Diversity and Distributions

ContextGenetic diversity is essential for the persistence and future adaptation of species. However, human-driven habitat fragmentation results in population isolation, often leading to rapid loss of genetic diversity and adaptive capacity. Genetic management of focal taxa may be overlooked in many threatened species conservation programs. The Endangered southeastern Australian mallee emu-wren Stipiturus mallee is a species that may benefit from genetic management. Its current range encompasses patchily distributed sub-populations, prone to bottlenecks and genetic drift. Thus, the reintroduction to areas from which the species has been locally extirpated requires careful selection of founders to maximise genetic diversity. AimsWe analyse reduced-representation genomic data from seven sampling areas across the global meta-population to design a translocation strategy that maximises heterozygosity and retention of mallee emu-wren allelic diversity. MethodsWe estimated genetic structure, genetic diversity within, and differentiation between subpopulations, thus testing previous inference based on 12 length-variable loci of low population differentiation with 10,840 genome-wide SNP loci. We also estimated effective population sizes to identify populations in need of genetic augmentation, Finally, we used metapop2 simulations to estimate the relative contributions of each population to global genetic diversity of the species and to estimate the source and number of founders that would maximise heterozygosity and allelic richness in a hypothetical newly established population. Key resultsWe found weak genetic structure across all sampling areas, supporting previous conclusions that the global mallee emu-wren population should be considered a single genetic unit for management purposes. Low but significant Weir and Cockerham pairwise FST among locations indicated differentiation between sampling areas, suggesting that contemporary gene flow is restricted. Effective population sizes for the two regions supporting the largest numbers of mallee emu-wrens were below the threshold associated with reduced adaptive potential. ConclusionsThe genetic health and adaptive potential of sampled mallee emu-wren sub-populations are at risk. Implications The global mallee emu-wren meta-population would likely benefit from genetic augmentation, including reciprocal gene flow between extant sub-populations. To maximise genetic diversity in newly established populations, managers should prioritise gene-pool mixing with founders sourced from all sampled areas.

Sky islands, mountain-top forests isolated by surrounding lowlands, offer unique opportunities to test how past and present landscapes shape species distributions. We examined the distribution of the Arizona gray squirrel (Sciurus arizonensis) across the Madrean Archipelago to test the constraint-based dynamic island biogeography (C-DIB) model, which posits that current occupancy in the sky islands reflects historical habitat size and connectivity. Using verified specimen records, we modeled climatically suitable habitats across four time periods: the Last Glacial Maximum (LGM), Mid-Holocene (MH), Present, and Future. For each mountain, we quantified suitable habitat area and estimated least-cost dispersal distances to assess both persistence and colonization potential. Our results suggest that species presence is best explained by LGM habitat metrics, which marginally outperformed models based on current conditions. Mountains that were large or well-connected during the LGM continue to support S. arizonensis, whereas historically isolated ranges remain unoccupied despite suitable contemporary habitat. These findings indicate a legacy of Pleistocene connectivity and reveal patterns of distributional disequilibrium. Furthermore, climatically suitable habitat for S. arizonensis has shifted both elevationally and geographically through time, reflecting long-term responses to climatic change. Together, these results emphasize the importance of protecting historically connected refugia, restoring riparian corridors that facilitate dispersal, and developing mountain range-specific management strategies that account for elevational shifts and potential downslope habitat recovery under future climate scenarios.

