos and can also rely on the geographical locations of animals to improve performance. While image classification models have difficulties in classifying low-quality images, unbalanced datasets, and with a small number of images, species distribution models have difficulty in classifying species that coexist in a given region. We propose here strategies for combining image classification models based on deep neural networks with species distribution models using genetic algorithms. The proposal is applied to a real-world dataset comprising fifteen classes of animals from the Brazilian fauna obtained from Fiocruzs citizen-scientist Wildlife Health Information System (SISS-Geo). The SISS-Geo photos portray the reality of animals in their environments, with varying quality, and pose numerous difficulties for classification. Experimental results demonstrate that the proposed integration consistently outperforms standalone models. While individual SDMs achieve Top-1 accuracies of 27.79% (MaxEnt) and 31.76% (Bioclim), and CNN-based classifiers reach 58.17% with ResNet50 and 64.13% with ResNet-152, the hybrid strategies yield substantial improvements. The genetic algorithm-based integration with a single global weight achieves up to 67.96% Top-1 accuracy, whereas the class-specific integration using fifteen parameters attains the best overall performance, reaching 69.03%.

Recent advances in automated technologies, such as passive acoustic monitoring, provide a powerful framework for surveying bird communities at broad spatial scales. Among the most widely used artificial intelligence tools for automated bird sound recognition is BirdNET, which can identify over 6,000 species worldwide. However, the effects of key user-defined settings, such as species filtering, remain poorly evaluated. Here, we assess how alternative species-filtering strategies influence BirdNET performance in describing bird communities worldwide. We analysed 5,047 minutes of sound recordings from 72 locations worldwide, comprising 1,192 bird species identified by expert ornithologists. We compared three common species-filtering approaches applied in BirdNET workflows to post-process its output: no filtering, spatial filtering (species present all-year at a given location), and spatio-temporal filtering (species present at a given location and week). The unfiltered approach maximised BirdNET species detection (recall) but suffered very low precision (had many misidentifications) and poor overall performance. In contrast, the other two filtering strategies greatly improved precision and overall performance, despite moderate reductions in recall. Among them, spatio-temporal filtering consistently achieved the best performance across most datasets and regions globally. Within this optimal filtering approach, we also evaluated the role of another parameter: occurrence probability thresholds. Intermediate values of this threshold (around 0.05) maximized BirdNET performance in community-level analyses. Our results demonstrate that species filtering is a key but often underappreciated component of BirdNET workflows. We hope our findings may guide future studies in selecting optimal species filters, while emphasising that filtering selection should be guided by study objectives and data context.

Artificial intelligence (AI)-based image analysis can efficiently estimate fish length, but differences in devices, imaging conditions, operators, and AI models limit comparability among surveys. We propose a standardization framework that estimates a bin-specific error matrix from paired reference measurements and AI-derived lengths and applies it to standardize (correct) AI-derived length-frequency distributions. The Richardson-Lucy expectation-maximization algorithm was used, with the number of iterations selected via cross-validation. Simulations based on empirical length-frequency data from 110 species showed that standardization reduced relative bias and distributional discrepancy; median relative-bias and root mean square error ratios were below 1, and the performance was more affected by the amount of paired data than by the number of cross-validation folds. In real data from 957 Japanese jack mackerel, standardized AI-derived distributions approached human-observer histograms, although discrepancies remained in the range of 160-230 mm. The proposed framework provides a practical approach for improving the comparability of image-derived length-frequency data using paired calibration data, without retraining the underlying AI model.

