Under changing climatic conditions, widespread generalists and warm-adapted species may benefit, while specialists and cold-adapted species may decline, with interspecific competition possibly adding to unfavourable conditions. A fascinating example are the cryptic butterfly species Aricia agestis and Airica artaxerxes. Here we investigate the climatic niches of both species to assess competition effects and model probable distributions under future climates. As the multivoltine, generalist A. agestis likely outcompetes the univoltine, more specialized A. artaxerxes, we assessed possible negative effects on the latter by combining both species’ projected occurrences in a new framework.

Species’ climatic niches indicated that A. artaxerxes was evading direct competition by occurring around the niche edges of A. agestis, which may underline the former being outcompeted by the latter at their climatic optimum. While we projected northward shifts in both species, A. agestis showed an increase in range size but A. artaxerxes’ range decreased. Including a possible competition effect only slightly heightened the negative prospects for the latter. Contrary to previous assumptions, the hybridization area may decrease in the future, diminishing the possibility of genetic exchange between the cryptic species.

Our analysis provides a glimpse into the possible future of a complex example of species interactions.

Audiovisual methods are increasingly being used in monitoring programmes to observe and describe the loss of biodiversity. There are a variety of methods, including camera traps and acoustic recorders among others, which are being used to record both vertebrates (mammals, birds, amphibians) and invertebrates (insects) in aquatic, terrestrial and marine habitats.

With this technology becoming readily available and easily affordable, it is increaingly being used and thus generates very large amounts of data, which must be archived and processed. Although this task is very time consuming, ideally this would be done almost in real time together with a stream-lined data analysis if possible, in order to develop the basis for instant recommended actions for conservation, society and policy makers.

The AI4WildLIVE project tries to achieve this and aims to bring together data repository, processing infrastructure, citizen science and data analysis multimodal monitoring data in one portal for the first time. We present a first showcase of the portal structure & utility using our own camera trap data collected in Bolivia and South Africa highlighting the different steps of the data archiving, data pre-processing (image classification by citizen scientists & AI) and subsequent geo analysis.

Phytophagous insects are closely associated with morphological and

metabolic traits of tree species. For the pedunculate oak (Quercus

robur) alone, at least 700 species of phytophages are known to affect

the health and development of the leaves and, in some cases, can lead

to severe leaf loss and tree death.

However, due to the lack of standardized methods for detecting

components of tree fitness, the complex spatial and temporal patterns

in leaf characteristics and phytophage interactions remain largely

unknown.

To address this knowledge gap, we propose a novel approach utilizing

imaging techniques and AI-based object detection to automate the

assessment of herbivory, leaf-herbivore interactions, and leaf health.

Our automated pipeline aims to overcome current limitations in

assessing leaf damage, particularly in leaves with non-entire margins.

By applying AI-based object detection techniques, we can accurately

identify characteristic damages and quantify herbivore and pathogen

damage in a standardized manner. Through comparisons with manipulated

test data and human assessed estimates, we will evaluate the accuracy

of our methods. The proposed approach will offer a wide range of

applications and facilitate a spatially and temporally fine scaled

monitoring of plants and associated biota.

In the climatically harsh areas of Antarctica and the subantarctic areas of southern South America crustose, saxicolous, lecideoid lichens establish a dominant vegetation-form. They form a symbiosis of fungal (mycobiont) and algal (photobiont) partners. It is now widely assumed that the photobiont is primarily determining the climatic niche for the whole lichen organism. Mycobiont species vary in their capability to associate with either one or more photobionts. Therefore, they differ in their ability to extend their climatic niche by a more generalist selection of photobionts.

In this study, we present a circum Antarctic dataset consisting of 662 samples of lecideoid lichens focussing on their associated dominant myco- and photobionts covering a habitat range from the 41°S to the 86°S latitude. We investigated how symbiotic associations between myco-and photobionts are mediated by climatic factors and in which extent the associations are changing in their environmental space along time. Furthermore, we aimed to model past symbiont switches and accompanied niche change based on the phylogenetic relationship of the fungal symbiont. With this approach conclusions can be drawn to what extent symbiotic networks change in their interconnectedness and how the degree of generalization may be an advantageous adaptation to a rapidly changing climate.