In cryptic or difficult-to-detect bird species, the monitoring schemes based on generalist detection methods may introduce bias into abundance estimates and population indices. This the case of the Coturnix coturnix (Common quail), a migratory Palearctic galliform, in which the use of passive detection methods within breeding birds monitoring schemes may not be efficient owing to its complex socio-sexual system and migratory behavior. For the first time, C.coturnix detectability was simultaneously compared using standard passive, generalist multispecies survey methods from the Pan-European Common Bird Monitoring Scheme (PECBMS) and a species-specific active survey employing female call playback. Surveys were conducted at 1,077 listening points within 107 transects over four breeding seasons (2022-2025) in open farmland landscapes dominated by cereal crops in Extremadura, south-western Spain. Detection counts differed substantially between methods: active surveys increased expected counts by 72% (95% CI: 59-85%) compared to passive surveys. The increase in C.coturnix detections elicited by playback showed a non-linear, density-dependent pattern, being highest at low passive abundances per listening point (maximum at 3-4 individuals) and stabilizing at intermediate abundances. This indicates that call playback is particularly effective at detecting individuals that would otherwise remain undetected. Our findings suggest that passive, multispecies surveys may underestimate C.coturnix abundance, especially in low-density populations. Integrating species-specific active methods into monitoring programs can improve detectability, generate more reliable population indices, and support evidence-based conservation and management strategies for this elusive species. LAY SUMMARYO_LIBird monitoring schemes guide conservation decisions across Europe, but generalist schemes based on passive methods may miss species that are hard to detect such as C.coturnix Common quail, a migratory farmland bird that hides in dense crops. In practice, only males spontaneously calling can be detected, hence passive methods could lead to underestimates of its abundance and even false absences in low-density areas. C_LIO_LIWe compared standard passive surveys with surveys that added a recorded female call (playback) to stimulate male responses. Across 1,077 listening points monitored over four breeding seasons in southwestern Spain, playback increased the number of birds detected by 72% compared with passive methods alone. C_LIO_LIThe improvement was strongest where C.coturnix numbers were low, showing that many individuals remain undetected without playback. Incorporating simple, species-specific methods into monitoring programs can produce more reliable population estimates and strengthen conservation and management decisions for this elusive species. C_LI

O_LIChanges in habitat structure across species distributions may contribute to the generation and maintenance of range limits, but few studies have evaluated this by directly measuring habitat availability across relevant spatial scales. C_LIO_LIHere, we test the predictions that coarse-scale and patch-level habitat availability decline towards and beyond the northern range limit of Pacific coastal dune endemic Camissoniopsis cheiranthifolia. We used aerial imagery and geographic information system (GIS) tools to measure the coarse-scale availability of coastal dune habitat in California and Oregon. The availability of finer-scale habitat patches specifically suitable for C. cheiranthifolia was measured in a 2-generation field survey of > 4,200 5m x 5m plots randomly distributed across 1100 km of coastal dune habitat transcending the species northern range limit. At each plot, we estimated the proportion of area that contained suitable habitat as well as recorded occupancy by C. cheiranthifolia. As an alternative approach to visually estimating habitat suitability, we recorded plant community composition at each plot to predict beyond-range habitat suitability using a random forest model. C_LIO_LIContrary to our predictions, we found that coastal dune habitat, measured coarsely from aerial imagery, was more abundant and continuous towards and beyond the northern range limit. At the fine scale, however, the proportion of plots with suitable habitat (patch suitability) and the proportion of habitat within plots that was suitable (patch size) declined across the range limit. Moreover, patches were more isolated from one another and, in one survey year, less temporally stable towards and beyond the range limit. Finally, occupancy by C. cheiranthifolia was less likely in smaller, more isolated, and temporally unstable patches, providing mechanistic insight to the previously observed decline in occupancy towards the range limit. C_LIO_LISynthesis: Taken together, our results suggest that fine-scale habitat patch configuration changes in ways that likely impede patch colonization, thereby reducing occupancy and limiting the species northern distribution. Thus, consideration of geographic variation in patch and landscape structure, rather than only coarse-scale habitat availability, may be essential for understanding the processes that limit species ranges. C_LI

