Ignasi Bartomeus’s research while affiliated with Doñana Biological Station and other places

Motivation Pollinators play a crucial role in maintaining Earth's terrestrial biodiversity. However, rapid human‐induced environmental changes are compromising the long‐term persistence of plant‐pollinator interactions. Unfortunately, we lack robust, generalisable data capturing how plant‐pollinator communities are structured across space and time. Here, we present the EuPPollNet (European Plant‐Pollinator Networks) database, a fully open European‐level database containing harmonised taxonomic data on plant‐pollinator interactions referenced in both space and time, along with other ecological variables of interest. In addition, we evaluate the taxonomic and sampling coverage of EuPPollNet, and summarise key structural properties in plant‐pollinator networks. We believe EuPPollNet will stimulate research to address data gaps in plant‐pollinator interactions and guide future efforts in conservation planning. Main Types of Variables Included EuPPollNet contains 1,162,109 interactions between plants and pollinators from 1864 distinct networks, which belong to 52 different studies distributed across 23 European countries. Information about sampling methodology, habitat type, biogeographic region and additional taxonomic rank information (i.e. order, family, genus and species) is also provided. Spatial Location and Grain The database contains 1214 different sampling locations from 13 different natural and anthropogenic habitats that fall in 7 different biogeographic regions. All records are geo‐referenced and presented in the World Geodetic System 1984 (WGS84). Time Period and Grain Species interaction data was collected between 2004 and 2021. Major Taxa and Level of Measurement The database contains interaction data at the species level for 94% of the records, including a total of 1411 plant and 2223 pollinator species. The database includes data on 6% of the European species of flowering plants, 34% of bees, 26% of butterflies and 33% of syrphid species at the European level. Software Format The database was built with R and is stored in ‘.rds’ and ‘.csv’ formats. Its construction is fully reproducible and can be accessed at: https://doi.org/10.5281/zenodo.14747448 .

With many species interacting in nature, determining which interactions describe community dynamics is nontrivial. By applying a computational modeling approach to an extensive field survey, we assessed the importance of interactions from plants (both inter‐ and intra‐specific), pollinators and insect herbivores on plant performance (i.e., viable seed production). We compared the inclusion of interaction effects as aggregate guild‐level terms versus terms specific to taxonomic groups. We found that a continuum from positive to negative interactions, containing mostly guild‐level effects and a few strong taxonomic‐specific effects, was sufficient to describe plant performance. While interactions with herbivores and intraspecific plants varied from weakly negative to weakly positive, heterospecific plants mainly promoted competition and pollinators facilitated plants. The consistency of these empirical findings over 3 years suggests that including the guild‐level effects and a few taxonomic‐specific groups rather than all pairwise and high‐order interactions, can be sufficient for accurately describing species variation in plant performance across natural communities.

Perennial woody crops, which are crucial to our diets and global economies, have the potential to play a major role in achieving multiple UN Sustainable Development Goals pertaining to biodiversity conservation, socio-economic development and climate change mitigation. However, this potential is hindered by insufficient scientific and policy attention on perennial woody crops, and by the intensification of perennial crop cultivation in the form of monocropping with high external inputs. In this Perspective, we highlight the potential of properly managed and incentivized perennial woody crops to support holistic sustainable development and urge scientists and policymakers to develop an effective agenda to better harness their benefits.

Pollination is a crucial ecosystem service, yet pollinator species diversity is declining as a result of factors such as climate change, habitat loss and agricultural intensification. While previous studies have often examined the extreme scenario of complete pollinator removal, showing negative impacts on plant reproductive success, we take a more realistic approach by focusing on the effects of decreasing pollinator diversity. Our global meta-analysis reveals a notable negative impact of reduced pollinator species diversity on plant reproductive success measures, such as seed set, fruit set and fruit weight. Notably, this effect varies across plant families, impacting both self-incompatible and self-compatible species. We also find that wild plant species suffer more than cultivated ones. Furthermore, the loss of invertebrate, nocturnal and wild pollinators has a more substantial impact than the loss of vertebrate, diurnal or managed pollinators. Overall, our findings consistently underscore the positive role of biodiversity in maintaining ecosystem functioning, highlighting the urgency of mitigating factors that lead to the decline in pollinator species diversity.

Tools to predict pollinator activity at regional scales generally rely on land cover maps, combined with human-inferred mechanistic rules and/or expert knowledge. Recently, Giménez-García et al. (2023) showed that, using large pollinator datasets, different environmental variables, and machine learning models, those predictions can be enhanced but at the cost of losing model interpretability. Here, we complement this work by exploring the potential of using advanced machine learning techniques to directly infer wild-bee visitation rates across different biomes only from land cover maps and available pollinator data while maintaining a mechanistic interpretation. In particular, we assess the ability of convolutional neural networks (CNNs), which are deep learning models, to infer mechanistic rules able to predict pollinator habitat use. At a global scale, our CNNs achieved a rank correlation coefficient of 0.44 between predictions and observations of pollinator visitation rates, doubling that of the previous human-inferred mechanistic models presented in Giménez-García et al. (2023) (0.17). Most interestingly, we show that the predictions depend on both landscape composition and configuration variables, with prediction rules being more complex than those of traditional mechanistic processes. We also demonstrate how CNNs can improve the predictions of our previous data-driven models that did not use land cover maps by creating a new model that combined the predictions of our CNN with those of our best regression model based on environmental variables, a Bayesian ridge regressor. This new ensemble model improved the overall rank correlation from 0.56 to 0.64.