In the field of ecology, scientific literature is a key data source that is vital for generating insights and testing theories across ecological systems. For example, large ecological databases like TRY, FishBase, Compadre or TetraDENSITY are commonly used. The extraction of new data from literature is tedious and time consuming as it often requires researchers to manually process thousands of articles. This hampers our ability to get the most out of existing data and makes it difficult to keep databases up to date. Methods available for text mining are rapidly evolving because of developments in machine learning-based language models. Here we assess the use of a natural language processing framework to parse computable trait data from scientific literature. To optimize and validate the framework, we will use a case study on the automated collection of mammalian home-range values based on the recently published HomeRange database. We will compare the home-range values from the automated data collection with manually collected values to evaluate the accuracy of the framework. Our ultimate goal is to have an automated workflow that can be applied to various ecological fields and will result in robust datasets that can be used to advance of ecological understanding.

We are currently facing a global-scale biodiversity crisis, with accelerating rates of extinction, habitat degradation, and ecosystem disruption. Notwithstanding, the recent past of many organisms is poorly known, especially when the available information is fragmented and scattered in old literature. This source of information is often neglected in research as extracting the information is rather time-consuming, e.g. due to language barriers, library or copyright restrictions and the sheer number of existing documents. Technological advances, such as optical character recognition, natural language processing and artificial intelligence can facilitate the acquisition of information hidden in text sources. Here, we present the Specialised Information Service for Biodiversity Research (BIOfid; https://www.biofid.de) which was launched to increase availability, accessibility and usability of contemporary and historic biodiversity information, with a focus on Central Europe. BIOfid helps to find information from historical biodiversity literature through innovative text mining tools based on large language models. BIOfid`s semantic search technology is backed up by taxonomical, morphological, and ecological ontologies. Our tool is currently evaluated for how well we can acquire information on biotic interactions and functional traits. It improves data extraction, knowledge discovery, and ecological monitoring for a better understanding of the past, present, and future of biodiversity.

Species' ability to cope with climatic instability varies greatly, influenced by factors like dispersal, physiological adaptations, and phylogenetic conservatism. Here, we investigate how burrowing behaviour – a key component of species’ endurance strategies and ecosystem functioning – shaped the distributional expansion and diversification of mammalian lineages. Analysing 4,407 terrestrial mammal species and novel trait data on 3,096 species, we reveal distinct responses to climatic factors between burrowing and non-burrowing species. Burrowing lineages are disproportionately species-rich at lower temperatures and productivity levels. Both range size and species richness steeply increase with climate seasonality in burrowing species, as opposed to non-burrowing species. Constituting 47% of all terrestrial vertebrates, the proportion of burrowing species increases latitudinally, and particularly regions with greater Pleistocene temperature changes are almost exclusively composed of burrowing species. Trait conservatism, higher diversification rates, and Eocene peak diversification provide the evolutionary context to these contemporary gradients, underscoring the role of burrowing for mammalian radiations into cold-temperate climates. Our study highlights the potential of readily available behavioural information in improving forecasts of species' responses to climatic changes, and showcases divergences of broad importance for targeted conservation efforts.

Southern Europe, characterized by the convergence of several biogeographical regions and diverse climates, hosts diverse vegetation types. Understanding their typology and distribution is crucial for biodiversity conservation and ecosystem management.

Utilizing data from the European Vegetation Archive (27,511 plots, 7,389 taxa), we applied the TWINSPAN algorithm to classify vegetation into broad types: alpine and subalpine grasslands, xeric and meso-xeric grasslands, rocky grasslands and low scrub, subalpine low scrub of Eastern Mediterranean, Anatolian scrub, Western-Mediterranean garrigue, Eastern-Mediterranean garrigue, and herbaceous Mediterranean grasslands.