Convolutional Neural Networks (CNNs) have become increasingly prominent in biodiversity monitoring due to their strong performance in accurately detecting species from sound recordings, overcoming some limitations of traditional methods such as point-counts. Yet, their use in urban ecosystems remains limited, highlighting the need for frameworks that identify modelling strategies to optimize their performance in these complex soundscapes. Here, we evaluated how preprocessing and labelling strategies, detection thresholds, sample size, and architecture affect the performance of CNNs for bird identification in urban tropical ecosystems. We also assessed its potential by comparing CNN-derived biodiversity estimates with those from point-counts and acoustic indices. For this, we used one week of recordings collected along urbanization gradients in five Colombian Andes cities to developed 11 multiclass CNN models varying in spectral representation, labelling strategies, training data source and backbone architecture. The best-performing model, evaluated with F1-scores, combined Log-Mel spectrograms, multispecies labels, ecosystem-specific recordings, a probability threshold of 0.3 and a ConvNeXt backbone with its performance generally improving with sample size. Although CNNs and point counts detected partially distinct assemblages, CNN-derived species richness was comparable to that estimated from point-counts. In addition, the Normalized Difference Soundscape Index (NDSI) was positively associated with richness, suggesting its potential as a biodiversity proxy in tropical urban soundscapes. Overall, by identifying effective modelling designs and monitoring strategies, our study advances the development of robust biodiversity assessment frameworks in urbanized ecosystems in the Neotropics whilst also providing methodological guidance for future research and practical insights for wildlife monitoring and conservation.

Explainable machine learning (ML) methods are gaining increasing attention in environmental and ecological research for their ability to reveal relationships between environmental drivers and population dynamics. However, there remain questions on the reliability of these tools, especially given recent research shows that these explanations can be highly sensitive to model architecture. In ecology, it is typical to use a single ML model, and a comparative evaluation of sensitivity of explainability for different ML approaches is overlooked. In this paper, we develop a novel framework that quantifies explanation consistency between multiple ML model architectures. This framework provides a discrepancy measure for each model prediction, with high discrepancy indicating substantive explanation disagreement across models and low discrepancy indicating strong consensus in explanations across models. We then demonstrate that low explanation discrepancy aligns well with ground truth mechanism. Furthermore, high explanation discrepancy provide a mechanism to identify areas for model refinement and further investigation by domain experts. We do this by using a simulation study based on synthetic coral cover data that incorporate spatio-temporal variability driven by known disturbance effects. Our method provides a quantitative approach to assess the sensitivity of explainable ML in the absence of ground truth. As a result, this enhances the utility of ML approaches in conservation and ecological management. While we focus primarily on ecological modelling for coral reefs, our methods are generally applicable to other ecological and environmental modelling settings.

BackgroundUnderstanding ecosystem dynamics is essential for assessing ecosystem health, yet remains challenging due to complex biotic and abiotic interactions. Microbial communities are valuable indicators of environmental change, but the high dimensionality of microbiome data requires advanced analytical methods. This study explores the use of topic modeling (TM), an unsupervised machine learning approach initially designed for text analysis, to analyze microbiome data from the dynamic Warnow Estuary on the southern Baltic Sea coast. ResultsWe applied TM to estuarine microbiome data and compared its performance to traditional dimensionality reduction methods, Principal Component Analysis (PCA) and Principal Coordinate Analysis (PCoA). Quantitative results indicate that TM performs comparably to conventional approaches in preserving ecological and functional information, and in certain aspects even superior. In addition, we show qualitatively that NNMF, a TM method, captures latent patterns in the data providing an interpretable perspective on the microbiome. In this exploratory framework, NNMF suggested five distinct sub-communities within the estuary that appear to follow a seasonal succession influenced by freshwater inflow. These sub-communities were associated with specific ranges of salinity and temperature and showed distinct taxonomic profiles, with shared characteristics across the estuarine system. ConclusionsOur findings suggest that TM is a useful tool for exploring complex environmental microbiome datasets, offering a complementary perspective that can provide additional ecological insights. TMs ability to highlight coherent microbial community patterns indicates its promise for supporting environmental monitoring and informing targeted ecosystem management in dynamic habitats, though further studies are needed to fully assess its applicability.