Natural history collections underpin our understanding of species distributions, yet some historical records remain embedded in modern avifaunal checklists despite limited documentation and no independent verification. One such case concerns the Dusky Parrot Pionus fuscus in Colombia: although reported from specimens collected by Melbourne A. Carriker Jr. in 1942 in the Serrania de Perija, the species has not been observed in the country for nearly eight decades yet continues to be included in national checklists and conservation assessments. We reassessed the validity of this record by applying a multi-evidence framework integrating historic archival reconstruction, specimen-based morphological comparisons, climatic niche analyses, biogeographic limit assessment and contemporary survey-effort data. Historical documentation and morphological evidence based on high-resolution specimen images and associated curatorial records demonstrate that the Carriker specimens correspond to Pionus chalcopterus, not P. fuscus. Climatic niche analyses reveal minimal environmental overlap between P. chalcopterus and P. fuscus, and place the Perija locality within the climatic niche of P. chalcopterus, while regional biogeography and extensive modern birdwatching coverage provide no support for the occurrence of P. fuscus in Perija. Together, these concordant lines of evidence demonstrate that P. fuscus does not occur in Colombia. Our findings support its removal from national bird lists and conservation assessments and highlight how integrated, multi-evidence reassessments of historical records strengthen ornithological baselines, improve biogeographic inference and ensure that conservation priorities rest on verifiable evidence.

AimUnderstanding whether invasive species retain or shift their ecological niches has traditionally relied on scalar overlap metrics that quantify the magnitude of niche change, but not its structure. Here, we test whether biological invasions involve a reorganisation of the environmental axes along which native and invasive ranges are differentiated, and whether the dominant axes of this reorganisation are consistently associated with invasion pathway type (intercontinental vs. within-continent). LocationGlobal (North America, Europe, Africa, Asia, Australasia). Time periodContemporary (environmental variables representing long-term averages, 1980-2021). Major taxa studiedFreshwater crayfish (Decapoda: Astacidea): Procambarus clarkii, Faxonius limosus, Pacifastacus leniusculus, Faxonius virilis, Faxonius rusticus. MethodsWe analysed native and invasive occurrences for five globally important crayfish invaders using [~]400 hydrologically resolved environmental variables from the Global Crayfish Database of Geospatial Traits. Classification models were used to quantify environmental differentiation between native and invasive ranges, and feature contributions were aggregated by environmental domain (climate, topography, soil, land cover). Patterns were evaluated across intercontinental and within-continent invasion pathways and assessed for robustness using cross-validation, permutation tests, sample-size sensitivity, and comparisons with classical niche overlap metrics. ResultsNative and invasive occurrences were consistently distinguishable across all species (accuracy 96.5-99.9%). A pathway-dependent pattern emerged: intercontinental invaders were primarily differentiated along climatic dimensions (58-76% of model importance), whereas within-continent invaders showed a more balanced contribution of climatic and topographic variables ([~]42% each), including strong signals from river network position. This contrast was stable across cross-validation folds (SD < 1.6%), and supported by permutation tests (P = 0.001). Classical niche overlap metrics (Schoeners D = 0.30-0.62) did not capture this qualitative distinction. Main conclusionsBiological invasions involve not only changes in niche position but a reorganisation of the environmental axes that distinguish species distributions. Our results suggest that the dominant axes of this reorganisation differ systematically with invasion pathway, reflecting whether species encounter novel climatic regimes or primarily shift within existing climatic space along topographic and network-position gradients. By resolving which environmental dimensions underpin native-invasive differentiation, this approach provides a complementary perspective to scalar overlap metrics and a basis for more mechanistic interpretations of invasion processes.

The scale at which diversity is observed shapes the patterns we find. While spatial scale is known to influence biodiversity patterns, the effects of temporal scale, namely the average duration of sampling (known as temporal span), have been mostly overlooked. Here, we investigate how temporal span affects species richness patterns, their environmental drivers, and species richness hotspots. We used species richness data from several large bird datasets from Czechia, with over 7000 observations, a spatial grain ranging from 0.03 to 100 km2, and a temporal span ranging from 1 to 36 years (1985-2017). Using Random Forests, we modelled species richness as a response to temporal span, while also including area, geographic location, time, and environmental and land-cover predictors. We found that the temporal span is consistently among the most important predictors of bird species richness. Moreover, temporal span interacts with key environmental conditions, particularly precipitation and water bodies, modulating their effects on species richness and revealing processes that differ from those traditionally attributed solely to spatial grain. We also found that using different time spans can shift the predicted locations of biodiversity hotspots. Our results provide empirical evidence that temporal span should be included in studies about biodiversity and conservation planning, given the urgent challenges arising from ongoing biodiversity change and the complexity of its drivers.