Our analysis revealed significant correlations between vegetation composition, climatic parameters, and environmental factors. Vegetation distribution primarily followed a gradient from cooler, mesic conditions at higher elevations to drier, xeric environments at lower elevations. Distinct patterns were observed between Western and Eastern Mediterranean regions. Furthermore, shifts in plant life forms were observed, with hemicryptophytes favouring mesic conditions at higher altitudes, chamaephytes preferring warmer and drier climates, and therophytes more common in areas with greater continentality.

Our research provides valuable insights for revising habitat and vegetation classification systems and developing effective conservation and management strategies. These findings contribute to the broader understanding of biodiversity patterns and their underlying drivers in Southern Europe.

Understanding global patterns of species richness requires reliable mapping of the composition, steepness, and extent of ecological gradients. Existing paradigms for land-cover classification of remote sensing imagery tend to favor sharp boundaries between biogeographical units, whereas gradients would better reflect reality. Consequently, transition zones are rarely considered when mapping habitats, ecosystems, or biomes. However, research suggests that these transition zones, also known as ecotones, could be centers of biodiversity and drivers of speciation. Therefore, reliable mapping and monitoring of ecotones across large spatial scales is a prerequisite to developing effective conservation strategies for these biodiversity hotspots, especially in the face of climate change-induced range shifts. Spectral communities, based on the spectral species concept, could offer a reliable method for mapping ecotones. With the recent availability of medium-resolution (30 m) hyperspectral satellite imagery (EnMAP), differentiating between even the most visually-similar ecosystems, such as laurel forest and fayal-brezal on the Canary Islands, becomes feasible. Using a global dataset of island and continental ecotones, we test the reliability of the spectral communities algorithm in mapping ecotones by comparing spectral gradients to both climatic (e.g. temperature and precipitation), topographic (e.g. elevation, aspect, slope), and biotic (e.g. species and traits) data.

Plant phenology, examining key life cycle events such as budburst, flowering, fruiting, and senescence, provides valuable insights into environmental responses. However, documenting these events across vast spatial and temporal scales presents ongoing challenges.

While global phenological networks collect individual-scale data, a decline in observers threatens these datasets. Conversely, plant identification apps collect abundant occurrence data with time stamps and images across large spatial scales. To derive nuanced phenological stages from these records efficient annotation methods are essential which ultimately allow harnessing this data for phenological monitoring.

This study aims to automate phenological stage annotation of images obtained from opportunistic plant observations. We extract deep features from annotated plant images using the Flora Incognita neural network to train a Support Vector Machine (SVM). Subsequently, the model will annotate a large corpus of images according to various phenological stages.

Furthermore, we used the day of year of each image to employ a predictive model to estimate the timing of several phenological stages based on geolocation and elevation. This enables the automatic linkage of observation data to respective phenological stages, facilitating phenological monitoring. Subsequently, we compare the opportunistic data with systematic records over multiple years and phenological stages.

Biodiversity loss is a main driver of global change and “insect decline” one if its current hot topics. Despite numerous studies, a clear identification of the causes and their relative importance, as well as regional variation and the effect of habitat protection measures, is hard to get. While there are many studies on very large or very small areas facing these topics, few address actionable spatial units and long-term trends for various insect species.

My doctoral project will narrow this gap. Utilizing a long-term dataset (> 30 years) from a governmental insect monitoring program in the state of Bavaria, Germany, I will employ spatially explicit occupancy modeling to (1) capture regionally diverse trends on insect species level. Further, I will (2) explain these insect trends with habitat variables such as remote sensing-based land coverage, climate data, and management tools like protected areas.

By understanding the influence of these habitat variables, (3) future insect trends can be modelled, using existing land cover and climate models. The insights gained from this research not only advance the use of spatial occupancy models for biodiversity research methodologically but also provide crucial insights for practical conservation efforts at the regional level.