Camera trap networks such as Snapshot Safari have generated millions of labelled wildlife images across Africa, enabling the training of deep learning models for automated species classification. However, deploying models trained in one African region to another remains poorly understood. To the best of our knowledge, this study presents the first systematic evaluation of geographic domain shift within the African continent for wildlife camera trap species classification, using the Machine Learning sub-field of Artificial Intelligence. We use three model architectures, each interacting with Snapshot Serengeti in a different way: BEiTV2is fine-tuned on Serengeti images as a supervised baseline; DINOv2 with FAISS uses Serengeti images as a retrieval index without any weight updates; and BioCLIP is a true zero-shot foundation model that receives no Serengeti training data at all. All three are then evaluated on two Southern African test sets, Snapshot Kgalagadi and Snapshot Kruger, as well as on locally collected wildlife photographs from Botswana. We conduct eight experiments covering in-domain baselines, cross-dataset transfer, data scaling, MegaDetector preprocessing, grayscale vs. colour image conditions, and per-species transfer analysis. This work provides the first empirical characterisation of intra-African domain shift across both supervised and zero-shot architectures, and offers practical guidance for conservation AI practitioners who need to deploy models across the diverse ecosystems of Southern Africa without collecting new labelled data.

This paper presents a six-stage methodological framework for Convolutional Neural Net-work (CNN)-based cetacean vocalization detection and classification in Passive Acoustic Monitoring (PAM), implemented as the open-source toolkit ai-pam-pipeline. The frame-work is generalizable across species and fully parameterised through a single configuration file, guaranteeing exact experimental reproducibility. Two experiments are reported. Experiment A examines the effect of FFT window length Nfft [isin] {256, 512, 1024} on binary Bottlenose dolphin (Tursiops truncatus) whistle detection using stratified 10-fold cross-validation on an in-domain dataset (Oltremare, 192 kHz) and a cross-domain benchmark (DCLDE 2022). In-domain performance is uniformly high (macro F1{approx} 0.98; Wilcoxon, all p > 0.05). Cross-domain results diverge substantially: Nfft = 256 is significantly superior (p = 0.006, rank-biserial r = 0.89). The mechanism is an upsampling amplification effect: coarser spectral bins produce wider, higher-contrast FM traces after bilinear resampling to fixed image dimensions. This superiority is threshold-invariant: precision equals 1.000 across all configurations and thresholds{theta} [isin] [0.1, 0.9], confirming that the advantage is not an artifact of threshold choice. These findings demonstrate that preprocessing choices -- often treated as secondary implementation details -- can significantly affect cross-domain generalisation. While Nfft serves here as a controlled case study, the framework is designed to enable systematic, reproducible evaluation of arbitrary preprocessing parameters within a unified experimental protocol. Experiment B demonstrates multiclass capability on five T. truncatus vocalization cate-gories (macro F1 = 0.843); inter-class confusion between click trains and burst-pulse sounds reflects biological signal overlap rather than classifier failure.

O_LIMonitoring and limiting the spread of invasive species on islands requires efficient detection and population estimation methods. However, elusive species can be difficult to monitor using traditional methods, making autonomous approaches such as camera trapping and acoustic monitoring increasingly valuable. C_LIO_LIOn the island of Okinawa, Japan, the small Indian mongoose ( Urva auropunctata) threatens many native species since its introduction in 1910. Listed among the worlds worst invasive species, effective monitoring of U. auropunctata in Okinawa is critical. The Okinawa Environmental Observation Network (OKEON) uses camera traps to detect U. auropunctata, but success depends on precise placement. Though OKEON also includes a high-resolution acoustic monitoring programme, no audio classification model currently exists for U. auropunctata. Developing such a model could improve substantially our capacity to detect and manage the species. C_LIO_LIUsing sparse U. auropunctata vocalisations collected from camera trap videos, we built a lightweight Convolutional Neural Network distilled from a more complex model for classifying contact calls and alarm calls of U. auropunctata. Our distilled model performed similarly to the full model at detecting vocalisations from training data, but was considerably faster. C_LIO_LIWe applied the distilled classifier to [~]486 hrs of audio collected over eight years from southern Okinawa, where we successfully detected U. auropunctata a handful of times in each year of recording. In spite of strong model performance on test data, our model did not transfer well to unseen data, perhaps owing to the rarity of U. auropunctata calls and consequent small training dataset size, limiting its utility for ecological monitoring. C_LIO_LIPractical implication. The use of sparse audio data from camera trap videos to train an acoustic classifier had limited utility to detect the rarely vocalising U. auropunctata from passive acoustic monitoring data. We provide several recommendations for enhancing classifier performance to provide robust actionable insights into the distribution and spread of U. auropunctata, and aid targeted conservation efforts for Okinawas threatened biodiversity. C_LI