Most wildlife species currently inhabit areas transformed by human activity, a hallmark of the Anthropocene. Habitat alterations caused by the creation of roads and other human-made infrastructures shape the spatial distribution of wildlife species and their interaction with the environment. While some sensitive species disappear, more tolerant ones thrive near humans. Therefore, a streamlined tool to quantify the tolerance of different species to human pressures is useful to conservation, in particular to identify more vulnerable species. Here, we present ecoTolerance, an open-source R package that calculates two complementary, continuous metrics: the Road Tolerance Index (RTI), derived from the distance of each occurrence record to the nearest road, and the Human-Footprint Tolerance Index (HFTI), based on the global human-footprint raster. This package is based on a workflow that includes separate functions and arguments to automate data cleaning, spatial thinning, distance extraction, species-level summarization and map generation. As an applied example of its use and application, we processed 3782 records of five species: Copaifera langsdorffii (1407 observations), Bradypus variegatus (724), Sylvilagus brasiliensis (274), Boana faber (1226), and Boana boans (151), revealing RTI values that ranged from 0.183 to 0.654 and HFTI values from 0.111 to 0.392. the values of the two indices varied according to the incidence of road kill, as well as the habitat preference of the particular species. These examples demonstrate that ecoTolerance facilitates a rapid and streamlined assessment of species tolerance and vulnerability, providing valuable insights with potential to inform conservation actions.

AimThe Global Biodiversity Information Facility (GBIF) is the most prominent source of species occurrence data for modeling climate niches, but exhibits strong unevenness in its data coverage across different geographic regions. The impact of this spatial bias on the reliability of GBIF-based plant climate niches in Europe remains unexplored. This study aims to address this gap, and to investigate whether the targeted integration of additional atlas data can reduce the potential impact of the spatial bias. LocationEurope. Time period1950s - 2024 Major taxa studiedEuropean grassland plant species. MethodsWe analyzed the climate niches of a large number of grassland species, with diverse distribution patterns across Europe, based on a) GBIF and b) on an enriched version of GBIF with national atlas data from Eastern European countries (GBIF+), where data coverage is currently low in GBIF. We followed best practices in niche characterization, particularly by performing environmental subsampling. The accuracy in climate niche properties was determined by comparing niches based on GBIF and GBIF+ data with niches based on a careful implementation of expert range maps as reference dataset. We focused on niche optimum position and niche similarity. Additionally, we investigated how biogeographical indicators can predict variability in climate niche accuracy. ResultsMost species exhibited reliable climate niche characterization using GBIF data, especially for widely distributed species. Yet, reliability decreased with continentality; that is, when species were primarily distributed in Eastern Europe. Integrating additional data did not significantly reduce this bias in niche characterization. Main conclusionsDespite the spatial bias in its records, GBIF can be used to reliably characterize the climate niches of many species in Europe if uneven sampling effort is accounted for. The laborious integration of additional data to address spatial bias does not yield the desired increase in niche reliability.

AimAnalysis of species distributions often rests on the assumption of environmental equilibrium. That is, the distribution of a species (as documented by observation records) captures the full range of environmental conditions under which that species can maintain viable populations. Despite the centrality of this assumption to a variety of biogeographic questions, it is rarely empirically tested. This is particularly critical for recently introduced invasive species that are characterized by rapid expansion in their introduced range, often coupled with a niche shift relative to their native distribution. Defining equilibrium under these dynamic conditions is difficult. We developed the concept of environmental stasis as a more tractable proxy for equilibrium. In the context of species invasions, we define stasis as a prolonged period without an increase in the environmental conditions occupied by a species. LocationNorth America Time Period1614 to 2020. Major Taxa StudiedInvasive plants MethodsWe applied the metric of climatic stasis to a suite of 258 invasive plant species in North America. We categorized their invasion trajectories into three classes (linear, two- and three-phase) based on theoretical expectations and then assessed how many had demonstrated environmental (climatic) stasis over a period of at least thirty years. ResultsMore than 80% of the species were best fit by two- or three-phase models, indicating a declining rate of expansion. Climatic stasis was only documented for 44% of the species. In contrast, 85% of the species were in climatic stasis in their native ranges. The time to reach stasis ranged from 30 to 145 years (mean 90), and species at stasis in their invaded range occupied 97% of the climatic space they occupied in their native range. Main ConclusionsThis assessment provides valuable insight into the unrealized threat posed by the majority of invasive plants that have not yet reached stasis, as well as identifying which species can be most appropriately evaluated by methods that depend on the equilibrium assumption. Our work also demonstrates the useful perspective provided by the environmental stasis concept, which enables empirical quantification of one of the key aspects of equilibrium.