Orchids constitute nearly 10% of Madagascar’s plant diversity but the majority of species may be threatened with extinction. Commonly, they would be often associated with forests rather than with open ecosystems, which happen to be significantly under-protected. Here, we conducted a review of all orchid species in Madagascar and assigned their ecosystem types as closed (forest) or open (grassland, marsh, savanna, Tapia woodland, ericaceous scrub, and rocky outcrop) using herbarium specimens, online biodiversity databases and taxonomic literature. We find that even up to 30% of species can be found in open ecosystems, and 15% of species are found there exclusively. We also assess the bioregionalisation patterns across the island which point to the spatial differences in the distribution of closed vs open ecosystem specialists. Finally, using data on species flowering times, we provide a direction for sampling efforts inclusive of the species phenology and past spatial bias. The results highlight Madagascar’s open-ecosystem orchids as a distinct but underappreciated source of biodiversity and demonstrate that conservation efforts should expand beyond the Madagascan forests.

Bats (Chiroptera) are one of the most species-rich orders of mammals and harbour a highly diverse virome. Some bat-associated viruses have spilled over to humans, livestock, and other mammals, causing zoonotic diseases. However, a comprehensive overview on the diversity and biogeography of bat-associated viruses is missing, but highly relevant for Planetary Health. Here, we provide a biogeographic overview of global hotspots and biogeographic patterns of bat-associated virus richness in relation to biomes and continents. Publicly available data on bat-associated viruses was combined with data on global bat distribution to identify hotspots and biogeographic patterns of bat-associated virus richness. These patterns were linked to biomes and continents to detect underlying drivers.  In general, bat-associated virus richness increases from the poles to the equator, following bat species α-diversity. Southeast Asia, tropical Africa and Central America form distinct hotspots of bat-associated virus richness. Biogeographic patterns for virus families differed. Biomes and continents as potential underlying drivers only explain a low amount of the observed variance. However, these hotspots need to be considered with caution due to biases and knowledge gaps especially in tropical regions. Future research efforts are needed to sharpen our understanding of the bat virome and to identify areas at risk.

Insects and their key ecosystem functions are in decline, but the distribution of global insect diversity and its threats from climate change remain little understood. Butterflies have the potential to serve as global insect model system given their ecological importance and extensive data. Here we show that tropical and sub-tropical mountain regions emerge as centers of butterfly richness, rarity, and phylogenetic diversity and that the geographically restricted temperature conditions underpinning these hotspots make them highly exposed to projected global warming. Integrating comprehensive phylogenetic and range data for 12,119 butterfly species, we find that while mountains only cover 38% of the Earth’s surface at our grain, butterfly hotspots are 3.5 times more likely to be located in mountains than in lowlands. Hotspots of butterfly diversity face more severe impacts of global warming than non-hotspots, with as much as 64% projected niche loss in tropical realms until the year 2070. However, only 14%-54% of the realm-level richness and rarity hotspots (top 5%) in butterflies overlap with current vertebrate-based conservation priorities. Our study identifies critical conservation needs for butterflies and illustrates how the consideration of at least select global insect systems is key for gaging biodiversity loss in a rapidly warming world.

The availability of trait databases enabled global analyses on the distribution and variation of trait values and the biotic and abiotic factors influencing them. In contrast to morphological trait values and their intraspecific variability, the distribution and responses of chemical traits to biotic and abiotic factors have been less well explored on global scales considering multiple species. However, the few global analyses on the distribution and variation in plant specialised metabolites show clear trends. Previous analyses showed that the plants’ metabolism responds to various environmental parameters including temperature, precipitation, CO2 levels and nitrogen levels. For floral scents, clear effects of environmental parameters have been found. Thus, analyses on and the assessment of global trends in the chemical phenotype of plants revealed general patterns and rules advancing the understanding of the ecology of plant metabolomes. However, chemodiversity, which is the richness, evenness, and disparity within plant metabolites, has not been considered in similar studies and basic information on the factors affecting chemodiversity is still lacking. Based on a well-documented literature survey, we aim to conduct a global analysis based on published data on the global distribution of plant chemodiversity and the biotic and abiotic factors influencing it.