Passive acoustic monitoring poses an immense potential to assess avian diversity in many habitats, including agricultural landscapes. At the same time, automated recorders generate large datasets which present a challenge for processing and effectively assessing biodiversity. Methods such as manual listening by experts, automated detection algorithms like BirdNET and calculating acoustic indices all present different trade-offs in assessment of biodiversity through passive acoustic monitoring. In the present study we recorded soundscapes in a low-intensity agricultural landscape in western Turkiye in all four seasons. Two expert ornithologists listened to a subset of these recordings identifying bird species from the recordings. We also ran the same sample of recordings on BirdNET to compare BirdNET detections with expert detections and calculated acoustic indices for each recording. The results showed that BirdNET detected more species than experts, although some may not be reliable detections. Two acoustic indices (bioacoustic index and acoustic complexity index) were correlated positively with number of species detected by experts and one (normalized difference soundscape index) with number of species detected by BirdNET but the correlations were modest. The results show that acoustic indices may have limited value in detecting biodiversity and automated detection algorithms may do a better job, although these may need to be trained with local data to improve detection and classification.

AimA comprehensive understanding of the spatial distribution of biodiversity is hindered by fragmented datasets, sampling biases, and inconsistent observation protocols. Here, we present a workflow that integrates disparate datasets to produce large scale maps of biodiversity metrics as a basis for management-relevant information tools. We use integrated species distribution modeling (iSDM) to account for sampling biases and disparate data collection techniques, taking advantage of the vast numbers of open datasets available in data aggregators like GBIF. LocationNorway (excluding Svalbard and Jan Mayen) TaxonVascular plants MethodsThe workflow consists of four main steps: data acquisition, data integration, integrated species distribution modelling (iSDM), and the production of derived outputs. Input data include structured surveys, opportunistic observations, and environmental covariates. These are standardised and integrated into a point-processed based iSDM framework to produce species richness maps, associated uncertainties, and sampling effort maps. The outputs are further processed to identify biodiversity hotspots or to summarise species-environment relationships. The workflow used vascular plant data from Norway, combining occurrence-only and presence-absence datasets with environmental covariates. Outputs were generated at a spatial resolution of 500 x 500 meters, balancing accuracy, computational feasibility and relevance for management decisions. High-performance computing resources were utilized for model fitting and predictions. A subset of available data was used to validate the species richness maps. ResultsWe produced detailed maps of species richness, uncertainties and sampling intensity across Norways heterogeneous landscape, incorporating 1218 species in our final results. The species richness patterns highlight patterns consistent with previous mapping efforts. Validation showed an increase in model accuracy when compared to models which did not use an iSDM framework. The workflow highlights limitations in the infrastructure of the currently openly accessible data, particularly the need for more structured presence-absence datasets and standardized metadata. Main conclusionsThis study underscores the potential of workflows that integrate disparate datasets for biodiversity modeling. To maximize accuracy and utility, future efforts should focus on improving data standardization, the publication and collection of more structured data, and fostering data-sharing collaborations. Advances in the workflow itself, including optimising modelling covariates and integrating more comprehensive spatio-temporal aspects, will also increase the relevance of the outputs. These advances will increase our ability to estimate species richness with a precision and accuracy that can reliably inform conservation and management decisions.