Understanding relatedness in sharks is challenging due to uncertainty in distributions, low population densities and difficulties in sampling across life stages. In Fiji, bull sharks (Carcharhinus leucas), with an effective population size estimate of [~]258, aggregate at the Shark Reef Marine Reserve (SRMR), but gravid females disperse at the end of the year to give birth in adjacent rivers. Questions remain regarding reproductive connectivity, female returns across years, and kinship structure. Using population genomics on 296 bull sharks across age classes (neonates, young-of-the-year, juveniles, and adults) collected over a decade at the SRMR and in three adjacent rivers, we assessed familial connections. Direct genetic links, including first- and second-degree relationships, connected SRMR adults with young age classes in the Navua and Rewa rivers, providing evidence of reproductive connectivity. Within rivers, genetic similarities across cohorts revealed reproductive philopatry. Remarkably, several individuals sampled years apart were assigned to the same sire-dam pairs, indicating repeated pairings across breeding seasons. However, the few related links detected between the SRMR and the rivers may reflect incomplete sampling. Altogether, bull shark reproduction in Fiji seems influenced by reproductive philopatry and repeated pairings, suggesting added complexity in their reproductive behaviour.

Community science data are increasingly recognized as important resources for biodiversity research, in part because of the spatial and temporal resolution that they afford. While these data are useful for applications such as describing occurrence patterns, tracking movement of migratory animals, and recording phenological events, they can also be probed for "second-order" purposes, such as documenting species interactions. Here, I present a dataset of more than 35,000 annotated interactions between monarch butterfly (Danaus plexippus) larvae and their associated host plants from the community science platform iNaturalist. I document more than 70 unique species of milkweed hosts (Apocynaceae: Asclepiadoideae) used by monarch larvae, including a number of previously undocumented interactions. Monarchs show strong seasonal turnover in the species of host plants used across the migratory cycle, highlighting the importance of early season hosts like Asclepias viridis and A. asperula in eastern North America and A. californica and A. cordifolia in the west. I also demonstrate that non-native horticultural milkweed species have increased the spatial extent of monarch breeding during winter (November - February) by more than 60%, a pattern previously suggested from observational data but not formally quantified until now. To my knowledge, this represents the largest analysis to date of species interactions using unstructured community science data and highlights the value of platforms like iNaturalist for conducting fundamental research in ecology and conservation.

Understanding how climate shapes intraspecific genetic turnover is critical for predicting biodiversity responses to global change, yet such analyses remain limited for systems where natural adaptation and human-mediated dispersal jointly structure diversity. Here, we investigate the spatio-temporal dynamics of genetic composition in the western honey bee (Apis mellifera) across Anatolia and Thrace, a major historical refugium harboring five subspecies. Using a dataset of 672 individuals genotyped at 30 microsatellite loci, we characterize population structure and model ancestry compositions as a function of environmental and geographic variables. We integrate Gradient Forests and Generalized Dissimilarity Modelling to identify key climatic drivers of intra-specific turnover and project future changes under multiple CMIP6 climate scenarios. We detect five major ancestral groups with widespread admixture structured by both spatial processes and environmental gradients. While geographic distance explains a substantial proportion of variation, climatic variables account for a large fraction of ancestry turnover. Spatial projections reveal distinct ecological regions corresponding to subspecies distributions, with high turnover zones aligned with major geographic and ecological barriers. Climate projections indicate substantial restructuring of ancestry compositions over the 21st century. Most ancestral groups show declines in persistence and resilience, whereas lineages associated with warmer and drier conditions expand under future scenarios. Regions of high uniqueness and refugia contract, while areas experiencing rapid turnover and novel ancestry compositions increase. Existing Genetic Conservation Areas provide incomplete representation of diversity and are projected to lose effectiveness under future climates. Our results demonstrate that climate change is likely to disrupt spatial genetic structure, promote admixture, and threaten persistence and resilience of honey bee populations. By modeling ancestry composition as a multidimensional proxy for genetic variation, for the first time to our knowledge, this study provides a scalable framework for forecasting intraspecific biodiversity dynamics and informing conservation and management strategies under global change.