Gridded biodiversity data (e.g. country-level or regional atlases) play a prominent role in macroecology, particularly in the study of patterns of species occupancy, geographic ranges, biodiversity, and their drivers and temporal dynamics. However, managing, exploring, and analyzing data in grids comes with problems.

We have reviewed the problems with gridded data, and the existing solutions. We identify problems of sampling (e.g. varying sampling method and effort in space and time, imperfect detection), geometry (e.g. varying grid cell area and shape, positional errors), taxonomy (e.g. misidentification, taxonomic splits and merges), and scale (e.g. unspecified or varying spatial and temporal grain).

The first type of solutions happens prior to gridding of the data; this includes selection of geographic projection, grain, and grid cell shape. The second type of solutions involves manipulation of the gridded data; examples are aggregation of cells to coarser grains or cell filtering. Third type of solutions is done during data analysis, usually by representing the problem by a covariate in statistical models. We hope to provide guidance particularly to early career biogeographers and macroecolgists who may otherwise struggle to make sense of the various solutions scattered through the literature.

Secondary woodiness that originates on oceanic islands, insular woodiness, is a remarkable island syndrome that evolved independently throughout the angiosperm tree of life. One of several hypotheses explaining this phenomenon states that burial of vegetation through tephra from volcanic activities on the island is a selecting driver towards insular woodiness. However, there is currently no conclusive evidence for this hypothesis.

Here, we test the role of volcanic tephra deposits as a driver for insular woodiness by analysing vegetation patters and plant functional traits related to three craters that erupted in historic times (75 a, 378 a, and 5 ka) on the island of La Palma.

Vegetation on lapilli fields is dominated by secondary woody and archipelago-endemic woody species, with younger tephra fields highly dominated by secondary woody species and older fields more dominated by ancestral woody species. Furthermore, the functional trait space seems constraint in lapilli vegetation in comparison to pine forest in the same elevation range. These observed patterns indicate potential abiotic filtering of tephra burial towards (insular) woody species and their traits. Our results thus provide the first empirical evidence for the “volcanic selection” hypothesis of insular woodiness.

Understanding biodiversity change across spatial scales is crucial in the Anthropocene. While existing analyses primarily focus on species range size (i.e. occupancy), spatial structure remains understudied. Spatial structure, indicating the degree of clustering or dispersion in distributions, can be quantified through spatial autocorrelation. This metric, influenced by endogenous (dispersal, demographic) and exogenous (environmental, climatic) processes operating at different spatial scales, can help identify species threatened by population isolation and separation.

Here, we explore spatial autocorrelation patterns of bird distributions across the Northern Hemisphere over time and spatial scales, using data from temporally replicated Breeding Bird Atlases. We computed multiple measures of spatial autocorrelation, including global Moran’s I and Join-counts, for both species’ distributions and richness at different grain sizes and analysed their trends over time and spatial scales. We observed a consistent decrease in autocorrelation with increasing spatial scale across all study areas and periods. Notably, there was no discernible temporal trend in the average spatial autocorrelation of species distributions or richness. Our analyses revealed that temporal trends in spatial autocorrelation were independent of occupancy, highlighting the distinctiveness of range size and structure and the importance of studying both.

Intraspecific genetic variation between plant populations is common and can reflect local adaptation. Understanding the phenotypic and genetic variation along a geographical gradient is crucial for predicting a species’ adaptive ability and assessing its vulnerability to climate change.

In this ongoing project, we study genetic variation in Hordeum murinum, a common wild annual with different ploidy levels. We combine common garden experiments with genetic tools to identify adaptive variation and its underlying genetic regions.