O_LIIdentifying individual animals from vocalizations is an emerging research area in computational bioacoustics. This non-invasive approach reduces reliance on physical capture and tagging for wildlife monitoring. Recent advances in this area leverage deep learning and bioacoustic foundation models adapted from species-level classifiers. However, these models typically rely on fixed-window inputs and may not fully capture temporal structure across extended or complex songs, which can contain information relevant to individual discrimination. C_LIO_LIHere, we evaluate whether modeling sequences of pretrained bioacoustic embeddings improves acoustic individual identification. We developed a framework that integrates transfer-learned spectrotemporal representations from BirdNET with a lightweight long short-term memory network. Unlike static baselines that use either the first embedding or an average over all embeddings, our approach processes time-ordered embedding sequences, allowing the classifier to use information distributed across multiple windows. C_LIO_LIWe evaluated the system using 87,865 vocalizations from 352 individuals across seven species. We used four publicly available vocalization datasets with individual-level labels, covering durations from 0.76 s (short calls) to 27.6 s (prolonged songs). Across five random seeds with stratified partitions, the framework achieved mean test accuracies between 93.9% and 98.3% and macro-F1 scores ranging from 93.2% to 98.3%, without data augmentation. The clearest gains over the strongest static baseline were observed for the great tit and the great spotted kiwi, reaching +1.3 and +2.8 percentage points, respectively. C_LIO_LIOur results indicate that the contribution of temporal sequence modeling depends on vocalization structure, rather than providing uniform evidence that chronological order drives performance. The benefit of recurrent aggregation was limited or absent for short calls but evident for long vocalizations with multiple informative windows. By combining bioacoustic foundation models with lightweight recurrent modeling, this approach provides a scalable, CPU-efficient tool for autonomous wildlife monitoring, particularly for species with extended or structurally complex vocalizations. C_LI

Freshwater monitoring programmes like the European Unions Water Framework Directive (WFD) provide a wealth of data on European lake status, including water quality and macrophytes (aquatic plants) as critical habitat features that support health of humans and wildlife. Easier WFD data access can enable external management and research to better safeguard human and natural freshwater use. We demonstrate a replicable workflow to easily download and process multi-year (2007-2024) observations of lake macrophytes (425 sites) and complementary water quality variables (202 sites) from Swedish WFD data. Then, we illustrate the value of improved data access to address ecological questions that drive conservation, investigating how spatial scales influence macrophyte richness and associated water quality relationships using a spatial random intercept model. Decomposing the spatial intercept links small scales (<10 km) to site-level gradients and large scales (>100 km) to biogeographical drivers. Stochastic and environmentally-structured processes coexisted at intermediate scales (10-100 km). Adding water quality rarely improved overall predictive performance of macrophyte diversity models but consistently influences the role of different spatial scales. Water quality variables showed consistent spatially structured variation at intermediate scales and unique spatial patterns in tandem, overlapping with large-scale biogeographical influences. Altogether, we show context-dependencies for spatial model interpretation and provide guidance in accounting for spatial confounding to improve inferential and predictive performance. Our workflow and results show a clear way forward for accessing high-quality macrophyte and water quality data sets and their utility for addressing ecological questions that guide macrophyte protection under the WFD. HighlightsO_LIyears Swedish of macrophyte and water quality monitoring data were extracted. C_LIO_LIrichness showed scale-specific patterns linked to geographic gradients. C_LIO_LIbest predictive models for richness had no water quality at all. C_LIO_LIoverlap in their spatial scales and must be carefully separated. C_LIO_LIpen access data and multiscale analysis can apply to many ecological questions. C_LI