Animal translocations are becoming increasingly popular as a tool for conservationists. Demographic factors can be crucial determinants dictating translocation viability in the short term. Translocated populations pass through artificial bottlenecks and can suffer from founder effects. Reduction in genetic variation relative to their source populations is likely, limiting their adaptive potential. Founder events can increase frequencies of deleterious alleles due to elevated rates of inbreeding and inbreeding depression. Here, we describe the effects of human-driven, serial population translocations on the genetic diversity of critically endangered Anegada iguanas (Cyclura pinguis) in the British Virgin Islands. Though founding populations were extremely small (N=8, N=4), the census sizes of translocated iguana populations increased dramatically over the first twenty years. This implies that these translocations were successful from a demographic perspective despite the small number of animals used, indicating a genetic paradox. To quantify genetic signatures in these bottlenecked populations, blood samples were collected from the source population and two translocated populations and genotyped at 21 microsatellite loci. We found that allele frequencies in translocated populations differed significantly from those of the source, with the translocated populations having less genetic diversity. However, common methods for estimating presence of genetic bottlenecks were non-significant. Estimates of internal relatedness by age class suggest that inbreeding depression may be elevated after translocation, likely reflecting the small initial population sizes associated with these translocation events. Anecdotally, our work shows that translocations may result in subtle genetic erosion that has long-term population viability impacts, even when census size indicates success.

Invasive alien species (IAS) expansions are increasingly impacting the biodiversity and economy of Europe. To more effectively allocate the limited resources available for their management, it is pertinent to accelerate detection of IAS spread and distribution. One largely untapped secondary data source showing much potential lies in the automated tracking of internet activity such as IAS search intensity or mentions across different internet platforms. In this study, we tested if internet activity increases systematically when IAS expand into new EU countries utilizing the combined data of 88 invasive species from various internet platforms. In total, 14 internet platforms were screened and evaluated based on their database accessibility, mined data quality and utility for systematic IAS expansion tracking. We found that the procedure to obtain researcher access to minimal data required for IAS tracking (i.e., information about location, time and place) varies widely across platforms, and is particularly difficult without incurring significant costs for many of the larger ones (X, Google and Tiktok). From the explored species, more charismatic species (i.e., mammals) overall gained more online traction than more cryptic ones (i.e., plants), though online activity of the first proved a worse representation of real-world occurrence patterns. Moreover, while the final five selected internet platforms showed increased activity surrounding the year of invasion in many of the explored invasion scenarios (particularly Wikipedia and Facebook), inconsistencies between species groups, trends per platform and the large variability in data quality currently still hampers systematic integration of such data into existing databases. We conclude that combining IAS activity data from various internet platforms shows potential to accelerate IAS expansion detection across the EU (especially for fish, crustaceans, reptiles, birds and plants). However, incorporation in automated early warning systems is currently hampered by variable data quality, limited researcher access to online data and the few open, accurate and generalizable species classification algorithms with API access.