In 2023 in collaboration with researchers from Europe, we scored H. murinum in wild across Europe to define its ecological niche and collected seeds. H. murinum grew in a wide range of environments and plants were larger in wetter and colder areas. In a preliminary common garden, we found significant phenotypic differentiation between populations in root traits and an increase in the root to shoot ratio in populations originating from warmer sites. The preliminary results from this year’s common garden experiment indicate that plants from warmer and drier areas flowered earlier than plants from colder regions.

Quartzite sites in southern Africa can be considered as terrestrial habitat islands due to their distinct environmental conditions, well-defined boundaries, and unique vegetation hosting a high number of endemic species. These quartzite islands are scattered across various regions of southern Africa, spanning different biomes such as Succulent Karoo, Nama Karoo, and Fynbos, across a wide bioclimatic spectrum. Thus, they present a promising subject for investigating key aspects of island biogeography and ecological dynamics on different spatial scales. This newly funded project aims to investigate and compare six spatially distinct island systems, regarded as archipelagos, located along a broad bioclimatic gradient with different precipitation regimes. For each archipelago, an extensive species survey will be conducted alongside the establishment of a functional trait database encompassing the majority of perennial plants inhabiting quartz islands. The study will be conducted at different spatial scales. At the island scale we will elucidate the influence of island biogeographic drivers on taxonomic diversity, functional diversity and endemism. Additionally, we will test the effect of regional climate on the expression of traits at the archipelago scale. Finally, we will perform a macroecological analysis of the quartz island flora in southern Africa.

More than 40% of the world’s plant species are rare and threatened by extinction. While destruction of the natural world has accelerated dramatically over the past decades, our knowledge of global plant biodiversity patterns has not increased at similar rates. As our understanding of plant biodiversity remains incomplete, all global studies of plant diversity patterns are based upon subset of species. Digitally available plant data suffers, however, from strong biases which severely distort our view of global diversity patterns.

In our study, we aim to better estimate plant diversity patterns from incomplete data by determining the number of species needed to accurately represent plant diversity distribution patterns at a global scale using the World Checklist of Vascular Plants as a baseline. We use the information available to determine how well datasets such as the IUCN Red List of Threatened Species and openly available plant occurrence data (GBIF) represent different dimensions of plant biodiversity such as species, taxonomic, phylogenetic, and functional diversity. The results have important implications for macroecology and conservation.

Stream restoration projects aim to improve water retention in the landscape and increase the ecological value of water bodies around the globe. In Europe, most restoration projects take place on small streams, yet the ecological benefits of such projects are poorly documented. This study focuses on the effects of small stream restoration on terrestrial biota, in particular plants, insects, and birds, as they provide important ecosystem services and have high conservation value. We compare 50 restored small stream sections in Hesse, Germany, with adjacent non-restored sections, using plant and bird surveys and malaise traps with DNA metabarcoding for insect identification. In preliminary field work, we found a significant increase in wetland plants and an increase in bird diversity and abundance in restored sites. With this ongoing study, we now aim to determine the value of different restoration measures across trophic levels. Findings will contribute to increase the ecological value of future restoration projects.

The world is undergoing significant environmental transformations, impacting biodiversity and ecosystem functions. Yet, obtaining temporal biodiversity data is challenging due to cost and monitoring limitations. To address this, we aimed to develop a machine-learning model that infers temporal trends of species’ occupancy without requiring temporally replicated data.

The approach involved analyzing static snapshots of species spatial distributions and their covariates to infer temporal change in occupancy for temperate breeding birds in the Czech Republic, Japan, New York state, and the whole Europe (N = 841 species). Although static patterns are only partially able to predict temporal change, we found that the predictive strength of static patterns is lower when predicting past change as compared to future change. Interestingly, geometric constraints of the study area and the species distribution explain a high proportion of the predicted change in occupancy.