BioBlitzes are widely used citizen science events that combine biodiversity monitoring, public participation, and environmental awareness through short and intensive observation campaigns. However, applying this model to marine environments presents additional challenges related to safety, access, weather dependency, specialised equipment, species identification, and sustained participation. This paper presents the BioMARathon model as a case study of how BioBlitz-inspired events can be adapted to marine citizen science contexts. The BioMARathon extends the conventional BioBlitz format into a longer, seasonal, and distributed engagement model designed specifically for marine and coastal environments. The paper describes the conceptual foundations of the model in the Janus Engagement Framework, which informed both the design of the BioMARathon and the adaptation of the MINKA citizen science observatory to better support participation, validation, feedback, and continuity over time. BioMARato Catalunya, launched in 2021, is presented as the founding implementation of the model and as the basis for later replication in Portugal. Between 2021 and 2025, BioMARato Catalunya showed continued growth in participation, observations, and taxonomic coverage, while also contributing to the detection of non-indigenous species, first regional records, and climate-related ecological impacts. Beyond biodiversity outcomes, the case suggests that extending participation across a season, distributing activities through local mobilising organisations, and combining expert validation with visible feedback mechanisms can support recurrent participation, retention, and community reactivation in marine citizen science. Rather than offering a formal causal evaluation, this article contributes practical lessons for the design of citizen science initiatives in challenging environments.

Invasive animal species are spreading rapidly across the globe, creating an urgent need for efficient early-detection and monitoring tools. Passive acoustic monitoring has become an established method in biodiversity research, but its application to invasive species monitoring has been less systematically explored. Here, we combine a systematic literature review with a field-based case study to evaluate the potential of passive acoustic monitoring for invasive animal detection. We identified 26 studies on acoustic monitoring of invasive animals, mainly addressing amphibians (11 studies), birds and fish (five each) with most studies from the USA and Australia. The use of acoustic monitoring of invasive species has increased during the past decade, with recent studies applying automated detection, machine learning, and large-scale monitoring frameworks. As a case study, we further tested the feasibility of low-cost acoustic monitoring of the invasive American bullfrog (Lithobates catesbeianus) in southwestern Germany, combined with automated identification using BirdNET. We successfully confirmed bullfrog presence in eight of the eleven monitored lakes, including sites close to a protected nature reserve. Our results highlight the growing potential of passive acoustic monitoring of invasive species under field conditions. In combination with automated species detection, manual validation, and emerging real-time monitoring devices, passive acoustic monitoring becomes an increasingly powerful tool for early intervention and scalable management of biological invasions.

Automated monitoring of insect pollinators in natural environments with insect camera traps and trained deep learning algorithms provides novel data for insect ecological studies. However, efficient and accurate image recognition analysis of the recorded images or videos is challenging, particularly for images containing small insects against complex backgrounds with diverse vegetation communities. Even when insects can be detected in images, identifying their taxonomy remains difficult, particularly in footage with low image resolution, light conditions, and distances from the plants, and in cases where insects appear blurry or only partially visible. In this work, we present InsectDCT, an AI-based pipeline for automated detection, hierarchical classification, and tracking of insects in footage of natural vegetation tested in different environments. The InsectDCT pipeline consists of three levels: insect Detection and localization, hierarchical taxonomic Classification, and spatio-temporal Tracking. In the first stage, insects are detected in time-lapse images or video recordings using the You Only Look Once (YOLO11) object detection architecture. Detection performance is improved using motion-enhanced images, which improve robustness in cluttered and 3 dimensional environments. The detector is trained on an extensive dataset that contains more than 60,000 images collected using camera traps deployed across a wide range of plant families and floral habitats. In the second stage, detected insects are classified using a hierarchical taxonomy-aware classification framework that covers 80 taxonomic groups. Classification is performed at multiple taxonomic levels, including order, family, and genus/species, allowing coarse and fine-grained ecological analyzes while accounting for varying levels of visual ambiguity. In the third stage, a multi-object tracking module is applied to high temporal-resolution image sequences and video data to associate detections of the same individual across time. InsectDCT code and all datasets are made publicly available. Author summaryInsects are declining worldwide, creating an urgent need for efficient methods to monitor their abundance, activity, and diversity. Traditional insect surveys often require extensive fieldwork and expert taxonomic identification, which limits the scale and frequency of monitoring. In this study, we developed InsectDCT, an artificial intelligence-based pipeline that automatically detects, classifies, and tracks insects in camera-trap recordings collected from natural and semi-natural environments. Our approach combines deep-learning methods for object detection, hierarchical taxonomic classification, and tracking of individual insect observations through time. Unlike many existing systems that are trained for a single habitat or plant species, we designed our framework using images collected across a wide range of flowering plants, camera systems, and insect groups. This makes the system more transferable to new ecological settings. The classifier can identify insects at multiple taxonomic levels and can return higher-level classifications when species-level identification is uncertain. We demonstrate that the pipeline can process large image datasets efficiently, including on low-power edge-computing devices such as Raspberry Pi systems. By providing both the software and the underlying datasets, we aim to support scalable, non-invasive insect monitoring and facilitate future ecological and conservation research.