Understanding how species that are threatened with extinction utilise human-modified landscapes is essential for evidence-based conservation. We investigated multi-scale habitat selection by the Forest Owlet (Athene blewitti), an Endangered species, endemic to central India with fewer than 1000 mature individuals, in the Dangs district of Gujarat, the westernmost extent of its range. Using a hierarchical Bayesian occupancy framework, we examined how forest cover and three agricultural land-use types (dry agriculture with trees, dry agriculture without trees, and intensive agriculture) affected occupancy across three nested spatial scales: regional (81 km2), landscape (4 km2), and territory (0.25 km2). At the regional scale, the forest x agriculture interaction term was significantly negative ({beta} = -6.82, 95% CI: -9.87 to -1.59), indicating that owlets favour agroforestry-dominated regions over forest-dominated landscapes. Conversely, at the landscape scale, a significant positive interaction ({beta} = 1.36, 95% CI: 0.41-2.50) revealed synergistic benefits from forest-agriculture mosaics. Agriculture type strongly influenced landscape-scale occupancy: dry agriculture with trees showed positive effects ({beta} = 1.17, 95% CI: 0.43-2.02), whereas dry agriculture without trees had significant negative effects ({beta} = -1.19, 95% CI: -2.28 to -0.29). These findings demonstrate that Forest Owlets are not forest-obligate specialists but occupy complex agroforestry mosaics, requiring multi-scale conservation strategies. We propose that the traditional Malki agroforestry system, which incentives tree retention on farmland, offers conditional compatibility with Forest Owlet conservation, provided that mature cavity-bearing trees and small forest patches are explicitly protected.

Accurate mapping of tree species from satellite data remains challenging in heterogeneous mountain forests due to environmental gradients, mixed stands, limited availability of high-purity training labels, and strong illumination-angle effects. Recent geospatial foundation models offer a new approach by learning generic, cloud-agnostic, information-rich representations from large multi-sensor archives suitable for a range of downstream tasks, but their ecological utility for species-level mapping remains incompletely understood. Here, we evaluate two geospatial foundation-model embeddings, AlphaEarth and Tessera, for tree species classification in the Trentino region of northern Italy, using parcel-level forest inventories as reference data (18 species and species groups). We compare their performance against conventional Sentinel-1+2 satellite composites across a series of controlled experiments examining classification accuracy, label efficiency, classifier complexity, robustness to label impurity, and temporal transferability. Foundation-model embeddings consistently outperform composite-based multispectral satellite baselines (weighted F1 = 0.83 vs. 0.80; macro F1 = 0.55 vs. 0.50), reaching near-asymptotic accuracy with as few as 5% of available training parcels and preserving ecologically meaningful structure aligned with functional and taxonomic groupings. However, realising this advantage requires a nonlinear classifier: a compact neural network provides better results than classic machine learning (i.e. Random Forest) and performs as well as deeper neural networks, while a linear classifier on foundation-model embeddings underperforms a neural network on conventional composites. Ancillary environmental covariates offer no additional classification benefit when added to embedding-based models. Classification accuracy remains robust to moderate levels of label impurity, allowing mixed parcels to be retained in the training dataset without substantial penalties, while training with parcel-level species proportions as soft labels achieves higher peak performance (macro F1 = 0.586 for Tessera, 0.589 for AlphaEarth) and lower Proportion L1 error than hard labels without requiring purity filtering, maximising the value of the full range of input data. However, temporal transfer across years reveals performance degradation, with weighted F1 declining by 9% for Tessera and 15% for AlphaEarth, and disproportionate losses for rare species. Overall, our results show that geospatial foundation models shift a primary bottleneck in species mapping from feature engineering toward the availability, quality, and temporal alignment of ecological reference data, while opening new opportunities for scalable biodiversity monitoring and the analysis of ecological change.