Our results suggests that the importance of processes that drive change in species spatial distributions through time does not stay constant, and that characteristics of the study area, such as shape and size may determine the amount of change that can be predicted from data.

cito is an easy-to-use R package for deep learning that allows users to build neural networks using the familiar formula syntax of many other R packages. Yet, cito allows flexible modification of network architecture and hyperparameters, allowing users to test and apply most deep learning techniques without having to focus on coding. In addition, cito includes many user-friendly functions for model interpretation based on explicable AI and additional features such as confidence intervals (p-values). We explain how to extract effects (similar to linear effects from regression models) and variable importance (similar to ANOVA) from the fitted DNN and how to interpret these effects. Finally, we will show how cito can be used to fit CNNs for image classification and regression.

VAT is a web-based GIS system that offers an interactive and intuitive way to visualize, analyze, and transform spatio-temporal data in NFDI4Biodiversity to serve the biodiversity community. In this workshop, we present a novel VAT workflow for joining spatio-temporal thematic environmental raster data with spatio-temporal species occurrence data from GBIF. The workflow shows the unique features of VAT exploring spatio-temporal data and deriving advanced FAIR data products. Through its seamless connectivity to Jupyter Notebooks, VAT also takes advantage of Jupyter's powerful AI offerings. In addition, VAT, with its docker container-based implementation, is one of the fully integrated critical services in the RDC, the cloud-based research infrastructure of NFDI4Biodiversity. This enables VAT to manage large amounts of data in the underlying object storage ARUNA and to offer outputs of its workflows as new FAIR data products. Overall, this showcases the synergetic value of the RDC with all its interconnected components like VAT and ARUNA.

The “mountain-geobiodiversity hypothesis” explores the interaction of topography, climate, and biology in the evolution of mountain biodiversity. We tested this hypothesis in the Himalayas and the Hengduan Mountains on a group of local caddisfly species. We investigated one caddisfly species pair from each mountain respectively, each pair containing one species inhabiting high elevation and one inhabiting low elevation. We incorporated genomic and ecological evidence to reveal population structure, demographic history, and potential habitat range at present and during the last glacial maximum. The results indicated that the high-elevation species showed strong local differentiation, while the low-elevation species were shaped by hydro-morphology. Caddisfly species in the Himalayas generally exhibited an East-West oriented dispersal, while species from the Hengduan Mountains showed greater connectivity on the North-South orientation. Results of demographic history and species distribution modelling demonstrated that a cold climate leads to an increase in potential habitats, thus causing population expansion. Moreover, most of the divergence and admixture events aligned with the climatic cycles from the middle Pleistocene until the present, suggesting a species-pump effect. Our study demonstrates that mountain topography and climate fluctuations interact and influence the diversification of caddisflies differently in the Himalayas and Hengduan Mountains, thus supporting the hypothesis.

Groundwater-dependent vegetation (GDV) is an important global biodiversity hotspot facing threats from climate and land-use change, requiring large-scale mapping efforts. A novel medium-resolution mapping approach is presented that aims to identify GDV in the Mediterranean biome using global plant community data (sPlot) and remote sensing.

Approximately 31,000 vegetation plots have been extracted from ‘sPlot - The Global Vegetation Database’. An expert system is being built to evaluate plots likely to represent GDV based on functional species groups. Lists of species were compiled regarding phreatophyte coverage, moisture values or climatic parameters. These lists will be used to extract species compositions indicative of GDV and to classify vegetation plots accordingly.

The remote sensing approach implements criteria aimed at 1) vitality during dry periods, 2) seasonal and 3) interannual variation in vitality, 4) high topographic potential for water accumulation, and 5) topography (elevation, slope) derived from Sentinel2 and SRTM. The reclassified vegetation plots will be used to train machine learning algorithms for the pixel-wise classification of GDV in the Mediterranean biome.

Species compositions indicative of GDV would support identification in the field. Furthermore, detailed maps of GDV will ensure sustainable groundwater management and thus protect GDV as local biodiversity hotspots.