Fire has acted as a major eco-evolutionary force since the evolutionary appearance of plants, shaping plant-traits, diversity, vegetation assembly, and ecosystem functioning. Its ecological role depends on long-term fire regimes. Anthropogenic land-use change and climate warming are disrupting these regimes, particularly in densely populated regions such as the Mediterranean Basin. In the Italian peninsula (Mediterranean region) fire activity peaks during the dry summer months and is projected to intensify under climate change scenarios. Recent methodological developments - based on emerging satellite data, ground-based observations combined with Random Forest (RF) habitat classification, and spectral indices such as the NDVI provide a robust framework for monitoring post-fire land-cover dynamics over time. In this study, we applied RF modelling to classify vegetation cover using a 2017-2024 satellite imagery time series of the Monte Pisano area (central Italy) to assess pre- and post-fire vegetation trajectories. Evergreen shrubs and trees exhibited rapid post-fire regrowth, whereas coniferous stands showed slower recovery rates. NDVI trends revealed an expected sharp decline immediately after the fire, followed by gradual recovery of broadleaf forests and shrubland communities. Moreover, our results indicated a progressive increase in the cover of native deciduous and evergreen species of high conservation value (listed under the Habitat Directive). The framework delivers spatially explicit insights into post-fire recovery, supporting targeted management, restoration under European Nature Restoration Regulation, and long-term monitoring in Mediterranean ecosystems. Incorporating fine-scale environmental variables may further improve classification accuracy and enhance assessments of vegetation resilience and ecosystem recovery following fire events. HighlightsO_LIRecurring fires strongly affect ecosystem structure and function in Mediterranean landscapes. C_LIO_LIIntegrating remote sensing with Random Forest models enables effective monitoring of post-fire vegetation recovery over time. C_LIO_LINDVI time series provide reliable proxies for tracking vegetation vigor and land-cover change. C_LIO_LIPost-fire recovery trajectories are shaped by fire severity, vegetation physiognomy, plant functional types, and soil conditions. C_LIO_LITargeted restoration and management interventions informed by spatial-temporal vegetation patterns are urgently needed. C_LIO_LIThe proposed framework aligns with objectives of the EU Nature Restoration Regulation for ecosystem and habitat recovery. C_LI

Invasive plants pose a major environmental problem, threatening biodiversity, altering ecosystem functions, and causing economic loss. Climate change is altering environmental conditions, potentially facilitating the spread of invasive plant species, posing challenges for ecosystem management and biodiversity conservation. Accurate predictions of invasive species distributions are therefore essential for effective monitoring and early intervention. Species distribution models (SDMs) have become an important tool for predicting species habitats, but many studies rely on traditional machine learning approaches, focus on single-species predictions and overlook uncertainty associated with future climate scenarios. This study aims to evaluate the performance of a deep learning-based SDM framework, Deepbiosphere, for predicting both native and invasive plant species distributions on a regional scale, the US state of Michigan, and to assess how climate scenario uncertainty influences spatial predictions of invasive species risk particularly on two focal invasive species. Results show that Deepbiosphere outcompeted other baseline models by on average of 10.98% with a mean AUC-ROC of 0.79 across 1553 vascular plant species. For two invasive species Rhamnus cathartica and Ailanthus altissima, Deepbiosphere respectively improved modeling accuracy by an average of 56.41% and 74.99%, suggesting its enhanced predictive capability for invasive species. Current predictions indicated that R. cathartica is already broadly suitable across much of Michigan, whereas A. altissima is currently more restricted to southern regions. Under future climate scenarios, both species were projected to expand northward, with a particularly strong expansion signal for A. altissima. Prediction uncertainty was spatially heterogeneous, where general circulation models (GCMs) were the dominant source of uncertainty across most of the state. By integrating citizen science, remote sensing, and deep learning, we produced high-resolution risk-uncertainty maps for key invasive species and highlighted the importance of explicitly mapping uncertainty to support more informed invasive species management under climate change.