O_LISpecies occurrence data are fundamental to understanding, predicting, and conserving global biodiversity. However, biodiversity datasets remain affected by substantial data-quality issues, particularly erroneous or imprecise geographic coordinates. Most available tools for identifying problematic records rely primarily on automated spatial or metadata-based checks and rarely integrate expert-curated species range information, which can reveal introductions or geographic errors that often escape standard validation procedures. C_LIO_LIHere, we introduce RuHere, an R package designed to manage species occurrence data, flag potential errors, and support the iterative exploration of problematic records. RuHere streamlines the data-cleaning process by integrating six main steps: (1) obtaining species occurrence records; (2) merging datasets and standardizing spatial information; (3) flagging records based on metadata; (4) flagging records using expert-derived distribution data; (5) visualizing, investigating, and summarizing flagged issues in the final datasets; and (6) exploring and reducing sampling bias. C_LIO_LIWe demonstrate the applicability of RuHere using occurrence data for a plant species (Araucaria angustifolia) and an animal species (Cyanocorax caeruleus). Nearly 75% of records were flagged as potentially problematic, including records identified exclusively by functions relying on specialist range information. C_LIO_LIThe main strengths of RuHere lie in its integrated and computationally efficient workflow, its tools for exploring and evaluating flagged records, and its ability to incorporate expert-derived distribution data to identify occurrences outside a species known natural range. By combining metadata-based checks, coordinate validation, and specialist knowledge, RuHere provides a robust and reproducible framework for improving the quality of species occurrence datasets. C_LI

O_LITropical dry forests are among the most threatened ecosystems globally, yet the consequences of livestock overgrazing for ant communities remain poorly documented, particularly in the Tumbesian biodiversity hotspot of southwestern Ecuador, where uncontrolled goat grazing constitutes the dominant disturbance agent. C_LIO_LIWe sampled ant communities (Formicidae) across a goat-grazing disturbance gradient in Zapotillo (Loja Province, Ecuador), establishing three disturbance levels (Dense, Semi-dense, and Open Forest) with nine 60 x 60 m plots per level (n = 27) and 486 pitfall traps. Community responses were assessed using abundance-based and presence-absence analyses of morphospecies richness, Hill-number diversity, community composition, beta diversity decomposition, and functional guild structure; vegetation structure was characterized using satellite-derived NDVI. C_LIO_LIWe recorded 47,459 individuals belonging to 22 morphospecies in six subfamilies. Morphospecies richness declined with disturbance (Dense: 19, Semi-dense: 15, Open: 12), with four specialist genera exclusive to Dense Forest. Beta diversity decomposition revealed a shift from turnover-dominated dissimilarity at moderate disturbance to nestedness-dominated dissimilarity at high disturbance, indicating progressive habitat filtering as the dominant community-restructuring process. C_LIO_LICommunity composition differed among disturbance levels (PERMANOVA: F = 4.49, R{superscript 2} = 0.272, p = 0.001) and was correlated with NDVI (r{superscript 2} = 0.341, p = 0.013). Cryptic/soil and Leaf-cutter guilds were nearly eliminated from Open forest while the Opportunist guild expanded markedly, indicating that functional homogenization precedes detectable taxonomic impoverishment. C_LIO_LIOvergrazing drives directional ant diversity loss and biotic homogenization at both taxonomic and functional levels in the Tumbesian dry forest, underscoring the conservation value of intact Dense forest. C_LI

Species interactions can determine species population sizes, geographic ranges, evolutionary trajectories, and responses to environmental change. Yet, despite their importance to many fundamental and applied questions, information on species interactions is often lacking due to constraints in data collection. Billions of text comments that have been submitted by millions of citizen scientists around the world have the potential to fill these gaps. Comments can be used to identify biotic interactions using advanced large language models (LLMs), providing a novel source of interaction data that is unusually high in spatiotemporal coverage, breadth, and resolution. This novel approach opens new avenues to evaluate species interactions on a broader scale, and to characterize and conserve biodiversity under pressing global change. Highlights- Although species interactions are central to biodiversity dynamics, progress in resolving their fundamental properties and forecasting their shifts under global change has been hindered by persistent data limitations - Citizen science platforms contain billions of observer text comments that often contain valuable information about species interactions, but the unstructured format of the information and the size of the datasets make these comments difficult to use - Large language models (LLMs) provide an unparalleled opportunity to collect and analyze species interactions from such comments - Using two case studies, we present a workflow that leverages LLMs to automatically collect species interaction observations from citizen science comments in multiple languages around the world - Such a novel source of data greatly expands the data coverage and resolution of species interactions across space and time and can help to answer both long-standing ecological questions and new, pressing questions about ecological responses to global change