It is observed an alarming trend of declines on biodiversity globally due to anthropogenic activities. However, those activities impact ecosystems and biomes differently and as a consequence, some species could be more threatened than others. It is known that small-range species are disproportionately more threatened, due to their expected lower tolerance to environmental changes and lower capacity to move to new suitable locations. In this study, we aim to estimate the impact of the loss of small-range species on biodiversity using desiccation-tolerant plants as model species. We performed sequential extinction simulations, from the small-range to broad-range species, to estimate decreases in phylogenetic diversity in all world biomes. The contribution of small-range species varied across taxonomic groups and biomes. Overall, the loss of small-range Pteridophytes was similar to the loss of broad-range species. Conversely, the impact of the loss of small-range Angiosperms was critical in three world biomes, such as tropical and subtropical moist broadleaf forests. Our findings suggest prioritizing small-range species from these biomes for conservation. Nevertheless, we also recognize the importance of assessing and monitoring small-range species from other biomes, since the loss of phylogenetically more redundant species reduces the options for the maintenance of ecosystem function and services.

Water deficit is one of the main drivers of plant mortality and is projected to be more critical owing to climate change. Because desiccation-tolerant vascular plants (DT plants) can cope with water deficit, a paradigm emerges associating those species to locations characterized by water deficit conditions. However, this paradigm is not supported by earlier studies, hampering our understanding of the species–environment relationships and the vulnerability of DT plants to climate change. Here, we tested this paradigm and provide an evaluation of the vulnerability of DT plants to climate change. We estimated the diversity and distribution of DT plants along water deficit gradients and assessed species vulnerability to climate change from a climatic perspective and over broad phylogenetic and macroecological scales. We found that the diversity and distribution of DT plants were neither associated with, nor restricted to, locations characterized by water deficits. Our findings suggest that the desiccation events DT plants undergo are rather promoted by topo-edaphic conditions than by climate, reinforcing the need for studies that investigate processes that operates across different scales. Moreover, species with narrow niche breadth might be the most vulnerable to climate change, we suggesting that ecologically restricted species should be prioritized for conservation.

Biodiversity is threatened worldwide by anthropogenic activities, and the identification of relevant areas for biodiversity is highly valuable to cope with cost-benefits trade-offs in conservation efforts. Such information is highly desired in the African continent. Therefore, we attempted to identify the African centers of diversity and endemism for desiccation-tolerant plants, a group of plants with a remarkable response to drought and largely neglected for conservation. For that, we compile a list of all known vascular species that occur in Africa and estimated the taxonomic and phylogenetic diversity, besides endemism richness and phylogenetic endemism, from a niche and spatial perspectives. We identified Southern Africa (e.g., Eastern highlands and the Great Escarpment) and the Eastern regions of the continent (e.g., East African Rift – Eastern Highlands – Drakensberg), as well as in Madagascar (e.g., Central Highlands) as centers of diversity for this continent. Our findings reveal priority areas for the diversity conservation in 19 different ecoregions of 3 biomes. However, we should not neglect regions rich in endemic species. Those areas covered 32 ecoregions of 4 biomes, extending the need for attention and monitoring in other regions like Ethiopian parts of East African Rift, Namibian Khomas Hochland, and in the Mascarene Islands.

Automated Red List assessments (AA) are rapidly developing approach to support extinction risk assessments. A major strength of AA is that they can speed up the time-consuming Red Listing process by helping experts to identify species for prioritisation and thereby mitigate a sever bottleneck for conservation practice . However, it is unclear how accurate AA reproduce expert-based Red Lists and for which species they can be applied reliably. Here, we present the results of an AA for Germany based on modelled distributions of 2,136 plant species at three time slices. Based on the modelled species ranges and changes in occurrence probability among time slices, we approximate indicators for three official Red List criteria (current population status, short-term trend, and long-term trend), which we then automatically combine into extinction risk assessments. Overall our AA agreed well with the expert-based Red List for Germany, with an average difference of only one category. In general, the automated Red List was more optimistic (suggesting lower extinction risk), although with some variation related to region and extinction risk category. Our results suggest, that AA can provide a reliable data-driven baseline to support expert Red Listing, in particular for species with medium to large distribution ranges.