Mutualistic interactions among organisms are fundamental to the origin and maintenance of biodiversity. Yet the study of community dynamics often relies on values averaged at the species level, ignoring how intraspecific variation can affect those dynamics. We propose a theoretical framework for evaluating the extent to which various forms of variation within populations can influence species’ persistence in mutualistic systems. Next, drawing from detailed empirical data on plant–pollinator interactions and plant fitness, we quantify intraspecific variation in the mutualistic benefits received by plants and incorporate this variation into estimations of the community’s structural stability, a robust theoretical measure of species’ likelihood of persistence. Through explicit consideration of intraspecific variation, we are able to demonstrate that having different combinations of specialized and generalized individuals within plant populations promotes the persistence of pollinator communities. Further, we find that these heterogeneous mixtures of plant individuals reduce the probability of exclusion of focal plant species by promoting indirect effects across the broader plant–pollinator community. By providing a framework that explicitly accounts for individual-level variation, we open the door to a better understanding of the mechanisms promoting biodiversity in mutualistic communities and beyond.

Evolutionary and ecological forces shape species coexistence, but how different ecological mechanisms drive coevolutionary dynamics remains poorly understood. Focusing on mutualistic communities, we explore how morphological and phenological trait matching can shape the coevolution of species traits, influence the evolutionary trajectories at the community level, and determine community stability. Using in silico experiments, we first show that because phenological traits can decouple interactions in time, their coevolutionary dynamics led to the emergence of interaction motifs promoting facilitation over competition. In contrast, coevolution driven by morphological traits led to poorly structured networks with higher connectance. As a consequence, phenological coevolution increased the ecological stability of the community, relative to those coevolved based on morphology, and dampened the diversity-stability trade-off observed in morphologically coevolved communities. Next, by using 17 empirical pollination networks, we show that phenological motifs promoting facilitation were abundant in natural communities, and that as predicted by the theoretical models, the phenological structure in species interactions was a major determinant of the structural stability of these empirical communities. These results show that modelling explicitly the basic mechanisms determining species interactions is crucial to understand how species coevolve, and the ecological properties emerging at the community level, such as community structure and stability.

Pollinators are essential for the health of ecological systems. They transfer pollen from one flowering plant to another, helping the plants complete their life cycle. In doing so, pollinators also provide important functions and services critical for human wellbeing. But many pollinator populations are in decline due to ongoing challenges such as climate change and habitat loss, and accurately predicting which populations are most at risk is vital to preventing their extinction but poses many challenges. To improve current prediction approaches, a new study combined mathematical modeling and data collected in the field over 6 years. This long-term dataset on plants and their pollinators across 12 sites revealed that it is the interactions among species that drive pollinator persistence, with the most persistent communities having a nested interaction structure characterized by a particular combination of specialist and generalist species. The most persistent communities were also found to inhabit the largest habitat patches, allowing them more resilience to environmental change. These findings suggest that a better understanding of species interactions can improve our ability to make sound species management decisions and highlight that preserving interaction networks is key to pollinator conservation under global change.

Agricultural practices shape arthropod communities in arable fields, consequently influencing their interactions and the resulting ecosystem services, in particular pest regulation. Predatory arthropods play a pivotal role by preying on herbivores, soil fauna, and on other predators. However, the intricate mechanisms through which agricultural practices shape the dietary preferences of predators, and regulate herbivore populations remain complex and inadequately understood. We assessed how fertilisation with organic fertiliser and extending crop rotations with perennial ley affected predation pressure across prey taxa. We mapped predator and prey trophic linkages with molecular analysis of carabid predator gut contents, and measured densities and taxonomic richness of predators, herbivores, and soil fauna in 19 cereal fields during three samplings across the growing season. We derived two food web structure metrics: prey vulnerability that is the average number of predators feeding on a selected prey, and predator trophic redundancy, that is dietary overlap. Prey vulnerability was compared among soil fauna, herbivores, and other predator species (that is interspecific intraguild predation) over the growing season, and across treatments. The mechanistic underpinnings of observed shifts in vulnerability of herbivorous prey at different crop stages were identified using information criteria to select among candidate variables related to the richness, density and interaction structure of the different guilds during both the current, and the previous crop stages. Agricultural diversification via organic fertilisation combined with perennial ley in the crop rotation decreased the vulnerability of both intraguild prey and soil fauna prey, and stabilised herbivore vulnerability. Mechanistically, the vulnerability of herbivorous prey at crop ripening emerges from the combination of predator richness and trophic redundancy during this sampling round, rather than from carryover effects from previous crop stages. Synthesis and applications: Our results suggest that locally provided resource continuity through diversified cropping practices bolster biological pest regulation, thus underline the importance of lesser disturbance in arable ecosystems for the provision of ecosystem services. Enhanced predator species richness together with availability of alternative prey through the season underpins this enhanced pest regulation.