Widespread arthropod declines pose risks to ecosystem functioning and agriculture. Assessing this decline or potential remediation implies the need for standardized and scalable population monitoring. Image-based methods, including camera traps and citizen science programs, are increasingly used, but the volume of data collected requires automated analysis. Robust arthropod detection is essential for individual counting or fine-grained classification, yet current datasets and algorithms do not address the vast morphological diversity across arthropod species and often overlook the variety of photographic contexts, such as differences in background, lighting, and image composition, in which arthropods are captured. To address this gap, we developed an arthropod detection dataset, covering all terrestrial families present in France with available validated images on the iNaturalist platform (749 families). To achieve this, we employed an iterative workflow in which a YOLOv11 model pre-annotated images -- using one representative species per family-- followed by manual correction and model retraining. Repeating this process progressively reduced annotation effort and improved model accuracy. The final outcome consists of a publicly available curated detection dataset and a robust arthropod detector for natural background scenes. The detector achieves an F1-score of 0.91, demonstrating strong performance despite substantial interspecific morphological variation and heterogeneity in photographic contexts. We further demonstrated the taxonomical universality of the model showing high F1-score and IoU averaged at the class (0.79, 0.85) and order level (0.82, 0.86) and also a good detection generalizability (F1-score>0.90, IoU>0.83) on species, genera and families never encountered by the model during training. Finally, we show how this model can be improved to generalize to new datasets using data augmentation, complementary training data or fine-tuning and increase detection of small objects. In particular, we report performance of the improved models on three use cases largely used in non lethal insect monitoring: (i) diurnal pollinator monitoring through citizen science or (ii) flower and nocturnal insects monitoring through smartphone time-lapse of a UV-illuminated white panel. These results mark an important step toward automated analysis of arthropod images in natural contexts, from both large-scale automated monitoring approaches or from citizen science monitoring programs.

Passive Acoustic Monitoring (PAM) has emerged as a transformative tool for biodiversity assessment in recent years. Despite widespread acceptance and application for conservation-related outcomes, the synergistic effects of hardware limitations, signal propagation, and environmental conditions on how far a signal can be reliably detected remain critically understudied. We quantified changes in signal detectability using Autonomous Recording Units (ARUs) in a tropical agroecosystem using playback experiments of standardised pure-tone (1-8 kHz) in fallow rice paddy fields. We deployed a four-ARU array and broadcast signals over a 50- 300 m distance gradient, and modelled operative detectability of signals using a binomial Generalised Linear Mixed-effects Model (GLMM). Our findings show that the detection space of an ARU is highly frequency-dependent and environmentally modulated. Detection probability for low-frequency signals (1 kHz) decreased rapidly (50% threshold at [~]100 m), whereas mid-range frequencies (4-6 kHz) occupied an acoustic window that remained reliably detectable up to 250 m. Higher relative humidity significantly enhanced overall detection, while increasing temperatures disproportionately reduced low-frequency detectability. The orientation of the ARU to the signal source was important as the detection probability declined from 81% for recorders facing the source (0{degrees}) to 14% for rear-facing units (180{degrees}). Our findings underscore the importance of determining the detection space before undertaking PAM. We propose a Decision Support Framework that provides a pathway for researchers to integrate focal taxa traits with technical constraints to determine detection space and optimise study designs when using PAM for monitoring biodiversity and assessing conservation action.